ホーム » 研究テーマ一覧(旧ver)

# 研究テーマ一覧(旧ver)

●テーマ：制御理論の研究とその応用

プログラミング

OS : Windows

C言語による割り込み処理，リアルタイム制御など諸々．

ロバスト制御：ロバスト安定性，H無限大制御，ミュー設計法，LMI，２次安定化コントローラ，ニューロコントローラ，適応制御，モデル予測制御，ファジィ制御

• 生理学的光合成モデルの制御理論的システム構成と多細胞同期現象の解明：シミュレーション
• 倒立振子制御系へのセンサ攻撃とその対策：シミュレーションと実験
• 力学モデルによる最適化計算と太陽電池制御系への応用：シミュレーションと実験
• マルサス係数推定による減衰/成長度評価と生体系への応用：シミュレーション
• クアッドロータの外乱推定と制御：シミュレーションと実験

シミュレーションで使用する基本ソフトウェア

• 倒立振子（サーボテクノ製）

• 慣性運動装置NAV440(Crosbow製)

・太陽光発電モジュールと組込システムμITRON

クアッドロータの外乱推定と制御

Research Interests:Control Systems Theory and Its Applications

Adaptive Control Theory（適応制御理論）, Nonlinear Control Theory, and Robust Control Theory（ロバスト制御理論

・松尾孝美：ハイゲイン入力推定器を用いた適応パラメータ調整則，電気学会九州支部，12-2P-07 (2008).
・Robust Adaptive Control with a Parameter Adjustment Law Based on a Riccati Equation, under preparation
Adaptive Control with a High-Gain Input Estimator,submitted (2008)
・Takami Matsuo , Kentaro Adachi and Haruo Suemitsu:Frequency Estimation with an LMI-based Adaptive Update Law, Int. J. International Journal of Advanced Mechatronic Systems, 1-2, pp.100-107 (2008).
Kentaro Adachi, Haruo Suemitsu, and Takami Matsuo:LMI-Bsed Frequency Estimator With Averaging,ICICIC 2007.
・Kentaro Adachi, Haruo Suemitsu, and Takami Matsuo:Frequency Estimator with LMI-type Adaptive Update Law,SICE-ICASE International Joint Conference 2006 (SICE-ICCAS2006) (2006.10).
・Riccati方程式に基づくパラメータ調整則とその応用，計測自動制御学会論文集， 35-12, 1631–1633 (1999)
・Robust Parameter Adjustment Law Based on a Riccati Equation,
Proceedings of Control 97, Sydney, Australia, pp.628-633 (1997)
・強正実条件を緩和した適応制御系の１設計法，計測自動制御学会論文集, 32巻9号, pp.1335-1344 (1996)
・A Design of Robust Adaptive Control System with a Relaxed Strictly-Positive-Realness Condition, Proceedings of the 34th IEEE conference on decision and Control, pp.2322-2327 (1995)

パラメータ調整則に周波数制限がついているので，高周波外乱に弱いのではないかとの 指摘をうけている．このための改善を検討している．

### 2）（非線形制 御）瞬時リアプ ノフ指数による非線形システムの同期判定

Yusuke Totoki and Takami Matsuo : Decay/Growth Rate Estimation Using Instantaneous Lyapunov Exponent, Int. J. Bifurcation and Chaos, 22-3, (2012) 1250047 (13 pages) DOI: 10.1142/S0218127412500472 (2012)

Yusuke Totoki, Akira Goto, Haruo Suemitsu, and Takami Matsuo: Synchronization Detection of Biological CAM Plants Using Instantaneous Lyapunov Exponent, proc. of SICE Annual Conference 2010, 514-519 (2010).
・Yusuke Totoki, Haruo Suemitsu, and Takami Matsuo:Decay Rate Estimation of Continuous Time Series Using Instantaneous Lyapunov Exponent, Proc. of ICCAS-SICE 2009, 5073-5077 (2009).

・十時優介，末光治雄，松尾孝美：瞬時リアプノフ指数を用いた連続時間時系列の減衰度推定，101C3,2008
・十時優介，末光治雄，松尾孝美：時系列データからの瞬時減衰度推定，電気学会九州支部,12-2P-08,2008
・十時優介，松尾孝美：瞬時リアプノフ指数を用いた非線形システムの同期判定，計測自動制御学会九州支部学術講演会,A102,2007.

Lyapunov指数のうち最大のものを，最大Lyapunov指数といい，非線形システムの解(フロー)が遠ざかっていく速さのもっとも大きなものを代 表している．この値が正であると，フローが現在の位置から離れていくことを意味しており，その位置が局所安定ではないことを物語っていると考えられてい る．Maybhateらは，同一構造の2つのシステム間の同期をパラメータ推定に応用し，その際に，Lyapunov指数の計算を用いている．しかしなが ら，つぎのような2つの問題点が指摘されている．最近，安定な同期多様体のいくつかの不安定不変集合の存在から，負のLyapunov指数がカオス同期の ための十分条件でも必要条件でもないことが報告されている．また，Lyapunov指数を正確に計算するためには，ダイナミクスが完全に既知でなければな らず，未知パラメータを含む場合には，計算することができない．そこで，Huangら\cite{huang}は，Lyapunov指数ではなく， Lyapunovの安定論に基づいて，時系列からすべての未知パラメータを同定するシステム論的な手法を提案している．このようなことから Lyapunovの安定論が安定性解析の主流となっていると考えられる．

そこで，本研究では未知パラメータを含む場合でも計算できる時系列を用いた新たなLyapunov指数である瞬時Lyapunov関数 (instantaneous Lyapunov exponent, ILE)を定義し，局所安定性判別や同期判定にどの程度有効であるかについて考察する．ILEは時系列の減衰率を推定する局所的指標であるが，基準関数と 呼ばれる関数を適宜選ぶことにより，指数安定以外の局所安定性の判別にも適用することができる．本報告では，まず，regularでないシステムと2次振 動システムに対して，瞬時Lyapunov関数を算出し，その特性を調べる．ついで，Huangらのパラメータ未知のLorenzシステムに対するカオス 同期システムの同期判定に対するILEの有効性を検証する．

### ３）（適応制 御）時変系の適 応制御と微分トラッカー設計

Papers

*Yasuhito Eguchi, Tsuyoshi Ohba, Haruo Suemitsu, and Takami Matsuo : Auto-Tuning Velocity Estimator by Using Adaptive Observer, ICIC Express Letters, 8-2, 427-433 (2014).
*Yasuhiro Kawakami, Yasuhito Eguchi, Tsuyoshi Nimiya, Haruo Suemitsu, and Takami Matsuo : Velocity and Acceleration Estimation by Iterative Learning Observer and Perfomance Validation with MEMS-Based Inertial Sensors, Int. J. of Advanced Mechatronic Systems, 5-2, 113-121 (2013).
*大羽　健,江口泰史,末光治雄,松尾孝美：オートチューニング適応微分推定器の安定性とその応用,第57回システム 制御情報学会研究発表講演会，114-6 (2013)
*大羽 健,江口 泰史, 末光治雄, 松尾孝美：慣性計測機器を用いたオートチューニング適応微分推定器の性能検証,第56回自動制御連合講演会,1259-1262 (2013).
*江口泰史，大羽 健，末光治雄，松尾孝美:反復学習オブザーバによる速度・加速度推定器の慣性計測機器による性能検証,第32回計測自動制御学会九州支部学術講演会(2013)
*大羽 健,江口 泰史, 末光 治雄, 松尾 孝美:オートチューニング適応ゲイン則による速度推定器の慣性計測機器による性能検証,第32回計測自動制御学会九州支部学術講演会(2013)
*Yasuhiro Kawakami, Tsuyoshi Nimiya, Haruo Suemitsu and Takami Matsuo : Performance Comparison of Velocity Estimators with MEMS-Based Inertial Sensors, Proc. of 2012 International Conference on Advanced Mechatronic Systems, pp.195-200 (2012).

*江口泰史，川上裕大，末光治雄，松尾孝美 :反復学習オブザーバを用いた速度・加速度推定器設計と性能検証,第55回自動制御連合講演会,1160-1165 (2012)
*江口泰史，川上裕大，末光治雄，松尾孝美 :適応ゲインのオートチューニングと微分推定器への応用,第31回計測自動制御学会九州支部学術講演会，5-6 (2012)
*川上裕大,江口 泰史,末光 治雄,松尾 孝美:慣性計測機器を用いたモデルフリー加速度推定器の性能検証,第13回システムインテグレーション部門講演会,1536-1541 (2012)
* 江口泰史，川上裕大，末光治雄，松尾孝美 :オートチューニング微分推定器の性能評価,第13回制御部門大会(2012)
*Tsuyoshi Nimiya, Haruo Suemitsu, Takami Matsuo, Sawut Umerujan, and Kazushi Nakano : Velocity and Acceleration Estimation of an Electric Throttle System by Adaptive Differential Filter, Proc. of IEEE/SICE-SII2011,409-414 (2011).
*川上裕大,荷宮 剛,末光治雄,松尾孝美 : 慣性計測機器を用いたモデルフリー速度・加速度推定器の推定性能比較,第12回計測自動制御学会制御部門大会(2011)
*川上裕大,荷宮 剛,末光治雄,松尾孝美 : 慣性計測機器を用いた実測値による速度・加速度推定器の推定性能検証,第30回計測自動制御学会九州支部学術講演会講演会(2011)

*Yuta Kitsuka, Tsuyoshi Nimiya, Haruo Suemitsu, and Takami Matsuo:Non-Model-Based Velocity and Acceleration Estimators for a Suspension System with Parallel Connection of a Hydraulic Actuator, proc. of 2010 IEEE Multi-Conference on Systems and Control, pp.549-554 (2010).
* Yuta Kitsuka, Tsuyosi Nimiya, Haruo Suemitsu, and Takami Matsuo: Feedback Control of a Suspension System with Non-Model-Based Velocity and Acceleration Estimators, proc. of SICE Annual Conference 2010, 484-489 (2010).
*Yuta Kitsuka, Tatsuya Nomura, Haruo Suemitsu, and Takami Matsuo:Adaptive Estimation of Velocity and Acceleration with High Gain Input Observer, proc. of Int Conf. on Modelling, Identification and Control, pp.436-441 (2010).

*Tatsuya Nomura, Yuta Kitsuka and Takami matsuo : Nonmodel-Based Estimation for Velocity and Acceleration by Adaptive Identification Method, IEEJ Trans. on Electrical and Electronic Engineering, 5-3, 372-374 (2010).
*野村達八，松尾孝美：適応微分推定器による非モデルベース加速度推定，自動制御連合講演会CD-ROM,H6-2 (2009).
*野村達八，末光治雄，松尾孝美：適応微分推定器による速度・加速度推定とその応用，SICE九州支部，101C4 (2008)
*野村達八，末光治雄，松尾孝美：リアルタイム加速度推定器の構成，電気学会九州支部，12-2P-06 (2008).
*Takami Matsuo, Shuhei Wada, and Haruo Suemitsu :Model-Based and Non-Model-Based Velocity Estimators for Mobile Robots,Int,J. Innovating Computing, Information and Control, 4-12, pp.3123-3133 (2008).
*Shuhei Wada, Haruo Suemitsu, and Takami Matsuo:Model-Based and Non-Model-Based Velocity Estimators and Its Application to Mobile Robots,A03-06.pdf (4 pages),Proc. of Second International Conference on Innovative Computing, Infromation and Control (ICICIC-2007).
*Shuichi Wada, Kouichi Mitsunaga, Haruo Suemitsu, and Takami Matsuo:Frequency Estimator with LMI-type Adaptive Update Law,SICE-ICASE International Joint Conference 2006 (SICE-ICCAS2006).
*Soon-Hyun Park and Takami Matsuo: Time-Derivative Estimation of Noisy Movie Data Using Adaptive Control Theory,International Journal of Signal Processing, 2-3, pp.170-177 (2005).

