Implementation of a Wavelet-Based MRPID Controller for Benchmark Thermal System

This paper presents a comparative analysis of the intelligent controllers for temperature control of a benchmark thermal system. The performances of the proposed wavelet-based multiresolution proportional-integral derivative (PID) (MRPID) controller, which can also be stated as a multiresolution wavelet controller, are compared with the conventional PID controller and the adaptive neural-network (NN) controller. In the proposed MRPID temperature controller, the temperature error of actual and command temperatures of a thermal system is decomposed into different frequency components at various scales of the discrete wavelet transform (DWT). The wavelet-transformed coefficients of temperature error at different scales of the DWT are scaled by their respective gains and then are added together to generate the control signal for the thermal system. The performances of these intelligent controllers are investigated in both simulation and experiments for different operating conditions of the thermal system. The performances of the wavelet-based MRPID controller are found superior to the conventional PID and adaptive NN controllers for temperature control of the thermal systems.