Data driven output joint probability density function control for multivariate non-linear non-Gaussian systems

This study presents a novel data-based joint probability density function (JPDF) control strategy for multivariate non-linear non-Gaussian stochastic systems so that the output JPDF of the system can be made to follow a desired JPDF. The output JPDF, which is usually immeasurable, is estimated according to the output sequence of the system. The multi-step-ahead cumulative performance index is constructed with respect to the control objective and is minimised based on an intelligent optimisation algorithm. By minimising this performance function, a new predictive controller design algorithm is established with more simple formulation and less computation load than existed results. Furthermore, a new approach is developed to guarantee `convergence in distribution´. Finally, simulations are given to demonstrate the effectiveness of the proposed control algorithm and some desired results have been obtained.