Neural Network Based Self-Tuning PID Controller

Ziegler and Nichols developed the well-known Ziegler-Nichols approach to modify PID controller coefficients in order to find conventional PID controller parameters that are adequate. This tuning approach is straightforward and provides constant coefficient values, resulting in a PID controller with poor adaptability to model parameter variations and changing operating circumstances. Therefore, in this study, a self-tuning PID controller based on Neural Networks (NN) is proposed. The combination of a neural network and PID control compensates for the simplicity of this method of control with artificial intelligence, hence making it robust to dramatic parameter changes. The suggested adaptive PID controller is evaluated in the MATLAB/Simulink environment using sinusoidal and step signals as the desired outputs. In this paper, neural network PID controller is used for disturbance rejection. The artificial neural network enables parameter adjustments for the PID controller to be made in a dynamic manner. At the end of this work, it can be concluded that neural network based PID controller has a robust response with time-delay distribution.