Cooperative Control of Recurrent Neural Network for PID-Based Single Phase Hotplate Temperature Control Systems

High-precision temperature control technology is currently more and more important in industrial thermal processing systems. In this paper, an RNN controller with integral-proportional-derivative (IPD) compensation driven by a reference model is proposed for single phase hotplate temperature control systems. A reference model is introduced based on the real controlled plant for the RNN controller to obtain better self-learning and adjusting efficiency by providing a more valuable teaching signal. Further, an Adam optimization algorithm is applied to improve the control performance of the RNN controller. The simulations were developed under a MATLAB environment and the experiments were performed on a temperature experimental platform that used a digital-signal-processor (DSP) as digital controller. The results of simulations and experiments were quantitatively compared with those for a conventional temperature control system which only had an IPD controller. The control efficiency of the proposed RNN method was successfully evaluated.