Abstract：

これらの手法は，時間関数に時変数の高次多項式が含まれる場合には，誤差となって現れるため，短時間での推定に向いている．さらに，スライディングモード制御の応用として，Levantは厳密微分器を提案している\cite{levant}．

この微分器はパラメータ推定を用いて得られるので，信号に付加される出力ノイズから微分推定値への伝達関数が
$\frac{\gamma}{s}$となり，高周波成分のゲインを低減させることができる．$\gamma$を大きくすると推定の速応性をあげることができるが，高周波帯域のゲイン特性は上方にシフトするだけであるので，近似微分器に比べて，ノイズ抑制に利点があると考えられる．しかしながら，提案した推定器では，1階微分の推定は可能であるが，相対次数の制約から，2階以上の微分値の推定を行うことができなかった．

### ４）（非線形制 御）Trajectory Tracking of Nonlinear Systems with Scalar Function

・Trajectory Tracking of Nonlinear Systems via Lie Derivative of a Scalar Function, IFAC Workshop on Periodic Control Systems (PSYCO 2004), 8/30-9/1/2004, Yokohama.

Energy is one of the fundamental concepts in science and engineering practice, where it is common to view dynamical systems as energy-transformation devices. The control problem can be recast as finding a dynamical system and an interconnection pattern such that the overall energy function takes desired form\cite{Chung,Astrom,Ortega}. This control scheme is called the energy shaping approach and is the essence of the passivity-based control. Positive definite function such as the storage function plays a key role in the passivity-based control\cite{Byrnes}. Moreover, the trajectory tracking control problem for nonlinear systems is one of the most challenging of the control theory\cite{Mazenc,Slotine}. It can be solved locally for small reference state trajectories using linear control techniques. However, a nonlinear control strategy is required to track bounded trajectories with large amplitude and obtain large basins of attraction. In this issue, the Lyapunov stability theory plays a key tool to derive a time-varying nonlinear controller and to guarantee its stability. Chung {\em et al.}\cite{Chung} discussed the control problems of a swinging pendulum using an energy function. \AA str\”{o}m {\em et al.}\cite{Astrom} investigated an energy control strategies for swing-up a pendulum for a cart-less pendulum under the condition that the input was the acceleration of the pivot. Moreover, Chatterjee {\em et al.}\cite{Chatterjee} dealt with the swing-up and stabilization problem of a cart-pendulum system with a restricted cart track length and restricted control force using generalized energy control methods.

In this paper, we propose a nonlinear state feedback control law to satisfy a constraint on state variables. The constraint is expressed as an algebraic equation containing a scalar nonlinear function such as an energy function or a limit cycle. Computing the time-derivative of the scalar function that defines the constraint of the state variables, we derive a nonlinear controller to regulate the state variable satisfying the constraint. A cart-less pendulum system is given to demonstrate the closed-loop performance of the proposed controller. In particular, we present a controller considering a viscous friction of the pendulum.

5)（非線形制御）ファジィ微分方程 式の求解と安定性解析

Teppei Hiroshige, Haruo Suemitsu, and Takami Matsuo : Stochastic Evaluation of Dissipative Observer for Multivalued Nonlinear System with Coulomb Friction, Proc. of IEEE/SICE-SII2011,392-397.

(2010)
ファジィ数とは，「だいたい5」をあらわすファジィ概念で，システムの記述に関わるパラメータや関数および方程式等のファジィ化には，ファジィ数が用いら れる．微分方程式の構造がわかっているが，パラメータや初期値が不確かな場合に，これらの値が曖昧だとして，ファジィ数として扱い，微分方程式にファジィ 概念を導入したものが，ファジィ微分方程式である．ファジイ微分方程式の解の存在条件，解の一意性，安定性などの研究が多く行われてきている．ファジィ微 分方程式の研究は，集合関数の微分定義の悪条件性から，つぎのように多様な解が存在する．
・拡張原理による解，Seikkala微分による解
・ Hukuhara微分による解\cite{kaleva2}，一般化Hukuhara微分による解
・$\pi$微分による解，微分包含による解，Liu過程による解
などがある．

Osorio {\em et al.} presented a dissipative based method to design observers for nonlinear discountinuous/multivalued systems \cite{osorio}. They modeled a multivalued system as a differential inclusion and derived the global exponential stability conditions in the LMI forms. In the example, they designed the full-order observer of the 2-cart system with coulomb friction as an example of the discountinuous/multivalued systems.Friction between sliding contact surface is a result of stochastic interactions between rubbing asperities \cite{anderson}.　The rubbing asperities cause friction by shearing surface materials, lubricants, or surface coating \cite{anderson}. Yet while the friction force in nearly all friction tests is highly stochastic in nature, with significant variations in both amplitude and frequency, most friction models do not take variations into account and instead represent the friction forces by a smooth mean value \cite{anderson}.　Anderson {\em et al.}\cite{anderson} introduced a additional noise to the smooth mean friction force in order to represent the stochastic nature of friction. They does not recommend white noise but adopt an average flow model using sinusoidal disturbance, because white noise contains frequencies that are infinitely high. However, Feng \cite{feng} used a Gaussian white noise for slip mode.
In this paper, instead of the multivalued function of the example in \cite{osorio}, we introduce a simulation model with uniformly distributed random number in the low velocity area in order to validate the performance of the observer using the multivalued systems design proposed in \cite{osorio}. We show that the stochastic approximation of the multivalued function affects the estimation performance and it is impossible to estimate the friction force near the zero velocity. Finally, we design the adaptive update law to estimate the Coulomb sliding friction force.

### ６）（ロバスト制 御）制御系への攻撃と検出モニタ

・Kotaro Murakami,Haruo Suemitsu and Takami Matsuo :Repeated Replay-Attack to a Feedback Control System, ICT-ROBOT2016, ThCT2.2, Sept 7-9,Bexico, Busan, Korea,2016.

サイバー物理系とは，物理システムと制御指令を出す情報システムとがネットワークを介して接続されているシステムを指 し，近年の通信技術の発展が可能にした統合システムである．コンピュータと物理的対象のシステム統合により工業オートメーションが実現し，人の手による操 作等，多くの作業をコンピュータで代替できるようになった．その結果，通信ネットワークによる情報システムと物理システムの統合によって，技術的かつ潜在 的利益がもたらされるようになった．しかし，サイバー物理系において，通信網の拡大やシステムの複雑化はセキュリティの観点からは，リスクの増大を意味す る．なぜなら，通信網の拡大やシステムの複雑化によって攻撃表面は拡大するので，センサを騙したり，システムに悪データを注入することは難しくな いからである．無人航空機へのGPS攻撃の既知の例がある\cite{Iran}．このように，サイバー物理系はサイバー攻撃を受ける可能性が高いが，現 時点では事例が少なく，セキュリティ対策が立てにくい状況にあるため，攻撃の事例収集や考えられるシナリオの作成の必要がある．

Teixeira ら\cite{Teix}は，敵対者のシステム情報，漏えい，障害要因によって定義された攻撃空間を導入し，この空間において，Zero dynamics attack, Bias injection attack, Denial-of-Service (DoS) attack, Covert attack, Eavesdropping attack, Replay attackなどのいくつかの攻撃シナリオを分類している．こ のような現状を踏まえ，Tiwari\cite{Tiwari}は攻撃検出のためのモニタの存在下で攻撃者が利用する有効な攻撃戦略として，Bias injection attackを導入し，一部の設定において有効性を確認した．しかし，Tiwariの提案した攻撃戦略は，状態制約をもつもつハイブリッドコントローラで あるけれども，通常のセンサ値にオフセットを乗せるものであるため，I制御を含むコントローラから構成されている制御系においては無効であると考えられ る．
そこで我々は，I制御を含む制御系においても有効な攻撃戦略として，リプレイ攻撃の新たなシナリオを考えた\cite {Murakami1,Murakami2,Murakami3}．攻撃者にとって，リプレイ攻撃を採用する利点は，たとえセンサ値が暗号化されていて も，暗号化された後のセンサ値を記録・再生するので，暗号化によるセキュリティを突破した上で，システムを攻撃できる可能性があるということである．
まず，システムには，ある閾値以上になると攻撃を知らせるモニタ(Failure detector/Safty monitor)が装備されているものと仮定し，そのシステムに検出されずにシステムを危険な状態に陥らせることができるかを検証した.
つ いで，2つの攻撃を検証した．第一は，Moら\cite{Mo}が定義したセンサ値を記録し，時間を遅れさせて再生するだけの攻撃であり，これを遅延リプ レイ攻撃と呼んだ．もう一つは我々が提案したもので，一定期間センサ値を記録し，その後，記録したセンサ値を繰り返し再生するという攻撃である．これを繰 り返しリプレイ攻撃と呼んだ.

\begin{itemize}
\item 開ループ型 : センサをフィードバック系から切り離し，侵入して記録したデータを観測値としてコントローラに送る攻撃
\item バイアス注入型 : フィードバック系はそのままで，侵入して記録したデータをセンサ値に付加する攻撃
\item 内部モデル型 : 繰り返し制御系と同様に周期信号を発生させる内部モデルをフィードバックループ内に仕込む攻撃
\end{itemize}
の3 つのタイプに分けて，適応クルーズコントローラの安定性を検証するとともに，各々の攻撃に対する攻撃検出するためのモニタをオブザーバ設計問題に帰着す る．シミュレーションでは，Tiwari\cite{Tiwari}が扱った適応クルーズコントローラを攻撃の対象とし，システムの状態変数を先導車と後 続車の車間にしたものを用いる．攻撃者は，システムの出力にあたる距離センサをハッキングするものとし，このシステムに対して，LQIコントローラを設計 し,繰り返しリプレイ攻撃を仕掛け，これらのオブザーバを設計することにより，繰り返しリプレイ攻撃に強いシステムが構成できるかどうかを検証する．

### ７）（ロバスト 制御）Robust Control by PID+Q Controller

・Robust Stabilization of closed-loop Systems by PID+Q Controller,International Journal of Control, vol.70-4, 631-650 (1998)
・Robust Control of Multivariable Systems by PID+Q Controller
Proceedings of American Control Conference, Albuquerque, USA, pp.3674-3678 (1997)
・Robust Stabilization of Closed-Loop System with PID Controller by use of Youla Free Parameter,
Proceedings of the 7th International Symposium on Dynamic Games and Applications,
Vol.2, pp.636-648 (1996)
・A Parametrization of All Stabilizing Compensators with a Static Output
Feedback in the Central Part and Its Application to PID Controller,
Preprints of 13th IFAC World Congress, Vol.C, pp.229-234 (1996)

・PID+Qコントローラのノミナル性能はPIDコントローラのそれと一致し，自由パラメータには依存しない．このため，３つのPIDパラメータはロバスト安定性やロバスト性能を上げるためのみに用いることができる．
・PID+Qコントローラは１型の閉ループ応答を保証するすべての安定化補償器を表現している．
・ロバスト安定性を向上するような自由パラメータは，ある種のH∞モデルマッチング問題を解くことに帰着され，これは LMIを用いて解くことができる．
さらに，つぎのような3条件を提案している．
・第1ロバスト安定条件
・第2ロバスト安定条件
・ノミナル性能回復条件

