模糊推論積分型滑動模式之小腦模型控制器設計

Design of Fuzzy based Integral Sliding Mode Using Cerebellar Model Articulation Controller

張冠文(Kuan-Wen Chang)

指導教授 ： 洪欽銘

國立臺灣師範大學/科技學院/工業教育學系/碩士(2001年)

摘要

本文針對傳統可變結構控制(Variable Structure Control，簡稱VSC)之
滑動(sliding)模式下會發生高速顫動(chattering)現象，將使系統產生不想要
的高頻成分，甚至造成系統不穩定，提出模糊推論積分型滑動模式之小腦
模型控制器(Fuzzy based Integral Sliding Mode Using Cerebellar Model
Articulation Controller，簡稱CFISMC)。滑動模式控制器具有系統參數變動
及雜訊干擾之不敏性，擁有強健性控制，積分器可有效消除系統穩態誤
差，並能夠提升系統控制穩定度，由模糊控制加入可簡化設計系統的複雜
性，控制法則簡單，易於實現並利用狀態點與滑動面之距離，進行適應性
控制增益之動態調整，使狀態點能快速到達滑動面並降低超越量，以提昇
系統之暫態響應品質，並引用小腦模型控制器加入，可輔助積分型滑動模
式控制器之系統架構，小腦模型控制器(CMAC)，是應用查表方式之類神
經網路， 對非線性系統具備優越之快速學習收斂速度及類化
（generalization）能力，以補償積分型滑動模式控制器，因系統控制器設
計之限制而使系統控制效能品質不理想。此外亦希望經由小腦模型輔助控
制器的加入，能夠縮短系統上昇時間並簡化系統設計之目的，以減少暫態
響應時間並使系統快速到達穩定狀態。最後並將本研究之架構模擬於球體
平衡桿系統(Ball-on-Beam Balancing) 控制與雙倒單擺系統(Tandem
Pendulum)控制，以驗證其控制效能。

關鍵字

可變結構控制 ； 滑動模式 ； 模糊推論 ； 強健性 ； 不敏性 ； 小腦模型控制器 ； 類化能力

並列摘要

The sliding mode causes high speed chattering phenomenon in the
traditional Variable Structure Control (VSC) , and produces unwanted high
frequency in the system and even creates instability. This paper proposes
Fuzzy based Integral Sliding Mode Using Cerebellar Model Articulation
Controller, abbreviated as FCISMC. The integral controller effectively clears
up errors in the system stability, and the sliding mode controller has invariable
property with variation in the system parameters and interfering surface noise,
thus excelling in robust control. By adding the fuzzy logic controller, it
simplifies system difficulty in design. Fuzzy logic control rules are simple to
make and easy to implement. We can regulate the control gain by the distance
between state point and sliding surface. In this way, the state point can reach
the sliding surface rapidly and reduce the overshoot. The transient response of
the system will then be improved. By adding the Cerebellar model articulation
controller, it aids the system structure of the integral sliding mode controller.
Also, with the surpassing nonlinear learning ability of Cerebellar Model
Articulation Controller (CMAC) and its sample generalization ability, it is
hoped to compensate the poor control efficiency caused by design limitation in
the conventional ISMC. Besides, adding the CMAC shortens the design
procedure and reduces the difficulties in design. This cuts down the
temporary state respond time and enables the system to reach its stable state.
Finally, one experiment for the integral sliding mode with CMAC is simulated
with the Ball-on-Beam Balancing control system and the Tandem Pendulum
control system to demonstrate the improvement in its control performance.

並列關鍵字

Integral Sliding Mode, ； Sliding mode, ； robust, ； invariance property, ； Cerebellar Model ； Articulation Controller, ； generalization

參考文獻

[3] Albus, J.S. Data Storage in the Cerebellar Model Articulation Controller (CMAC).

[4] Albus, J.S. A New Approach to Manipulator Control: The Cerebellar Model

Algorithm to the Control of Robotic Manipulators. The International Journal of

Robotic Research, 6, 2, 84-98,1987.

[6] Shelton R. O. & Peterson & J. K. . Controlling a Truck with an Adaptive Critic CMAC