基于神经网络的异步电动机离散容错控制

Neural Networks-Based Discrete-Time Fault-Tolerant Control for Induction Motors

提出了一种基于神经网络的异步电动机离散容错控制方法,同时考虑了执行器失效故障和偏差故障。使用欧拉公式建立考虑执行器故障的异步电动机系统离散模型;利用自适应神经网络技术实现异步电动机控制系统的容错控制;结合动态面技术和反步法,解决了反步法应用到离散系统中产生的“计算复杂性”和“因果矛盾”问题。通过Lyapunov稳定性分析,证明了闭环系统是半全局一致最终有界的,仿真结果表明该控制方法在执行器故障发生后仍可保证位置跟踪性能,验证了该方法的有效性。

A discrete-time fault-tolerant control for induction motors was studied,considering actuator fault containing both loss of effectiveness and bias.The continuous dynamic model of induction motor drive systems considering actuator fault was transformed into discrete-time form by using Euler formula.The adaptive neural networks was employed to realize fault-tolerant control of induction motors.Dynamic surface control was used to overcome the"explosion of complexity"and noncausal problem emerge in the design process of traditional backstepping method.It is proved by Lyapunov theorem that the closed-loop system is semi-global uniform and ultimately bounded.Simulation results show that the proposed approach can guarantee the position tracking performance after the actuator fault occurs,which verifies the effectiveness of the method.