The Method Study Of Temperature Control Based On Multi-chips MOCVD System

This paper first briefly introduces the development of control theory and temperature control theory, and then introduces the MOCVD system, as well as the temperature control object, control mode and control characteristics of the Multi-chips MOCVD equipment. The equipment is characterized as " non-linear ", " large time-delay ", so it is very necessary to make research on the methods of the temperature precise control of the system.According to the principles of system identification, analysis the experimental data and gets dynamic model of the temperature control system. To the dynamic model, we firstly study the typical control algorithm used in temperature control: PID Control, Internal Model Control, Fuzzy-PID Control, Smith-predictor Control, and under the condition of the SIMULINK in MATLAB we compares advantages and disadvantages of these methods. In order to propose adaptive algorithm, the neural network is studied in the paper. The simulation of the BP neural network algorithm and the neuron algorithm of Smith-predictor shows that the neural network algorithm is the direction of our further research.