Study And Improvement Of Predictive Control Algorithms Based On PAC

The concept of Programmable Automation Controller (PAC) was proposed by the automated research institutes from the United States. PAC is a controller which is based on the industry standards and has universal open platform. It combines the advantages of PLC and PC control. Once it has been introduced, major automation companies, like GE, Rockwell Automation, NI and others have launched their own PAC products.Control algorithms which run in PAC also receive increasing attention, predictive control algorithm is a research direction. This control algorithm does not require a very precise mathematical model and its robustness is good. Now it has been successfully used in chemical, petroleum, electric power and other industries. The widely applicable predictive control algorithms are Dynamic Matrix Control (DMC), Model Algorithm Control(MAC) and Generalized Predictive Control (GPC).This paper focuses on the research of these three predictive algorithms which are used in PAC and an algorithm improvement-the combination of fuzzy control and predictive control. First of all, the background and present situation of PAC system, predictive control and fuzzy predictive control are introduced in this paper. And then the principles of algorithms and fuzzy control are detailedly explained. Based on these theoretical knowledge, I design the Predictive Control Simulation System in MATLAB. This system includes GUI interface and MATLAB programming. In the PAC system, algorithms require C language form, so it is important to program C language algorithms in this research. The outputs of C language algorithms and simulation system algorithms are compared in this paper. Robustness and algorithm computation time are two problems that must be solved for algorithms' application in PAC. Several industrial methods are described to improve robustness in this paper. They are very useful for algorithm research. Fuzzy predictive control which is based on fuzzy rules optimization is proposed for saving running time. I researched the simulation of fuzzy predictive control and C programming, and then tested the running time of fuzzy predictive control and predictive control. The results show that fuzzy predictive control can save a lot of computation time. In the last part of this paper, I conclude the study results and propose the deficiency and improvement direction.