Study On Internal Model Control Of Non-Square Systems

Chemical and industrial practice often encounter in a class of non-square processes which have unequal number of inputs and outputs. Increasing industrial control requirements, constraints in inputs and the changes of the operating conditions are all the reasons that non-square systems exist. For this kind of system is quit different from the normal systems, the traditional control methods are not adaptive to them.As one of the advanced control methods, internal model control (IMC) is famous for the simple structure, rapid response and high robustness. With decades of research and application, the theoretical structure of IMC has been founded. In this paper, it will be extended to the non-square process. The main contributions of the paper are as follows:Firstly, the structure, principle and design method of IMC strategy for multivariable systems are introduced. Then, the robustness of the system is analyzed. By simulation of a two-input two-output process, the efficiency of IMC is proved.Secondly, the "squared" methods based on loop pairing are discussed, including RGA, ERGA and NRGA. A new loop pairing method used for non-square system directly is proposed.Several methods of synthesizing a controller for the original non-square system are compared, including Moore-Penrose inverse control, multivariable coordinated control, and centralized control and so on. The possibility of their combination with IMC is discussed.A design method of decoupling IMC is proposed for non-minimum phase non-square processes with different time delays. By inserting compensated terms which are derived by analytical design, the realizable decoupling controller is obtained. Meanwhile, a control model reduction method based on NLJ algorithm is introduced to simplify the controller matrix. The simulation results show its decoupling capability and robustness.At last, the application of IMC-PID in non-square processes is studied.