On-line Multistep Predictive Control Of Tempreture Based On Matlab And Kingview

Predictve control is one of the most advanced control methods in modern control theories. Classical PID method is very simple and convenient in application., but its precision is not high. In recent years, the auto-adaptation method and the auto-correction method have resolved the indetermination problem to some extent, but their essence is that still requests to on-line cognize the model of the object, and the adaptive ability is weaker. The robust needs to be improved. Predictive control method includes their virtue. As a excellent arithmetic, it faces industry process characteristics, and its model precision may be not very high, on-line calculation is convenient. The development of digital computer toward the small scaled, high-speed and big capacity, low cost direction, provides the material foundation for the realization of this new calculate way.The paper applies predictive control theory into practice on the base of PCT-III process control system device. The device is made up of water system and temperature system. The experiment adopts the temperature system. The control software is KingView produced by Yakong Company in Beijing. It is very good at designing interface and communication, can write simple program too. But its ability is very limited for multi-step temperature predictive control because of its poor calculation and program ability. So MATLAB is introduced into the experiment. MATLAB is very good at calculating and programming. The problem that need to be resolved is the data exchange of the two kinds of software. Fortunately the two kinds of software both support dynamic data exchange (DDE), so DDE will be a key problem.There are several predictive control arithmetic. Multi-step temperature predictive control experiment adopts dynamic matrix control method(DMC). The method is very convenient to establish predictive model. The predictive model that can descript the dynamic state behavior of process may be received through a simple experiment. The internal mechanism of the arithmetic don't need to be understood deeply. The DMC method adopts rolling and optimization strategy. The process of optimizationis not onetime off-line implement, but on-line calculation again and again and roll over carry on. It can make up for indetermination caused by unmatched model, time variation and disturbance in time. Moreover, one more outstanding character is to correct the model error through feedback. It compares the practical output with model output, and the roll over optimization is established on base of feedback of error. It improved the precision and robust of control system.The paper mainly includes five parts, which studies theory of predictive control method and how to connect the KingView and MATLAB, brings the advanced theory into practice. The experiment is very successful, which shows that predictive control method is really excellent in industry control.