控制理论(社会学)
李雅普诺夫函数
Lyapunov稳定性
滑模控制
理论(学习稳定性)
控制工程
机械系统
工程类
计算机科学
控制(管理)
非线性系统
物理
人工智能
量子力学
机器学习
作者
M. R. Homaeinezhad,F. FotoohiNia
出处
期刊:Robotica
[Cambridge University Press]
日期:2021-08-09
卷期号:40 (4): 1168-1187
被引量:2
标识
DOI:10.1017/s0263574721001004
摘要
Abstract In dynamically switched systems with unknown switching signal, the control system is conventionally designed based on the worst switching scenario to ensure system stability. Such conservative design demands excessive control effort in less critical switching configurations. In the case of continuum mechanics systems, such excessive control inputs result in increased structural deformations and resultant modeling uncertainties. These deformations alter differential equations of motion which cripple the task of control. In this paper, a new approach for tracking control of uncertain continuum mechanics multivariable systems undergoing switching dynamics and unknown time delay has been proposed. Control algorithm is constructed based on the mathematical rigid model of the plant and a Common Lyapunov Function (CLF) is proposed upon sliding hyperplane regarding all switching configurations. Considering the model-based nature of sliding mode control (SMC) and inevitable uncertainties induced from modeling simplifications of continuum system or parameter evaluation errors, Finite Element Analysis (FEA) is utilized to approximate total model uncertainties. To obtain robust stability, instead of conventional switching functions in the construction of control law, the control inputs are selected such that system dynamics reside within stability bounds which are calculated based on the Lyapunov asymptotic stability criterion. Therefore, the unwanted chattering issue caused by continuous switching is not observed in control input signals. Eventually, the accuracy of the proposed method has been verified through the student version of ANSYS ® mechanical APDL-based simulations and its effectiveness has been demonstrated in multiple operating conditions.
科研通智能强力驱动
Strongly Powered by AbleSci AI