Research On Multivariable System With Multiple Time Delays Based On Internal Model Control

Multivariable system with time delays is common used in industrial processes, while the traditional control methods for this type of process are often hardly to achieve the desired control effects. Because of increasing industrial control requirements, non-square multivariable systems which needed higher requirements of control method are becoming more common. Therefore, the researches to find effective and practical control algorithm for these complex controlled objects is one of the hotspots of the current research in the field.With decades of researches and applications, internal model control(IMC) for multivariable systems has the advantages of simple structure, less parameters and higher robustness. In this dissertation, aiming at multivariable systems with time delays and taking internal model control as the basic control method, some solutions and improvements are proposed for the inadequacies of the control methods for these complex systems.Firstly, the principle, structure and design method of internal model control for single-variable system are systematically represented. A variety of time delay approximation methods had been compared. Then, the characteristics and decoupling methods of multivariable system, which are the foundation of the research In this paper, are analyzed.Secondly, research on model reduction and parameter optimization based on genetic algorithm are given. For making the objective function more reasonable, local variable weight is applied to a modified objective function. Compared with the traditional objective function, the modified function is more comprehensive, flexible and practical.In addition, a modified two-degree-of-freedom internal model control(M-TDOF-IMC) is proposed for non-square systems with multivariable time delay. By designing the closed-loop transfer function, the unrealizable factors of the derived controller are removed. In the proposed control structure, both the set-point tracking responses and the load disturbance rejection responses can be tuned by two controllers separately at the same time. The proposed control system not only has good decoupling performance and tracking performance, but also has good disturbance resistance. Then, applied parameter optimization based on genetic algorithm to M-TDOF-IMC. Through simulation analysis, the optimized control system has better control performances.At last, for the calculation of the pseudo-inverse is large and complex, a simplified control method for non-square multivariable systems based on the static decoupling is given. The coupled problems of non-square process are solved by the pseudo-inverse of model steady-state gain matrix. Appling internal model controller to SMITH predictor control directly by take advantage of equivalence between internal model control structure and SMITH control structure. Thus, eliminating the need for translate internal model controller into PI controller, while reducing the number of controller parameters. The simulation studies shows that the modified method is more direct, at the same time it has a wider range of applications.