In this paper, critical analysis of performance comparison of various optimization techniques for a MRPID (Multi Resolution Proportional Integral Derivative) thermal system controller is presented. This controller controls the uncertainty or mismatch between real and reference temperature. Wavelet coefficients of error between the two temperatures and their corresponding gains are added to produce a control signal for the thermal system. The main purpose of the proposed method is to optimize the tuning parameters of a wavelet-based MRPID controller to regulate the switching pulse of the thermal system. Modeling and simulations are carried out using MATLAB/Simulink@2015. Figures-of-merit such as disturbance rejection, system stability and transient response are also compared using various techniques like Particle swarm optimization (PSO)-MRPID, Genetic Algorithm (GA)-MRPID (existing techniques), Firefly-MRPID controllers and Elephant herding optimization (EHO) based Recurrent neural network (RNN)-MRPID algorithm. Firefly-MRPID and EHO with RNN-based MRPID are proposed for determining unique (best) operating conditions of the thermal system in this paper.

在本文中，对 MRPID（多分辨率比例积分微分）热系统控制器的各种优化技术的性能比较进行了批判性分析。该控制器控制实际温度和参考温度之间的不确定性或不匹配。将两个温度之间的小波误差系数及其相应的增益相加，以产生热系统的控制信号。所提出方法的主要目的是优化基于小波的 MRPID 控制器的调整参数，以调节热系统的开关脉冲。使用 MATLAB/Simulink@2015 进行建模和仿真。还使用粒子群优化 (PSO)-MRPID、遗传算法 (GA)-MRPID（现有技术）、Firefly-MRPID 控制器和基于循环神经网络 (RNN)-MRPID 算法的大象放牧优化 (EHO)。本文提出了 Firefly-MRPID 和具有基于 RNN 的 MRPID 的 EHO 来确定热系统的独特（最佳）运行条件。