Iterative Learning Control Over Fading Channel

The networked control structure can effectively enhance the robustness,stability and flexibility of the entire system,which is widely used in control systems.However,the networked control structure introduces various types of uncertainties,such as data packet loss,communication delay,sampling and fading channels,and channel noise.In particular,fading channel involves damage of transmitted data over certain propagation media due to attenuation.Compared with other randomness caused by the channel(such as data packet loss and communication delay),the fading channel has attracted less attention from the control community.Therefore,this paper mainly studies iterative learning control over fading channels.The main work of this paper includes:(1)The iterative learning control of linear systems over fading channels is studied.Fading communication introduces multiplicative randomness to the transmission signal,resulting in biased information.A method of directly correcting the received measurement value using the statistics of the fading phenomenon is proposed.A P-type learning algorithm is proposed to deal with the output fading situation and the input fading situation.The proposed control schemes have been proved to converge in the mean-square sense.Simulation examples verifies the effectiveness of the algorithm.(2)The learning control strategy of the network stochastic system in which the output and input data are transmitted through multiple independent fading channels is studied.According to where the fading channel appears.The traditional P-type learning control scheme was improved respectively,using variable learning gain instead of constant learning gain and introducing average operator to suppress the influence of various uncertainties.Under the phenomenon of random fading and system noise,zero error convergence of the input error can be achieved by the algorithm.Simulation examples verifies the effectiveness of the algorithm.(3)The iterative learning control of stochastic linear systems where system information is known over fading channels is studied.The Kalman filtering technique was applied to design an optimal time-and iteration-varying learning gain matrix.The proposed control schemes have been proved to converge in the mean-square sense.Two illustrative simulations are provided to verify the theory.(4)The point-to-point tracking problem of network stochastic systems with fading communication based on iterative learning control is studied.An auxiliary matrix is introduced to illustrate the relationship between the output points in the entire time and the required tracking target.Fading communication introduces multiplicative randomness to the transmission signal,resulting in biased information.This chapter considers two different initial states,and proposes a corresponding learning control scheme,introducing a decreasing gain to ensure stable convergence under various random conditions.A simulation example verifies the effectiveness of the algorithm.(5)The batch learning consistency tracking problem of multi-agent systems with fading neighborhood information is studied.A method of directly correcting the received measurement value using the statistics of the fading phenomenon is proposed.On this basis,a distributed learning scheme of variable gain sequence is proposed to achieve the control goal.The linear system and the nonlinear system are studied separately,and the convergence analysis is carried out in the sense of mean square.