ロバスト性のみの向上を行うコントローラであり，基本性能はPIDコントローラのパラメータの選定に依存している．また，全体のコントローラの次数が大きくなるため，低次元化が実装の大きな問題になる．

### ８）（システム 理論）閉ループ 応答の最大振幅指定のためのマトリクス不等式の導出

・Takami Matsuo :
Some Transfer-Function Conditions for a Desired Maximum Amplitude or Exponentioal Envelope of a Closed-Loop Transient Response, Proceedings of the 33rd IEEE Conference on Decision and Control, pp.2659-2660 (1994)

### 論文：

Shinkyo Hashizu,Haruo Suemitsu and Takami Matsuo :Cell Synchronization of Crassulacean Acid Metabolism plant with Feedback Loop, ICT-ROBOT2016, ThCT2.1, Sept 7-9,Bexico, Busan, Korea,2016.
・Ryo Sakamoto, Chisato Matoba, Haruo Suemitsu and Takami Matsuo: Feedback Control of Biological Rhythm in Crassulacean Acid Metabolism by CO2-Uptake Signal,　Preprints of the 4th IFAC Conference on Analysis and　Control of Chaotic Systems, Tokyo, Japan, August 26-28, pp.59-64 (2015)

Takami Matsuo, Yusuke Totoki, and Haruo Suemitsu : Adaptive Estimation of Biological Rhythm in Crassulacean Acid Metabolism with Critical Manifold, ISRN Applied Mathematics, vol. 2013, Article ID 856404, 9 pages, 2013. doi:10.1155/2013/856404.
・Ryo Sakamoto, Akira Goto, Haruo Suemitsu and Takami Matsuo : Phase Shift of Biological Rhythm in Crassulacean Acid Metabolism by Controlling Light Intensity, Proc. of SICE2013, 783-786 (2013).
・松尾孝美，後藤　彬，坂本　亮，末光治雄 :ヌルクライン設計及び入力変数指定による光合成リズムのフィードバック制御,第55回自動制御連合講演会,1304-1209 (2012)
・ 坂本 亮,末光 治雄, 松尾孝美：二酸化炭素取り込み量のフィードバックによるCAM植物の周波数と位相制御,第56回自動制御連合講演会,1160-1162 (2013).
・坂本 亮,末光 治雄, 松尾 孝美:二酸化炭素取り込み量を用いたCAM植物のフィードバック制御,第32回計測自動制御学会九州支部学術講演会(2013)

Akira Goto, Yusuke Totoki, Haruo Suemitsu, and Takami Matsuo : Control of Biological Clock in Crassulacean Acid Metabolism Using Nullcline Design, Proc. of IEEE/SICE-SII2011,1084-1089 (2011).
・Akira Goto, Yusuke Totoki,Haruo Suemitsu, and Takami Matsuo : Adaptive Observer-Based Controller for Biological Clock in Crassulacean Acid Metabolism, Proceedings of 2011 8th Asian Control Conference (ASCC),Kaohsiung, Taiwan, 500-505 (2011).
・ 後藤 彬,十時優介,末光治雄,松尾孝美 : Critical Manifoldを用いたCAM植物の非線形ダイナミクス推定,第53回自動制御連合講演会,587-592(2010)
Yusuke Totoki, Akira Goto, Haruo Suemitsu, and Takami Matsuo: Synchronization Detection of Biological CAM Plants Using Instantaneous Lyapunov Exponent, proc. of SICE Annual Conference 2010, 514-519 (2010).

・Yusuke Totoki, Haruo Suemitsu, and Takami Matsuo:On Robustness against Temperature Changes of Adaptive Estimator for CAM Plants,Proc. of the 34th ISCIE International Symposium on Stochastic Systems Theory and Its Applications, 7-12 (2010).
Yusuke Totoki, Haruo Suemitsu and Takami Matsuo:Nonlinear Dynamics Estimation of CAM Plants Using Slow Manifolds,,Proc. of SICE Annual Conference 2008, 1877-1882 (2008).
・十時優介，末光治雄，松尾孝美：CAM植物の非線形ダイナミクス推定，第8回適応学習制御シンポジウム，7-10 (2008).
・Masayuki Oka, Yusuke Totoki, Haruo Suemitsu and Takami Matsuo: Adaptive Observer for Biological Clock of Crassulacean Acid Metabolism with Partial States,C07-09.pdf (4 pages),Proc. of Second International Conference on Innovative Computing, Infromation and Control (ICICIC-2007).
・Masayuki Oka, Nodoka Goto, Haruo Suemitsu, and Takami Matsuo:Adaptive Estimator for Biological Clock of Crassulacean Acid Metabolism,SICE-ICASE International Joint Conference 2006 (SICE-ICCAS2006) (2006.10).
・岡　昌志，後藤　和，末光治雄，松尾孝美：CAMモデルの体内リズム推定のための適応オブザーバ設計，信学技報，NLP2006-28 (2006.7).
・Adaptive Observer-Based P-Controller for Biological Clock of Crassulacean Acid Metabolism, IFAC Workshop on Adaptation and Learning in Control and Signal Processing (ALCOSP 2004), 8/30-9/1/2004, Yokohama.
・Nonlinear Estimator and Controller for Biological Clock of Crassulacean Acid Metabolism, 6th Japan-France Congress on Mechatronics & 4th Asia-Europe Congress on Mechatronics, Saitama, 2003..

ほとんどすべての生物は，概日リズムをもっている．概日リズムは，普遍的なつぎの3つの性質を持っている．(1) 自由継続性：外部刺激が無くても約24時間周期のリズムが自律的に継続する．(2) 温度補償性：生理学的な温度範囲において，リズム周期が温度に影響されにくい．(3) 光位相同調性：明暗サイクルに同調することができる．高等植物の概日リズムは，遺伝子発現，光合成，成長などの多くの生理学的過程において，有用な役割を 担っている．哺乳類の概日時計は，視交叉上核の主要時計から構成され，他の概日時計を同調させている．ところが，高等植物においては，朝に生まれる植物の 時計たんぱく質の標的遺伝子が最近の遺伝レベルの研究で発見などあり，時計候補遺伝子は発見されているものの，主要な時計遺伝子が未だ確定されていない状 況であると思われる．また，植物は動物と異なり，動物の脳のような中枢は存在しないため，細胞間の同調機能がどのようになっているのかは，長い間未解決の 問題であった．植物の概日時計では，植物細胞間の結合が弱く，概日時計機能は，長い間，細胞自律的であると考えられていた\cite{Endo}．最近， 維管束と葉肉の概日時計が階層的にカップリングしていることがC3植物であるシロイヌナズナの実験により示されている\cite{Endo, Fukuda1,Takahashi}．植物の光合成には，C3型，C4型，およびCAM（Crassulacean Acid Metabolism）型がある．サボテンなどの多肉植物は，CAM型光合成(crassulacean acid metabolism,ベンケイソウ型有機酸代謝)と呼ばれる乾燥環境に高度に適応した炭素代謝機構を進化させている．我々は以前，制御理論の立場から， CAM生体リズムモデルを研究する場合の問題には，推定の問題，制御の問題，および同期の問題があることを言及した\cite{matsuo1}．今回の 解説では，最近の研究結果までを概観するとともに，細胞間同期モデルについて取り上げる．まず，Blasiusの1細胞モデルを状態方程式で記述し，連続 光下でのシミュレーション波形を示す．ついで，1細胞モデルであるOwenら\cite{Owen}のSD(System Dynamics)モデルとBlasius\cite{blasius2,beck}のモデルを比較し，各々モデルの優位点を概観する．また，Owenら が代謝パスウェイをモデル化するのに用いたVenSim(MIT開発言語)により，Blasiusのモデルを表現した結果を示し，2つのモデルの相違を確 認する．さらに，甲斐らのBlasiusのモデルを用いた多細胞結合モデルを，細胞外への炭水化物の流入出が細胞間結合であるという仮定に基づき，細胞内 のCO$_2$濃度の流束だけに細胞間のCO$_2$濃度の差がフィードバックされる2細胞同期モデルを提案し，このフィードバック結合が同期にどのよう な影響を与えるのかを，MATLAB/Simulinkによりシミュレーションを行う．

### ２)（システム 生物学系）モデ ルベースおよび非モデルベース心電図解析

*Akira Marui, Chisato Matoba, Takami Matsuo, and Yusuke Totoki : Decay/Growth Rate Evaluation by Malthusian Parameter Estimation and Its Application to ECG Signals, Proc. of 2016 IEEE 5th Global Conference on Consumer Electronics (GCCE), Oct.11-14, Kyoto, pp.539-540, 2016.
*

*
Chisato Matoba, Haruo Suemitsu and Takami Matsuo: Dual State-Parameter Estimation of ECG Signals with Recursive Bayesian Filters,The 47th ISCIE International Symposium on Stochastic Systems Theory and Its Applications, 12/5-12/8, 2015.
*
Chisato Matoba, Keisuke Imamasu, Haruo Suemitsu and Takami Matsuo:Parameter Estimation of Normal and Abnormal ECG Signals Using Recursive Bayesian Filters, SICE Annual Conf. 2015, July 28-30, 2015,  Hangzhou,China, 129-134 (2015).
*

*

*C. Matoba, K. Imamasu, H. Suemitsu and T. Matsuo:Detection of Cardiac Diseases from ECG Using Nonlinear Stochastic Filters, The 46th ISCIE International Symposium on Stochastic Systems Theory and Its Applications, Notre Dame Hall, Kyoto Institute of Technology, Kyoto, Nov.1-2,  (2014).
*Keisuke Imamasu, Chisato Matoba, Haruo Suemitsu and Takami Matsuo: Parameter Estimation of  Heart Rhythm Dynamics Using Adaptive Observer, Int. Conf. on Advanced Mechatronic Systems (2014).
*Chisato Matoba, Keisuke Imamasu, Haruo Suemitsu, and Takami Matsuo : Dynamics Estimation of ECG by Recursive Bayesian Filtering, SICE 2014 (2014).
*今増啓介, 的場ちさと, 末光治雄, 松尾孝美：心臓リズムの非線形モデルと実データによるダイナミクス解析,第56回自動制御連合講演会,1163-1166 (2013).
*今増啓介, 的場ちさと, 末光 治雄, 松尾 孝美:非線形振動子モデルを用いた心臓ダイナミクス推定,第32回計測自動制御学会九州支部学術講演会(2013)

### 論文：

Jun Mukae, Yusuke Totoki, Haruo Suemitsu, and Takami Matsuo : Parameter and Input Estimation in Hindmarsh-Rose Neuron by Adaptive Observer, Proc. of IEEE/SICE-SII2011, 1091-1095　(2011)

Ryuta Ito, Yusuke Totoki, Haruo Suemitsu, and Takami Matsuo: Adaptive Estimation of Firing Patterns of Hindmarsh-Rose Neurons and Synchronization Detection with Instantaneous Lyapunov Exponents, proc. of SICE Annual Conference 2010, 1743-1748 (2010).

