The Research On Multivariable Internal Model Control Algorithm

Nowadays, advanced control technology has been applied widely in many areas of industry, and it is obvious that this technology brings great economical effects. Although single loop PID controllers are still being employed in about 80% to 90% of the control systems.However, the PID controllers could not be competent in controlling some complex systems. Expecially for the MIMO systems, the processes usually exhibit severe interaction effects and consequently are very difficult to control, thus the single loop PID could not deal with this problem effectively. So the strategies for MIMO processes have been studied much, in the hope that the effective closed loop control of coupling systems can be achieved, and the economy benefits can be enhanced, the energy consumption will be reduced, and the high quality of products could be guaranteed.IMC technique is chosen to control the MIMO systems here. During decades of research and application, IMC is found to be of many advantages, such as excellent servo performance, high robustness, particularly in case of model uncertainty, effective hog-tieing, and so on.Otherwise, variables interact and constraint each other in some complex industry systems. Sometimes operational variables cannot control systems in time because of being constrainted, as a result, the structure of control systems transform between square and non-square systems. From the outlook of a device or an unit, non-square system is also common, such as fat systems whose vector of operational variables is more than that of control variables. It is not rare that time delay problems exhibit in industry processes. Usually, for MIMO systems, time delay of every loop is different. The focus puts on how to track the inputs without coupling for multi-time delays systems.Internal model controller has been designed for non-square systems in this paper. For the fat systems, the number of operaitonal variables is more than that of control variables, so the generalized inverses of matrices is used to design the internal model controller by caculation in frenquency domain. The simulation results are presented showing good performance about dynamic decoupling, multi-time delays control, the response velocity, and the robust. For the thin systems, the number of operaitonal variables is less than that of control variables, so selection control system is used to carry out complex control performance. Precise control and borderline control are chosen according to different control variavles. Then safe production and production quality can be ensured by switching controllers. Good performance is showed by the simulation results.