Decentralized PID Controllers Based on Probabilistic Robustness

Abstract

A tuning method for decentralized PID controllers was developed based on probabilistic robustness for multi-input-multi-output plants, whose parameters vary in a determinate area. The advantage of this method is that the entire uncertainty parameter space can be considered for controller designing. According to model uncertainties, the probabilities of satisfaction for every item of dynamic performance requirements were computed and synthesized as the cost function of genetic algorithms, which was used to optimize the parameters of decentralized PID controllers. Monte Carlo experiments were used to test the control system robustness. Simulations for five multivariable chemical processes were carried out. Comparisons with a standard design method based on nominal conditions indicate that the method presented in this paper has better robustness, and the systems can satisfy the design requirements in a maximal probability.