・十時優介，末光治雄 ，松尾孝美：Hindmarsh-Roseニューロンの発火パターンの適応推定と瞬時リアプノフ指数を用いた同期判定，第54回システム制御情報学会研究発表講演会,583-584 (2010).
・伊東隆太，十時優介，末光治雄 ，松尾孝美：Hodgkin-Huxlayニューロンモデルにおける入力電流の適応推定，SICE九州支部，171-174 (2009).
・ Yusuke Totoki, Kouichi Mitsunaga, Haruo Suemitsu and Takami Matsuo : Firing Pattern Estimation and Synchronization Detection of Synaptically Coupled Hindmarsh-Rose Neurons, Neural Network World, 19-5,483-497 (2009).
・ Yusuke Totoki, Ryuta Ito, Haruo Suemitsu, and Takami Matsuo:Simultaneous Parameter and Input Estimation of Hindmarsh-Rose Neuron by Adaptive Observer,Proc. of ICCAS-SICE 2009, 4896-4901 (2009).
・Ryuta Ito, Yusuke Totoki, Haruo Suemitsu, and Takami Matsuo:Adaptive Input Estimation of a Hodgkin-Huxley Neuron, Proc. of ICCAS-SICE 2009, 5230-5235 (2009).
・Yusuke Totoki, Kouichi Mitsunaga, Haruo Suemitsu and Takami Matsuo:Firing Pattern Estimation of Synaptically Coupled Hindmarsh-Rose Neurons by Adaptive Observer，Lecture Notes in Computer Science 5164, 338-347, Springer (2008).
・Kouichi Mitsunaga, Yusuke Totoki and Takami Matsuo:Firing Pattern Estimation of Biological Neuron Models by Adaptive Observer, ICONIP2007, Part I, Lecture Notes in Computer Science 4984, 83-92, Springer (2008).
・ Hiroyuki Fujikawa, Kouichi Mitsunaga, Haruo Suemitsu, and Takami Matsuo:Parameter Estimation of Biological Neuron Models with Bursting and Spiking,SICE-ICASE International Joint Conference 2006 (SICE-ICCAS2006) (2006.10).

ニューロン同期のモデルは，積分発火(integrated-and-fire)モデルとコンダクタンスベースのスパイキング・バースティングモデルに分 けることができる\cite{belykh}．バースティングは，膜電位がゆっくり変化する静止期と周期的なバースト期の繰り返す現象である．
ニューラルネットワークダイナミクスのどの研究においても，次の2つの決定的問題が存在する\cite{Izh}．
\begin{enumerate}
\item どのようなモデルでニューロン単体のスパイクダイナミクスを表現できるか．
\item ニューロンの結合モデルでのスパイクダイナミクスはどのようになっているのか．
\end{enumerate}
Izhikevichは，スパイキングニューロンの多種なモデルを比較することにより第1の問題を考察した\cite{Izh}．彼は，単純なDCパルス 入力を加えて，tonic spiking, phasic spiking, tonic bursting, phasic burstingなどの皮質ニューロンの20種類のダイナミクスについてまとめた．
その結果，Hodgkin-Huxley(HH)モデルが最善であることを示唆した．また，Arenaは，モデル方程式のパラメータがニューロンを動的挙動を決定していることを指摘している\cite{Arena}．

そこで我々の研究室では，モデルが簡単でバースティングを再現できるHindmarsh-Rose(HR)ニューロンをモデルとして，膜電位の計測データ から単体HRニューロンの内部状態を推定する3つのオブザーバを提案した\cite{mitsu}．ここでは，Kalman-Yakubovich補題を 用いて，標準的な適応制御理論の手法から，適応オブザーバの誤差システムの漸近安定性を示した．これらの推定器を用いることにより，内部状態と内部パラ メータを推定することができ，推定したパラメータ値から，バースティングかスパイキングか区別のつかない早期のダイナミクスデータから発火パターンを特製 することが可能になる．さらに，シナプス結合された2つのHRニューロンのダイナミクスに焦点を当てた研究を行った\cite{totoki}．

4)
（メカトロ ニクス系）倒立振子の非線形安定化およびスウィング制御

Hiroya Oka, Yuji Maruki, Haruo Suemitsu, and Takami Matsuo :Nonlinear Control for Rotational Movement of Cart-pendulum System Using Homoclinic Orbit,International Journal of Control, Automation, and Systems, vol. 14, no. 5, pp.1270-1279, 2016.
・Hiroya Oka, Haruo Suemitsu and Takami Matsuo:Stabilizing Control of Inverted Pendulum via Interlaced Backstepping and Forwarding Method,
SICE Annual Conf. 2015, July 28-30, 2015,  Hangzhou,China,  608-613 (2015).
Yuji Maruki, Kohei Kawano, Haruo Suemitsu, and Takami Matsuo : Adaptive Backstepping Control of Wheeled Inverted Pendulum with Velocity Estimator, International Journal of Control, Automation and Systems,12-5 (2014).

Yuji Maruki, Hiroya Oka, Haruo Suemitsu, and Takami Matsuo : Forwarding Control of Cart-Pendulum System by Following Homoclinic Orbit and Stabilizing Cart, ICCAS2014 (2014).
Hiroya Oka, Kohei Kawano, Yuji Maruki, Haruo Suemitsu, and Takami Matsuo : Adaptive Stabilization of Inverted Pendulum by Two-Step Control Method, SICE 2014 (2014).

Kohei Kawano, Yuji Maruki, Haruo Suemitsu, and Takami Matsuo : Computational and Experimental Validation of Partially Backstepping Controller for Inverted Pendulum, ICIC Express Letters, 8-2, 553-559 (2014).
・丸木勇治, 岡 宏弥, 川野晃平, 末光治雄, 松尾孝美：Forwarding手法による倒立振子のスウィング制御,第58回システム 制御情報学会研究発表講演会,342-1,5/21-5/23/2014.

・Yuji Maruki, Kohei Kawano, Haruo Suemitsu,and Takami Matsuo : Swinging and Stabilization of Inverted Pendulum with Homoclinic Orbit, Proc. of SICE2013, 721-724 (2013).
・川野晃平，岡 宏弥，甫立 悠，末光治雄，松尾孝美:倒立振子の部分的バックステッピング制御の実験的検証,第32回計測自動制御学会九州支部学術講演会(2013)
Kotaro Mihara, Junichi Yokoyama, Haruo Suemitsu, and Takami Matsuo : Swing-Up and Stabilizing Control of an Inverted Pendulum by Two Step Control Method, Proc. of 2012 International Conference on Advanced Mechatronic Systems, pp.323-328 (2012)
・横山淳一，三原晃太郎，川野晃平，末光治雄，松尾孝美:2段階制御法を用いた振子の振り上げおよび安定化制御，第3回プラントモデリングシンポジウム，105-112 (2012)
・川野晃平，三原晃太郎，末光治雄，松尾孝美 :２段階制御法を用いた倒立振子の部分的バックステッピング制御,第55回自動制御連合講演会,1629-1634 (2012)

Junichi Yokoyama, Kohtaro Mihara, Haruo Suemitsu, and Takami Matsuo : Swing-Up Control of an Inverted Pendulum by Two Step Control Strategy, Proc. of・ IEEE/SICE-SII2011,1061-1066.

・Tatsuya Nomura, Yuta Kitsuka, Hiroki Nakamura, Haruo Suemitsu, and Takami Matsuo:Adaptive Backstepping Control for a Two-Wheeled Autonomous Robot, Proc. of ICCAS-SICE 2009, 4687-4692 (2009).
Nonlinear Control of a Single Pendulum Based on a Constraint Functional,International Symposium on Advanced Control of Industrial Processes,Seoul,Korea, 8/22-8/23/2005
・Trajectory Tracking of Inverted Pendulums via Lie Derivative of a Scalar Function, SICE Annual Conference 2004 in Sapporo, 8/4-8/6/2004.

Energy is one of the fundamental concepts in science and engineering practice, where it is common to view dynamical systems as energy-transformation devices. The control problem can be recast as finding a dynamical system and an interconnection pattern such that the overall energy function takes desired form\cite{Chung,Astrom,Ortega}. This control scheme is called the energy shaping approach and is the essence of the passivity-based control. Positive definite function such as the storage function plays a key role in the passivity-based control\cite{Byrnes}. Moreover, the trajectory tracking control problem for nonlinear systems is one of the most challenging of the control theory\cite{Mazenc,Slotine}. It can be solved locally for small reference state trajectories using linear control techniques. However, a nonlinear control strategy is required to track bounded trajectories with large amplitude and obtain large basins of attraction. In this issue, the Lyapunov stability theory plays a key tool to derive a time-varying nonlinear controller and to guarantee its stability. Chung {\em et al.}\cite{Chung} discussed the control problems of a swinging pendulum using an energy function. \AA str\”{o}m {\em et al.}\cite{Astrom} investigated an energy control strategies for swing-up a pendulum for a cart-less pendulum under the condition that the input was the acceleration of the pivot. Moreover, Chatterjee {\em et al.}\cite{Chatterjee} dealt with the swing-up and stabilization problem of a cart-pendulum system with a restricted cart track length and restricted control force using generalized energy control methods.

In this paper, we propose a nonlinear state feedback control law to satisfy a constraint on state variables. The constraint is expressed as an algebraic equation containing a scalar nonlinear function such as an energy function or a limit cycle. Computing the time-derivative of the scalar function that defines the constraint of the state variables, we derive a nonlinear controller to regulate the state variable satisfying the constraint. An inverted pendulum system is given to demonstrate the closed-loop performance of the proposed controller. In particular, we do not neglect a viscous friction of the pendulum and derive a cart controller to generate a desired acceleration of the pivot.

An inverted pendulum is one of the typical examples of nonlinear control systems. Saeki proposed a new method to design an exact linearized controller. The controller design is performed in the following two steps: 1) the desired signal of acceleration of the cart is determined for stabilizing the pendulum using the nonlinear dynamic equation of the pendulum. This is a kind of the impedance control methods, 2) the control input to cart is determined so that actual acceleration of cart may follow the desired signal. The controller can be used to expand the range of initial angle from which the system can be stabilized. Yamakawa {\it et al.} generalized the previous method using a potential function and showed that the controller made the range of initial angle larger from which the pendulum can be stabilized. The potential function is one of the energy of systems. Energy is one of the fundamental concepts in science and engineering practice, where it is common to view dynamical systems as energy-transformation devices. The control problem can be recast as finding a dynamical system and an interconnection pattern such that the overall energy function takes desired form. This control scheme is called the energy shaping approach and is the essence of the passivity-based control. Positive definite function such as the storage function plays a key role in the passivity-based control. Moreover, the trajectory tracking control problem for nonlinear systems is one of the most challenging of the control theory. It can be solved locally for small reference state trajectories using linear control techniques. However, a nonlinear control strategy is required to track bounded trajectories with large amplitude and obtain large basins of attraction. In this issue, the Lyapunov stability theory plays a key tool to derive a time-varying nonlinear controller and to guarantee its stability. Chung {\it et al.} discussed the control problems of a swinging pendulum using an energy function. Astrom {\it et al.} investigated an energy control strategies for swing-up a pendulum for a cart-less pendulum under the condition that the input was the acceleration of the pivot. Moreover, Chatterjee {\it et al.} dealt with the swing-up and stabilization problem of a cart-pendulum system with a restricted cart track length and restricted control force using generalized energy control methods.

