Probabilistic tracking control for non-Gaussian stochastic process using novel iterative learning algorithms

A new generalised iterative learning algorithm is presented for complex dynamic non-Gaussian stochastic processes. After designed neural networks are used to approximate the output probability density function (PDF) of the stochastic system in the repetitive processes or the batch processes, the complex probabilistic tracking control to the output PDF is simplified into a parameter tuning problem between two adjacent repetitive processes. Under this framework, this article studies a novel model free iterative learning control problem and proposes a convex optimisation algorithm based on a set of designed linear matrix inequalities and L1 optimisation index. It is noted that such an algorithm can improve the tracking performance and robustness for the closed-loop PDF control. A simulated example is given, which effectively demonstrates the use of the proposed control algorithm.