Advanced 5G system chip MOCVD process parameter control and simulation based on BP neural network and Smith predictor

Abstract

in the advanced 5G chip process system, MOCVD is one of the key film processes in modern wafer manufacturing. MOCVD technology has a very high application value and broad application prospects that mainly used in semiconductor materials, nano materials and other material growth fields. The application field of MOCVD makes its process require high accuracy of temperature, flow, pressure and other parameters control. However, based on the conventional Smith predictor, this paper simulated the system with Simulink, and found that the system has overshoot and long regulation time, and the control effect is not ideal. In view of MOCVD parameter control, the BP neural network theory is combined to improve the controller, and a Smith predictive controller based on neural network is obtained. The simulation results show that the intelligent PID control based on neural network has a good effect on MOCVD temperature, flow and other parameters control, with lower overshoot and less system regulation time.