This paper presents a semi-global stabilizer and a limit cycle generator for a single pendulum. First, an exact linearized input is derived and the swing-up control law is obtained using a potential function that guarantees the convergence of the actual acceleration to the desired value. Next, we propose another nonlinear state feedback control law so as to satisfy a constraint functional on state variables. The functional is expressed as an algebraic equation containing a scalar nonlinear function such as an energy function or a limit cycle. Computing the time-derivative of the functional that defines the constraint of the state variables, we derive a nonlinear controller to regulate the state variable satisfying the constraint. In particular, we do not neglect a viscous friction of the pendulum and derive a cart controller to generate a desired acceleration of the pivot. Finally, the MATLAB simulations are performed to demonstrate the validity of the proposed control laws.

Ａｂｓｔｒａｃｔ(2013):
In this paper, we introduce a backstepping control design of a wheeled inverted pendulum. Based on a second-order motion equation of the body angle, an adaptive integral backstepping controller is designed to stabilize the body angle. It is shown that the $\sigma$-modification rule in the adaptive update law guarantees the boundedness of the errors in estimating the time-varying signal that is an output of a linear system with every bounded input signal. Then, the stabilizing controller for the wheel angle is constructed by a PD-type positive feedback.
The derived controller requires the full-state measurements. In the output feedback case, the $K$ filter or the observer backstepping is needed. However, the structure of the controller becomes complicated. We propose a non-model-based differentiator based on the adaptive update law. Since the non-model-based differentiator does not require any knowledge of the dynamic structure of the signal, we can use it as a velocity estimator for unknown nonlinear systems. Therefore, we replaced the velocity measurement with the estimates by the non-model-based differentiator.  Finally, simulation results for the proposed controller are presented.

### 5)（メカトロニクス系）慣性計測機器を用いたモデルフリー速度・加速度推定器の性能 検証

・岡　宏弥，末光治雄, 松尾孝美：適応微分推定器を用いた勾配計算と最適化計算への応用,第34回計測自動制御学会九州支部学術講演会, 102A3, 11/28-11/29 (2015).
・Yasuhito Eguchi, Tsuyoshi Ohba, Haruo Suemitsu, and Takami Matsuo : Auto-Tuning Velocity Estimator by Using Adaptive Observer, ICIC Express Letters, 8-2, 427-433 (2014).
・Yasuhiro Kawakami, Yasuhito Eguchi, Tsuyoshi Nimiya, Haruo Suemitsu, and Takami Matsuo : Velocity and Acceleration Estimation by Iterative Learning Observer and Perfomance Validation with MEMS-Based Inertial Sensors, Int. J. of Advanced Mechatronic Systems, 5-2, 113-121 (2013).
・大羽　健,江口泰史,末光治雄,松尾孝美：オートチューニング適応微分推定器の安定性とその応用,第57回システム 制御情報学会研究発表講演会，114-6 (2013)
・大羽 健,江口 泰史, 末光治雄, 松尾孝美：慣性計測機器を用いたオートチューニング適応微分推定器の性能検証,第56回自動制御連合講演会,1259-1262 (2013).
・江口泰史，大羽 健，末光治雄，松尾孝美:反復学習オブザーバによる速度・加速度推定器の慣性計測機器による性能検証,第32回計測自動制御学会九州支部学術講演会(2013)
・大羽 健,江口 泰史, 末光 治雄, 松尾 孝美:オートチューニング適応ゲイン則による速度推定器の慣性計測機器による性能検証,第32回計測自動制御学会九州支部学術講演会(2013)Yasuhiro Kawakami, Tsuyoshi Nimiya, Haruo Suemitsu and Takami Matsuo : Performance Comparison of Velocity Estimators with MEMS-Based Inertial Sensors, Proc. of 2012 International Conference on Advanced Mechatronic Systems, pp.195-200 (2012).

Tsuyoshi Nimiya, Masayasu Sakamoto, Yasuhiro Kawakami, Haruo Suemitsu, and Takami Matsuo:Experimental Validation of Adaptive Velocity and Acceleration Estimators with MEMS-Based Inertial Sensors, pp.223-228, Proceedings of the 43rd ISCIE International Symposium on Stochastic Systems Theory and Its Applications, Shiga, Oct. 28-29, ( 2012).

Tsuyoshi Nimiya, Haruo Suemitsu, Takami Matsuo, Sawut Umerujan, and Kazushi Nakano : Velocity and Acceleration Estimation of an Electric Throttle System by Adaptive Differential Filter, Proc. of IEEE/SICE-SII2011,409-414(2011).

また，Levantは，スライディングモード制御を用いた厳密微分推定器が提案している．

1)手動作業によるパラメータチューニングをなるべく避けて，なるべく汎用性のある推定器を設計できる．
2)パラメータチューニングを手作業を行うことにより，通常の定数の適応ゲインよりも，推定性能を向上させることができる．
その有効性を検証するために，1については，スカラー系の2つの定数パラメータ推定問題に適用し，性能を比較する．さらに，2については，慣性計測機器NAV440を用いた実測値を用いて，その推定性能を比較検証する．

### 6)（メカトロニクス系）適応制御手法を用いた自動車制御系の非モデルベース状態推定

Yuta Kitsuka, Tsuyoshi Nimiya, Haruo Suemitsu, and Takami Matsuo:Non-Model-Based Velocity and Acceleration Estimators for a Suspension System with Parallel Connection of a Hydraulic Actuator, proc. of 2010 IEEE Multi-Conference on Systems and Control, pp.549-554 (2010).

Yuta Kitsuka, Tsuyosi Nimiya, Haruo Suemitsu, and Takami Matsuo: Feedback Control of a Suspension System with Non-Model-Based Velocity and Acceleration Estimators, proc. of SICE Annual Conference 2010, 484-489 (2010).
・木束裕太，野村達八，末光治雄，松尾 孝美：適応同定手法による非モデルベース微分推定器設計とアクティブサスペンション系への応用，第54回システム制御情報学会研究発表講演会,547-548 (2010).
・木束裕太，野村達八，末光治雄，松尾 孝美：油圧アクチュエータサスペンション系の非モデルベース速度・加速度推定，SICE九州支部，177-180 (2009).
Yuta Kitsuka, Tatsuya Nomura, Haruo Suemitsu, and Takami Matsuo:Adaptive Estimation of Mass Flow Rate of a Four-Cylinder Spark Ignition Engine, Proc. of ICCAS-SICE 2009, 5224-5229 (2009).

Abstract:
In recent years, several new sensor technologies have been developed and implemented within automotive industries due to the increasing requirements for improved engine performance and emission reduction. It requires detailed and specified knowledge of the combustion process inside the engine cylinder along with a sophisticated technique in engine diagnostics and control.
Internal combustion engines (ICE’s) are controlled based on a manifold absolute pressure (MAP) and mass air flow (MAF) signals that are generated by MAP and MAF sensors, respectively. A controller controls emissions and engine performance characteristics of the ICE based on the MAP and MAF signals. MAF sensors are commercially available and have been used with ICE’s to provide the required MAF information. MAF sensors, however, are relatively expensive as compared to other sensors implemented with the ICE.
The estimation of the internal states of ICE’s such as the mass flow rate is very useful in engine-related studies.
Advantages of a sensorless system are clear, because of the difficulty of installing the sensor in many applications.
Many papers and the patents have dealt with these issues\cite{lee,shin,kim,USpatent}. The conventional techniques are less accurate than desired and sensitive to sensor noises.
We have recently presented a new time-derivative tracker of a nonlinear function based on the adaptive control theory\cite{park}. The proposed differential filter was applied to estimate the time-derivative of a nonlinear function of the intensity (nonlinear intensity) for each pixel to detect moving objects within a scene acquired by a stationary camera. The proposed method allows us to attenuate the influence of noises in estimating the time-derivative of sequential data. However, the time-derivative tracker has the tracking error in fast-varying signals since the estimator is based on the adaptive update law for constant parameters and the fuzzy basis function expansion. Moreover, we proposed another time-derivative tracker whose update law consists of a $signum$ function and compared with the exact differentiator \cite{wada}.

In this paper, we propose the estimator for the mass flow rate, the manifold pressure, and the torque produced by the engine based on the engine angular speed and the spark advance measurements. The proposed estimator requires the estimates of the time-derivative of signals. The need of velocity estimation from measured position data is still a difficult task and a challenging problem. The simplest numerical method for differentiating a signal is the backward difference. However, the backward difference operator has a noise-amplifying characteristic. To obtain a robust estimator to measurement noises, we adopt the adaptive differential filter\cite{wada}. The differential filter does not require the knowledge of the dynamics of systems and are more robust to output noises than the other differentiators.

### 7)（電力系統）適応オブザーバとTwistingを用いた太陽電池パネルのMPPT制御

論文：

・Tsuyoshi Ohba and Risa Matsuda, Haruo Suemitsu and Takami Matsuo: Improvement of EMC in MPPT Control of Photovoltaic System Using Auto-Tuning Adaptive Velocity Estimator, Journal of Robotics and Mechatronics, 27-5, 489-495 (2015),doi: 10.20965/jrm.2015.p0489.
・Tsuyoshi Ohba, Risa Matsuda, Haruo Suemitsu and Takami Matsuo:Improvement of EMC in MPPT Control of Photovoltaic System Using Adaptive Observer
, Int. Conf. on Advanced Mechatronic Systems (2014).

Abstract:
The output characteristic of a photovoltaic array is nonlinear and changes with solar irradiation and cell’s temperature. The Maximum Power Point Tracking (MPPT) technique is needed to maximize the produced energy. Most MPPT techniques contain the time derivative of the current and the voltage. These electric signals are disturbed by the high frequency noises such as the switching noises of the power device. The low-pass filter is employed to reduce the noises of the circuit. However, the estimation error occurs in calculating the maximum power point. Thus, we apply the adaptive observer to estimate the time derivative of noisy signals. Moreover, we improve the incremental conductance method by using the adaptive observer.

### 8)（メカトロニクス系）非線形摩擦の適応推定と制御

・Tatsuya Nomura, Yuta Kitsuka and Takami matsuo : Nonmodel-Based Estimation for Velocity and Acceleration by Adaptive Identification Method, IEEJ Trans. on Electrical and Electronic Engineering, 5-3, 372-374 (2010).
・Kouichi Mitsunaga and Takami Matsuo:Adaptive Estimation of Friction Forces with Differential Filter,Int.J. Systems Science, 40-12, 1263 – 1271(2009).

Kouichi Mitsunaga and Takami Matsuo:Adaptive Compensation of Friction Forces with Differential Filter, Int. J. of Computers, Communications & Control, 3-1, pp.80-89 (2008).
Kouichi Mitsunaga and Takami Matsuo:Adaptive Estimation of Friction Forces under a Sector Condition,Proc. of Second International Conference on Innovative Computing, Infromation and Control (ICICIC-2007)(2007.9)
・Kouichi Mitsunaga and Takami Matsuo:Adaptive Estimation of Friction Forces with Fuzzy Basis Function Expansion,SICE-ICASE International Joint Conference 2006 (SICE-ICCAS2006) (2006.10)

1)Abstract:
Friction is one of the greatest obstacles in high precision positioning systems. Since it can cause steady state and tracking errors, its influence on the response of the systems must be considered seriously\cite{papa}. Many friction models have been proposed that differ on the friction effects that are modeled in a lubricated contact. These models are divided into two categories: the kinetic and dynamic friction models. The kinetic friction models take into account the friction effects such as the viscous friction, the Coulomb friction, and the Stribeck effect. Another category of friction model includes dynamic friction model that embody the natural mechanism of friction generation such as the LuGre model\cite{[10]}.
A number of methods for friction estimation and compensation have been proposed for these models: adaptive control, joint torque control, learning control, variable structure control, and so on.
Tomei\cite{tomei} considered the tracking problem for robot manipulators with unknown parameters and dynamic LuGre model using full state variable information. Sato {\it et al.}\cite{sato} proposed an adaptive friction compensation method with an $H_\infty$ performance using the neural network approximation that is equivalent to the radial basis function approximation or fuzzy basis function approximation. They used the neural network approximation to parameterize the nonlinear characteristics of the dynamic LuGre model. A sliding-mode type error function is introduced that requires the position and velocity measurements. Putra {\it et al.}\cite{putra} proposed an observer-based friction compensation for a class of the kinetic friction models with known system parameters based on the strictly positive real condition. In the case of a partial state measurement, an adaptive estimator that does not require the strictly positive realness of the plants is needed. In this paper, we design adaptive observers to estimate the nonlinear friction model. First, two adaptive observer are designed to estimate the viscous friction force in the position measurement. Secondly, the kinetic friction force and the dynamic friction force are estimated by the fuzzy adaptive observers in the velocity measurement and in the position measurement, respectively. Finally, simulation results for the proposed observers are presented.

