The Research On Robustness Of Predictive Functional Control Systems

Model Predictive Control (MPC) is one of the most popular and successful advanced control strategies, which has been widely applied in many fields. As the third generation MPC, Predictive Functional Control (PFC) is a new predictive control algorithm to meet the fast tracking control needs of actual industrial process control. PFC has got a great development in theory and application, but as a control method developed form practice, the theoretical research is much backward relative to the engineering application. The research on stability and robustness of PFC systems is not mature and the problem about performance of close loop is not solved, especially in multivariable systems.The theories and applications of PFC are studied in this dissertation. Based on the basic principles, the detailed state space PFC algorithm for SISO systems is deduced and the stability and robustness are discussed. A single-value PFC method and a dual-value PFC method are respectively proposed for first-order pure delay process which is common in the industrial processes and the stability and robustness are also discussed.A multivariable PFC algorithm based on a two-input/two-output system with the transfer function model is presented and an explicit analytical solution is obtained for control variable, which is promoted into n-input/m-output system. The stability and robustness are discussed by internal model control (IMC) theory, which provides theoretical evidence for the effectiveness of this algorithm.A predictive functional control strategy based on wavelet function is developed and a controller is designed for diesel engine speed regulation system of ship power system with PFC theory using Morlet wavelet as basis function. The whole optimization of predictive horizon is damaged because the basis functions of traditional PFC are almost global functions and can't be set flexibly according to the system's controlling precision and the different requirements of approximation precision at different intervals of the horizon. This method using the properties of compact support and multi-scale analysis of wavelet function, the number and position of basis functions can be set flexibly. The theoretical analysis and simulation results show that this approach has favorable tracking effect (the time to set value decreases from 13.5 to 6) and anti-interference ability (the overshoot decreases from 10.6% to 5.7% when 20% disturbance is introduced) to restrain the influence of parameter variation and load disturbance compared with traditional PFC, also it does better in improving the dynamic precision of diesel engine and enhancing the frequency stability of a ship power system.