Minimum entropy tracking control for non-Gaussian systems using proportional-integral strategy

This paper proposes a practical PI control strategy to realize minimum entropy tracking control for general nonlinear stochastic systems with non-Gaussian noises. As the extension of the minimum variance control for linear Gaussian systems, minimum entropy control aims at decreasing the uncertainties of nonlinear non-Gaussian systems. The whole control process is divided into some batches. Within each batch, we employ a coefficient-fixed PI controller, compute the probability density functions of the tracking error based on the distributions of input noise, and then construct the sampling instant performance index and the overall performance index respectively. Between two adjacent batches, an iterative learning control based algorithm is used to adjust controller coefficients and improve the system tracking performance batch by batch. In addition, the stability analysis is performed to guarantee the boundedness of system variables. Finally, simulations are given to demonstrate the effectiveness of the proposed approach.