2)Abstract:
In this paper, we design adaptive observers to estimate the nonlinear friction model when the output is the position. First, an adaptive observer is designed to estimate the viscous friction force with full-states measurements. In the case of position measurements, the velocity measurement is replaced to the estimate by using the adaptive differential filter. Secondly, the dynamic friction force is estimated by a fuzzy adaptive observer with position measurements. Finally, simulation results for the proposed observers are presented.

### 9)（メカトロ ニクス系）非ホ ロノミック移動ロボットの適応制御

Shuhei Wada, Kouichi Mitsunaga, Haruo Suemitsu, and Takami Matsuo:Model-Based and Non-Model-Based Velocity Estimators and Its Application to Mobile Robots, ICICIC 2007
・Adaptive Control of Nonholonomic Mobile Robots using Control Lyapunov Function, 6th Japan-France Congress on Mechatronics & 4th Asia-Europe Congress on Mechatronics, Saitama, 2003

### 10)（メカトロニクス系）PID+Qコントローラによるリフターの振れ止め制御系設計

・Nominal Performance Recovery by PID+Q Controller and Its Application to Anti-Sway Control of Crane Lifter with Visual Feedback, IEEE Trans. on Control Systems Technology,12-1, 156-166 (2004).
・Robust Anti-Sway Control of Crane Lifter with Visual Measurement by PID+Q Controller,
Proceedings of IMEKO-XV World Congress, Osaka, Vol.X, 41-48 (1999)

Crane lifters are widely used in industrial plants and rehabilitation equipment. To improve the efficiency of cargo handling with cranes, it is necessary to control the crane trolley position so that the swing of the handling load is minimized. Tanaka et al. proposed an automatic dynamic measurement and control system which measures the attitude of a crane lifter by using an accelerometer located on the lifter. Another way  to  measure and control  the lifter is the visual sensing of the attitude by tracking a mark on the lifter with cameras. Though the visual measurement allows us to get an accurate on-line data of the attitude of the lifters, it involves the time delay caused by the image processing and the data transfer. The controller should be robust against the measurement signals with time delays. Ohyama et al.\ designed visual servo controller for an inverted pendulum with the video sensor called by VIDEO TRACKER G220 using the LQG control theory and  the delay compensator. On the other hand,  Matsuo et al. proposed the PID controller-based compensator with a free parameter for multivariable systems and called this compensator the PID+Q controller. It is an important problem in designing a PID controller to achieve a specified  robust performance and/or robust stability, since PID controllers are widely used in industrial processes with time delays or other uncertainties. The free parameter in the PID+Q controller can be used to improve the performance or the stability of the control systems with PID controller.
In this paper, we  point out that the free parameter can be designed to improve the robust stability and the robust performance of the system. To design the robust stabilizer, the first and the second robust stability conditions are introduced. To improve the robust performance of the system, the nominal-performance recovery problem is presented. Moreover, we  apply the PD+Q controller to the robust anti-sway control of crane lifters in order to compensate the visual measurement data with time delays.
The crane lifter system with  the visual sensor is modeled  by a pendulum control system with a time delay. Then, the robust PD+Q controller is designed to the model equation using MATLAB LMI toolbox. Finally, the designed controller  is implemented on a table-size crane lifter kit with the  visual sensor which is same as the experimental equipment in [2]. The simulation and the experimental results show that the proposed controller reduces the load swing in some degree and is robust to the time delays caused by the visual sensor.

### １１）（メカト ロニクス系）適応オブザーバを用いたクアッドロータの外乱および質量推定

Yu Hodate, Yuta Kondo, Haruo Suemitsu, and Takami Matsuo : Disturbance and Mass Estimation for Quad-rotor Flight Control Using Adaptive Estimator, ICT-ROBOT2016, ThCT1.3, Sept 7-9,Bexico, Busan, Korea,2016.

### 12)（情報通信系）非線形適応アルゴ リズムを用いたカオス暗号通信に関する研究

Takami Matsuo, Yusuke Totoki, Hauro Sumitsu:Robustness and Security of H_infinity-Synchronizer in Chaotic Communication Systems, ” Challenges and Paradigms in Applied Robust Control Book edited by Andrzej Bartoszewicz”, ISBN 978-953-307-338- 5(2011).
Tasuya Nomura, Takahiko Irie, Haruo Suemitsu, and Takami Matsuo:Stochastic Security Testing for Chaotic Communication Systems against Error Function Attack, IEEJ Trans. on Electrical and Electronic Engineering, , 6-5, 450-456 (2011).
Takami Matsuo, Yusuke Totoki, Hauro Sumitsu:Robustness and Security of H_infinity-Synchronizer in Chaotic Communication Systems, “Robust Control/ Book 3”, ISBN 978-953-307-562-4.
・H$\infty$-Synchronizer for Chaotic Communication Systems, Int. j. of Bifurcation and Chaos, 18-4 (2008).
・Takahiko Irie, Ikuto Kugimiya, Haruo Suemitsu, and Takami Matsuo:Security Validation of Continuous-Time Chaotic Communication Systems by Error Function Attack,SICE-ICASE International Joint Conference 2006 (SICE-ICCAS2006) (2006.10).
・Yoshinori Toshimitsu, Takahiko Irie, Haruo Suemitsu and Takami Matsuo:Robustness and Security Testing of H$\infty$-Synchronizer in Chaotic Communication Systems,SICE-ICASE International Joint Conference 2006 (SICE-ICCAS2006)(2006.10)..
・Yoshinori Toshimitsu, Haruo Suemitsu, and Takami Matsuo :Robustness and Security Improvement of H -Synchronizer in Secure Communication Systems,
Proc. of International Symposium on Advanced Control of Industrial Processes,Seoul,Korea, 8/22-8/23/05,pp.282-287 (2005.8).
・Zeros and Relative Degree Assignments of Adaptive Chaotic Communication Systems, Int.J. of Bifurcation and Chaos, 14-12, 4233-4247 (2004)
・カオス暗号通信系への制御理論的接近，計測と制御，42-12，1044-1051 (2003.12)
・カオス暗号通信における非線形復号器の設計，電気学会論文誌C, 9月号，2003
・A Design of Chaotic Cryptosystem Based on Linear Controller Parameterization, Proc. of SSS (2003)
・Adaptive Decrypter of Chaotic Communication Systems Based on Riccati Equation, Proc. of Asian Control Conference, Singapore, 2002
・非線形適応アルゴリズムを用いたカオス暗号通信に関する研究， 大分大学ベンチャービジネスラボラトリ成果報告書，2001
・非線形オブザーバによるカオス復号器の設計， SICE制御部門大会，2001

Abstract:
A general methodology for designing chaotic and hyperchaotic cryptosystems has been developed using the control systems theory\cite{[9],[10],[11],[12]}. The chaotic communication system is closely related to the concept of chaos synchronization. An overview of chaotic secure communication systems can be found in \cite{Yang2004}. He classified the continuous-time chaotic secure communication systems into four generations. In the third generation, the combination of the classical cryptographic technique and chaotic synchronization is used to enhance the degree of security. Specifically, Yang {\em et al.} proposed a new chaos-based secure communication scheme in an attempt to thwart the attacks proposed recently \cite{[9],[10]}. They have combined both conventional cryptographic method and synchronization of chaotic systems. Their cryptographic method consists of an encryption function (the multi-shift cipher), a decryption function (the inverse of the encryption function), a chaotic encrypter that generates the key signal for the encryption function, and a decrypter that estimates the key signal. The approach has a limitation since the cryptosystem design may fail if different chaotic circuits are utilized. So far, this generation has the highest security in all the chaotic communication systems had been proposed and has not yet broken. From the control theoretic perspective, the transmitter and the receiver in the chaotic communication system can be considered as the nonlinear plant and its observer, respectively. Grassi {\em et al.} proposed a nonlinear-observer-based decrypter to reconstruct the state of the encrypter \cite{[11],[12]}. They extended the Chua’s oscillator to the observer-based decrypter. The cryptosystem does not require initial conditions of the encrypter and the decrypter belonging to the same basin of attraction. If we can design a decrypter without the knowledge of the parameters of the encrypter, the chaos-based secure communication systems are not secure, because the parameters of the encrypter is selected as static secret keys in the cryptosystem. Parameter identification and adaptive synchronization methods may be effective for intruders in building reconstruction mechanisms, even when a synchronizing system is not available. Therefore, it is important for secure issues to investigate whether adaptive identifiers without the system information of encrypter can be constructed or not.

We have recently designed an observer-based chaotic communication system combining the cryptosystems proposed by Grassi {\em et al.}\cite{[11]} and by Liao {\em et al.}\cite{[12]} that allows us to assign the relative degree and the zeros of its encrypter system \cite{[30]}. Specifically, we constructed three cryptosystems based on a Chua’s circuit by assigning its relative degree and zeros. The cryptosystem consists of an encryption function (the multi-shift cipher), a decryption function (the inverse of the encryption function), a chaotic encrypter that generates the key signal for the encryption function, and a decrypter that estimates the key signal. The proposed cryptosystem allows us to assign the relative degree and the zeros of the encrypter dynamics by selecting an output vector that generates a transmitted signal as partial states of the encrypter. As in Fradkov {\em et al.}\cite{[23],[24]}, we can design an adaptive decrypter for minimum-phase systems with its relative degree 1. Therefore, the encrypter dynamics should be design such that its relative degree is more than two and its zeros are unstable so as to fail to synchronize the cryptosystem adaptively. At the same time, the designed cryptosystem should be robust with respect to uncertainties of the transmission lines such as a time delay, and noises. Suykens {\em et al.}\cite{[31],[32]} presented a nonlinear H$_\infty$ synchronization method for chaotic Lur’e systems based on the dissipativity of nonlinear systems to minimize the influence of the exogenous input such as the message signal and channel noises.

