Research and Design of Reactor Temperature Control System Based on RBF Neural Network

Abstract

Aiming at the time-varying, nonlinear and large time-delay problems of chemical reactor temperature control systems, this study focuses on the continuous stirred tank reactor (CSTR) for aniline hydrogenation to cyclohexylamine, and conducted modeling, advanced control algorithm design and simulation verification of the temperature control system. The temperature control mathematical model was established by fitting process data with Aspen software and parameter identification via MATLAB System Identification Toolbox. Cascade control schemes of traditional PID, BP neural network PID and LPSO-RBF-PID were designed, and their control performances were compared through Simulink simulation. Results show that the proposed control scheme has significantly better anti-interference ability than traditional ones, providing an effective technical solution for precise temperature control of industrial reactors.