Iterative Learning-based Fault-tolerant Control Method For Networked System With Quantization And Packet Loss

Repetitive systems(such as: servo motors,reciprocating pumps,etc.)are an important part of industrial systems,and fault-tolerant control of repetitive systems has great significance to the dynamic reliable and safe operation of industrial systems.The iterative learning theory uses the information of the previous iteration to improve the control behavior of the current iteration,iteratively corrects the tracking trajectory,and achieves high-precision tracking and control,which has become the mainstream research method of repetitive systems.However,due to the development of networked and distributed,the fault-tolerant control of repetitive systems needs to address the network environment and solve the problems caused by communication transmission,including quantization problem and packet loss.Therefore,this paper introduces an iterative learning strategy for the problem of fault estimation and fault tolerance control in the network environment,and studies active fault tolerance control of repetitive systems from typical network problems—quantization problem and packet loss.The main works of this article are as follows:(1)Considering the problem of iterative learning fault estimation for discrete-time systems,an observer based fault estimator is established,an objective function is constructed,and the convergence conditions of the fault estimator are designed from the dimensions of the iteration axis and the time axis.Finally,the performance of the fault estimator is verified through numerical experiments.(2)For the quantization problem of networked systems,the expression of the quantization error fluctuation range is determined to adapt to different quantizer errors.On this basis,the tracking error equation of the control system and the PD-type iterative learning controller are built,combined with the fault estimator.Then,the objective function is designed,and the convergence conditions of fault estimator and iterative learning controller are established.Finally,the tracking performance is verified through numerical experiments.(3)For the problem of packet loss in networked systems,the packet loss process is considered as a random switch variable subject to the Bernoulli process,and intermittent update strategy and continuous update strategy are introduced.Based on the iterative learning framework,the fault estimator and controller are constructed respectively,under packet loss.Under the influence of random switch variables,the objective function is constructed,and LMI conditions that meet the convergence of system performance indicators are also analyzed.Finally,the tracking performance is verified through the numerical experiment of the networked robot arm control system.