However, many proposed systems with robustness against parameter uncertainties and signal uncertainties are difficult to implement in practice with a reasonable degree of security. The basic difference between the conventional cryptography and the chaos cryptography is that the conventional encryption is defined discrete sets and the chaos encryption is defined on continuous sets. This makes the keyspace behavior of chaotic systems vary different that of conventional systems. Due to the continuous-value property, keys in chaotic cryptosystems form a key basin around the actual secret key. When one key is very close to the real one, it could decrypt part or all of the ciphertext\cite{Alvarez}. To avoid brute-force attacks, a secret parameter should be sensitive enough to guarantee the so-called avalanche property: even when the smallest change occurs in the parameter, the ciphertext will change dramatically\cite{Alvarez}. Various attacks such as the nonlinear forecasting, the return map, the adaptive parameter estimation, the error function attack (EFA), and inverse computation based on the chosen cipher attack, are proposed to recover messages from the chaotic ciphers. Guojie {\em et al.} discussed the secure property of chaos communication based on chaotic parameter modulation from the chosen-ciphertext attack under the Kerckhoff principle. Wang {\em et al.}\cite{Wang} presented the error function attack to evaluate system security as an efficient cryptanalysis tool with a known plaintext and public structure cryptanalysis method. By defining the EFA function, eavesdropper can scan the whole keyspace to find out the proper key that satisfies the EFA function with zero value. They defined the key basin width by the distance between two trial keys located on the two sides of the key basin, which exceeds the average EFA result of the entire keyspace. Anstett {\em et al.}\cite{Anstett} proposed a general framework based on identifiability for the cryptanalysis of chaotic cryptosystems. They also pointed out that cryptosystems involving polynomial nonlinearities are weak against a known plaintext attack.

In this paper, we propose an $H_\infty$ synchronizer in order to improve the robustness of chaotic communication systems with respect to delays in the transmission line based on the standard linear H$_\infty$ control theory. To begin with, we derive an error system between the encrypter and the decrypter and reduce the design problem of the cryptosystem to the stabilization problem of a generalized plant in the robust control theory. Next, we give a synchronizer parameterization and an $H_\infty$ synchronizer based on the robust control theory. Furthermore, the decrypter dynamics is designed via the linear controller parameterization to make the decrypter robust against disturbances in transmission line and/or sensitive to modeling errors of the decrypter. We present two design requirements on the robustness and the security. We need to design the free parameter such that both the requirements are satisfied. Since we cannot get this solution simultaneously, we design the dynamical compensator so as to satisfy the robustness requirement and then check the sensitivity to the key parameter mismatches whether the parameters in encrypter may play the role of the secret key or not, numerically. Finally, the proposed system is compared with that proposed by Grassi et al. using MATLAB simulations.

### 13)（画像処理 系）消化管造影X線写真のコンピュータ支援診断

Nobumichi Nagano and Takami Matsuo:Edge Detection and Feature Extraction of Gastrointestinal Radiographs Using Adaptive Differential Filter and Level Set Method,Int. J. of Advanced Mechatronic Systems, 3-4, pp.279-287 (2011).

Nobumichi Nagano, Takami Matsuo:Computer-Aided Diagnosis of Gastrointestinal Radiographs Using Adaptive Differential Filter and Level Set Method, proc. of Int Conf. on Modelling, Identification and Control, pp.441-446 (2010).
・Nobumichi Nagano, Takami Matsuo:Computer-Aided Diagnosis of Gastrointestinal Radiographs Using Adaptive Differential Filter, Proc. of ICCAS-SICE 2009, 4793-4797 (2009).
・長野宣道，松尾孝美：適応微分フィルタを用いた消化管造影X線写真のコンピュータ支援診断，SICE九州支部，203B4 (2008).

Abstract:
We have recently proposed an estimator to detect moving objects using the adaptive identification method [3]. The estimator is constructed, based on an adaptive differential filter of an intensity nonlinear intensity function. The nonlinear intensity function is introduced to reduce the effect of noises for specified intensity.
In this paper, we discuss the preprocessing methods for the cancer classification. Preprocessing is a task of removing unimportant information from data. We focus on the edge detection method, the segmentation of the detected edges and the feature extraction of the image segments. The adaptive differential filter is applied to detect the spatial derivative of contrast radiographs in order to detect lesions of gastric intestines. First, we modify the differential filter to estimate spatial derivative of the nonlinear intensity functions. The filter is used to detect the edge of the images weighted by the nonlinear intensity functions. Next, the level set method [5,6] is adopted to carry out the segmentation of the edges. Furthermore, we calculate the circularity and the iregularity to extract the histological characteristic features of the cellular image. Finally, the numerical examples are carried out, using the case radiograph of the gastric intestines medical image database produced by National Kyushu Cancer Center [8].

### 14)（画像処理系）微分トラッカーの 画像処理への応用

・朴　淳鉉，松尾孝美：適応同定法による時間微分フィルタと移動体検出への応用，電気学会論文誌C　(2006.11).
Soon-Hyun Park and Takami Matsuo: Time-Derivative Estimation of Noisy Movie Data Using Adaptive Control Theory,
International Journal of Signal Processing, 2-3, pp.170-177 (2005).
・Soon-Hyun Park and Takami Matsuo:Differential Filter for Detection of Moving Objects and Background Estimation Using Adaptive Identification Method,Modeling Decisions for Artificial Intelligence, CD-ROM,Tsukuba (2005).
・Soon-Hyun Park and Takami Matsuo: Adaptive Estimation of 3-D Motion Parameters and Optical Flow, Proc of 2003 IEEE International Symposium on Computational Intelligence in Robotics and Automation, Kobe, pp.1294-1299 (2003).
・Soon-Hyun Park, Haruo Suemitsu and Takami Matsuo: Recursive Detection of Moving Object by Optical Flow Information,　Proc of 6th Japan-France Congress on Mechatronics and 4th Asia-Europe Congress on Mechatronics, Saitama, Japan, pp.309-314 (2003).
・Takami Matsuo, Soon-Hyun Park, Haruo Suemitsu, Kazushi Nakano: Velocity Estimation of Moving Object by Optical Flow Information,　Proc. of International Symposium on Advanced Control of Industrial Processes, Kumamoto, 203-208 (2002).

The necessity to evaluate the time-derivative of signals arises frequently in many areas of research.
In tracking or detecting moving objects in an image sequence, the need of velocity estimation from measured position data still a difficult task and a challenging problem. The motion detection can be performed by calculating spatio-temporal gradient. A large set of motion detection algorithms has already been proposed in the literature. The first one is based on temporal gradient, {\it i.e.} the time-derivative: a motion likelihood index is measured by the instantaneous change in the image intensity computed by differentiation of consecutive frames.
The second one is the background subtraction techniques. In the absence of any a priori knowledge about target and environment, the most widely adopted approach for moving object detection with fixed camera is based on the background subtraction. The principle of these methods is to build a model of the static scene (i.e. without moving objects) called background, and then to compare every frame of the sequence to this background in order to discriminate the regions of unusual motion, called foreground (the moving objects). Many algorithms have been developed for the background subtraction. The most important steps in the background subtraction algorithm are the background modeling and the foreground detection. In particular, background modeling is at the heart of any background subtraction algorithm. Much research has been devoted to developing a background model that is an estimate of background. The background model is computed by the difference between the current frame and the current background model\cite{chcc}. The background model should be robust against environmental changes in the background, but sensitive enough to identify all moving objects of interest. A significant number of the described algorithms use a simple IIR filter applied to each pixel independently to update the background and use thresholding to classify pixels into foreground/background\cite{halevy,pfinder}. In particular, Kalman filter is a widely-used recursive technique for tracking linear dynamical systems under Gaussian noise. The dynamical systems such as the motion vector or intensity are assumed to be Markov models\cite{kuo,karmann}. These Markov models do not have physical meanings. Therefore, the estimator of the time-derivative for a signal, {\it i.e.} the motion vector, is required without any knowledge of its dynamics.
The third type of approach is based on the computation of the local apparent velocity (optical flow) that is used as input of a spatial segmentation. This method is in general more computationally complex and it is sensitive to the reliability of the optical flow.
The fourth method is based on morphological filters. By using spatio-temporal structuring elements, a local amplitude of variation can be computed as motion likelihood index. Such measure can be useful to detect small amplitude motion, but it is sensitive to outliers\cite{richefeu}. In particular, Richefeu {\em et al.}\cite{richefeu} presented a new differential operator based on a hybrid filter, combining morphological and linear operations. It computes a pixel-wise amplitude of time-variation over a recursively defined “temporal window”. It is particularly suited to small and low amplitude motion.

Recently, Ibrir have presented a time-varying linear system to estimate the first $(n-1)$th derivative of any bouded signal\cite{ibrir1,ibrir2}. The time-derivative observer was formulated as a high-gain observer where the observer gain was calculated through a Lyapunov-like dynamical equation\cite{ibrir2}. This paper presents a new differential filter of a nonlinear function based on the adaptive control theory\cite{narendra,sun}. The proposed differential filter is applied to estimate the time-derivative of a nonlinear function of the intensity (nonlinear intensity) for each pixel to detect moving objects within a scene acquired by a stationary camera. The estimated variety of the nonlinear intensity is obtained by using the adaptive parameter adjustment law\cite{narendra,sun}. We select two nonlinear functions, a quadratic function and logarithmic function, in order to attenuate the variety of disturbances with small amplitudes, because the filter should have certain noise immunity such as waving bushes or flowing water. As our differential filter recursively updates the estimate of variety of nonlinear intensity, its computational cost is very low. The proposed method allows us to attenuate the influence of noises in estimating the time-derivative of sequential data. The MATLAB simulations are performed to demonstrate its performance.

### １５）（画像処 理系）オプティ カルフローを用いた移動体の運動パラメータ推定に関する研究

・Adaptive Identifier for Estimating Optical Flow and Position of 3D Moving Object, submitted (2007)
・適応同定法による時間微分フィルタと移動体検出への応用，電気学会論文誌C. (2006.11).
Time-Derivative Estimation of Noisy Movie Data Using Adaptive Control Theory, International Journal of Signal Processing, 2-3, pp.170-177 (2005.10).
・Soon-Hyun Park and Takami Matsuo:Differential Filter for Detection of Moving Objects and Background Estimation Using Adaptive Identification Method, Modeling Decisions for Artificial Intelligence, CD-ROM,Tsukuba, 7/25/-7/27/05.
・Soon-Hyun Park, Haruo Suemitsu, and Takami Matsuo:Depth Estimation of Moving Objects Using Adaptive Identifier,
Proc. of International Symposium on Advanced Control of Industrial Processes,Seoul,Korea, 8/22-8/23/05,pp.233-238 (2005.8).
・Depth Estimation of Moving Objects Using Adaptive Identifier,International Symposium on Advanced Control of Industrial Processes,Seoul,Korea, 8/22-8/23/05 (to appear)
・Optical Flow and Depth Estimators of 3D Moving Objects via Fuzzy Adaptive Algorithms, IFAC Workshop on Adaptation and Learning in Control and Signal Processing (ALCOSP 2004), 8/30-9/1/2004, Yokohama.
・Recursive Detection of Moving Object by Optical Flow Information, 6th Japan-France Congress on Mechatronics & 4th Asia-Europe Congress on Mechatronics, Saitama (2003.9)
・Adaptive Estimation of 3-D Motion Parameters and Optical Flow, Proc. of IEEE Int’l Symposium on Computational Intelligence in Robotics and Automation(Kobe) (2003.7)
・Velocity Estimation of Moving Object by Optical Flow Information,
Proc. of International Symposium on Advanced Control of Industrial Processes,Kumamoto, 203-208 (2002.6)
・オプティカルフローを用いた移動体の速度推定,大分大学工学部研究報告，No.45,51-55 (2002.3).

The recovery of 3-D motion parameters and structure of an object from its 2-D dynamic picture image is one of the most important problems in computer vision. A fundamental problem in the image processing is the measurement of optical flow that is an approximation to the 2-D motion field of an image sequence. The 2-D image motion is the projection of the 3-D velocities of 3-D object points onto the image plane that results from the relative motion of the sensor and/or the environment. Optical flow can be used to infer the 3-D motion and structure parameters of the camera and the scene, to perform motion detection and object segmentation, to perform motion-compensated encoding, and to compute stereo disparity. Mukai et al. proposed a method for recovering motion and structure from perspectively projected optical flow of feature points that move rigidly. Their method allows us to recover the motion parameters such as the velocity and the angular velocity of a moving object except for a scaling factor. Since the scaling factor of the motion parameters is unknown, we cannot get the real motion parameters exactly. When the spatial and temporal derivatives can be computed reliably, the surface geometry can be recovered by computing optical flow. However, the local measurement and computation of optical flow are generally noisy and inaccurate, as opposed to global behaviors of the field.
In this paper, we present a recursive algorithm for the estimation of optical flow, which allows us to detect the edges of moving objects. Moreover, in the constraint equation of the gradient method only parallel motion is considered. If rotate or magnification motion exist, it is difficult to extract the accurate optical flow from sequential images correctly by using local method. Therefore, we propose an estimation method of the speed in the optical axis direction of a moving object using optical flow information estimated by data with a fixed camera. Under the assumption that the size of the moving object is known, we derive the relationship between the speed along the focal axis and the optical flow of a moving object. Since a major problem which has emerged in research on optical flow interpretation is sensitive to noise, it is necessary to develop a robust and simple technique for computing optical flow field or to derive an robust estimator for 3-D motion parameters without optical flow information.
Finally, we present an adaptive estimator of 3-D motion parameters from an image sequence directly.

We discussed the problem of recovering 3-D information from time-varying images \cite{Park1,Park2}. After deriving parametric model to apply the robust adaptive update law for single-input and single-output systems with input nonlinearities\cite{Matsuo1,Matsuo2}, we presented an adaptive estimator of 3-D motion parameters from an image sequence directly under the assumption that the motion parameters were slowly time-varying. In this paper, we propose an adaptive update law to estimate the time-varying parameters using the fuzzy basis function expansion. The proposed adaptive law allows us to estimate the time-varying 3-D motion parameters and its optical flow without gradient operations of image sequences. In particular, we give a fuzzy adaptive update law to estimate unknown time-varying 3-D parameters. Moreover, adaptive algorithms to estimate the optical flow and depth of a moving object are presented. Finally, we will conclude by demonstrating simulation results via MATLAB/Simulink.

(1) The recovery of the 3-D motion parameters and structure of an object from its 2-D dynamic picture image is one of the most important problems in computer vision. The aim of this paper is to propose a real-time estimator of 3-D motion parameters, 2-D velocity, and 3-D position without the spatial and temporal derivatives. First, we present a robust adaptive update law that allows us to estimate time-varying parameters by using the fuzzy basis function expansion. Next, we design adaptive estimators of 3-D motion parameters and 2-D velocities from image sequences directly without differential operations. Furthermore, under the assumption that the motion parameters are known, a depth estimator is given by a first-order adaptive identifier. Finally, simulations in MATLAB/Simulink are carried out to demonstrate the performance of the estimators.
(2) This paper presents a new algorithm to detect moving objects within a scene acquired by a stationary camera. The algorithm consists of an adaptive differential filter of an intensity nonlinear function and a decision rule to detect moving objects. We adopt two nonlinear intensity functions to reduce the influence of noises. The derivatives of the nonlinear intensity functions are estimated by an adaptive observer with $\sigma$-modification update law.

### 16)（メカト ロニクス系）ス パイラルベクトルを用いた誘導機およびブラシレスDCモータのロバスト適応制御

・Robust Control of Brushless DC Motors by PID+Q Controller, 6th Japan-France Congress on Mechatronics & 4th Asia-Europe Congress on Mechatronics, Saitama, 2003
・Fuzzy Adaptive Position Control of Brushless DC Motors with Spiral Vector, to appear in Proc. of Asian Control Conference, Singapore, 2002
・Robust Adaptive Control of Induction Motors with Spiral Vector, to appeared in IFAC World Congress 2002.
・Robust Adaptive Control Based on Riccati Equation and Its Application to DC Servomotor， Proc. of IEEE IECON’00,CD-ROM, 446-451 (2000)
・スパイラルベクトルに基づく切り換え型適応制御系設計とモータ制御への応用，精密工学会第5回知能メカトロニクスワークショップ pp.197-203，(2000)
・スパイラルベクトルに基づく適応制御系設計と過渡応答性能，第19回適応制御シンポジウム,平成11年1月

1)定常電気回路の解析にはフェーザが用いられることが多い．これは制御理論における周波数伝達関数に対応している．ところが，電気回路の過渡現象を扱う 際には定常解と過渡解の整合性が悪く，両者を結合させるのに不便なことが多い．このため，山村は負の実部をもつ複素指数関数を用いることにより，過渡一般 解と定常解を統一的に表現する方法を提案し，スパイラルベクトル理論と名付けたが，解を仮定するような手法であり，ラプラス変換や演算子法のほうがすぐれ ているといった指摘がなされ，十分に受け入れられているとは言い難い．本報告では，制御理論で減衰度指定のために用いられる虚軸シフトとスパイラルベクト ルの関係を述べ，さらに，適応制御系設計において減衰度向上のために虚軸シフトを行う方法を提案する．最後に，直流サーボモータの速度制御系に本手法と Riccati方程式に基づくパラメータ調整則\cite{matsuo}を適用し，その有効性を確認する．

2)本報告では，制御理論で減衰度指定のために用いられる虚軸シフトとスパイラルベクトルの関係を述べ，さらに，適応同定において減衰度向上のために虚軸 シフトを行う方法を提案する．適応同定とは，プラントのパラメータを入出力データから適応的に推定する手法のことである．適応同定において強正実性は重要 な役割を果たしている．適応同定機構の誤差システムが強正実条件を満たしているならば，グラディエント型のパラメータ調整則により，誤差の漸近的安定性を 保証することができる．しかしながら，相対次数が2以上の誤差システムでは強正実性は満足されず，このような場合には，プラント出力と同定器の出力との誤 差信号だけでなく，プラントの状態変数と同定器の状態変数のすべてを利用したパラメータ調整則が使われている．最近我々は，強正実条件のかわりに1本の Riccati方程式条件を満たす誤差システムの信号の有界性を保証するパラメータ調整則を提案した．このパラメータ調整則を用いることにより，状態変数 すべてを必要とするのではなく，入出力データのみから適応同定器を構成することができる．本報告では，このパラメータ調整則の収束速度をスパイラルベクト ルを用いて

Abstract:Preprints of IFAC World Congress
Our approach has the following features:
\begin{enumerate}
\item The solution for the Riccati equation exists when the relative degree of the plant exceeds unity. Thus, we can deal with plants with a relative degree $>= 2$ in much the way same as with a relative degree = 1.
\item By increasing the exponential part of the spiral vector, we can reduce overshoots of the adaptive control systems.
\item We can reduce the stability of the error system with bounded disturbances to the existence of a positive definite solution for a Riccati equation, and derive an $L_2$ performance condition.
\end{enumerate}

Abstract: Asian Control Conference
\begin{itemize}
\item The solution for the Riccati equation exists when the relative degree of the plant exceeds unity. Thus, we can deal with plants with a relative degree more than or equal to 2 in much the way same as with a relative degree equal to 1.
\item By increasing the exponential part of the spiral vector, we can reduce overshoots of the adaptive control systems.
\item We can reduce the stability of the error system with bounded disturbances to the existence of a positive definite solution for a Riccati equation, and derive an $L_2$ performance condition.
\item The nonlinear functions of the EMF’s of brushless DC motors and the Coulomb friction are approximated by the fuzzy basis function expansion.
\end{itemize}

### １７）（電 力系統）故障検 出器およびロバストコントローラを用いた統合的電力系統安定化システム設計

・厳密線形化によるAVRのロバスト設計，準備中(2002)
・リッカチ方程式に基づく電力系統の適応安定化制御 電気学会論文誌C分冊,121-C-2,366-374 (2001)
・Fuzzy Adaptive Fault Detector of Power Systems Based on Riccati Equation,
Proceedings of IEEE International Conference on Systems, Man, and Cybernetics, Tokyo
VI-179–VI-184 (1999)
・Adaptive Identification of Power Systems Based on Riccati Equation,
Preprints of 11th IFAC Symposium on System Identification, Kitakyushu, pp.1279-1284 (1997)
・An Adaptive Control System via a Riccati Equation and Its Application to Power Systems,
Proceedings of the 22nd IEEE International Conference on Industrial Electronics, Control, and Instrumentation, pp.1933-1938 (1996)

The adaptive techniques have been applied to power systems control by many researchers. Irving et al. designed a model reference adaptive control system for a linearized model around an operating point and their experiment confirmed their controller was effective for network perturbations. Taranto et al. constructed a robust controller for a linearized model around an operating point using $H_\infty$ control theory. However, these designs are not available after a large sudden fault, because a linearized model around an operating point cannot approximate the real nonlinear power system. Recently, Wang et al. designed the adaptive controller for the exact linearized model obtained by the direct feedback linearization.  The exact linearized model by the direct feedback linearization gives the exact solutions of the nonlinear model. Therefore, the controller for the exact linearized model is effective for all operating points. Wang et al. employed the feedback linearization technique and proposed the switching control between the power system stabilizer(PSS) and the automatic voltage regulator(AVR) in exciting the generator. The PSS is used to attain transient stability enhancement when a transmission-line fault occurs. Moreover, Matsuo et al.\ proposed an adaptive nonlinear controller of the exciter without the power angle feedback-path for Wang’s power systems model. We employed the adaptive control design method with a Riccati equation. This method allowed us to design the parameter adjustment law only using the estimate of the power angle. The adaptive nonlinear controller needs the time when a fault occurs since it uses the switching control of PSS and AVR.
Thus, it is important to know when a fault occurs in the transmission lines.
In this paper, we design an adaptive fault detector for transmission lines of the power system using the partial states. To design the fault detector, we propose the new adaptive parameter adjustment law that does not need the full state variable. Two types of the fault detectors are given. MATLAB simulation is done to verify the performance of the proposed fault detectors.
1)Fault detector by exactly linearized model: Using an exactly linearized model of the system that has the regressor generated by the available nonlinear signals, we derive an adaptive identifier based on the proposed adaptation law by fed back the output of delta-detector as a test signal.
2)Fault detector by Fuzzy Model: Using a fuzzy model of the system expanded by the fuzzy basis function, we derive an adaptive identifier based on the proposed adaptation law that does not require the strictly positive realness.

### １８）（電 力系統）SMES による電力系統ロバスト制御

・Robust Transient Stabilizer for Power Systems with Superconducting Magnetic Energy Storage Unit,
Proceedings of the 22nd IEEE International Conference on Industrial Electronics, Control, and Instrumentation, pp.1687-1692(1996)

SMESを含む電力系統のモデリングを行い，ディスクリプタ型非線形状態方程式の形で記述した．ここで，興味深いのは，入力がエネルギーであるため，状態変数の非線形関数となっていることである．この状態方程式を直接用いて，制御するためには，Behavioral approachのような手法が必要であると思っているが，いまのところペンディング状態である．そこで，上記の論文では，通常の状態方程式に変形するために，近似線形化を行い，LTVシステムを導出し，ロバストコントローラを設計した．