Adaptive Direct Current Motor Proportional Integral Derivative Controller based on Deep Learning Model

Proportional–Integral–Derivative (PID) controller is widely used in industries due to its implementation method is simple. There is a challenge to optimize the PID gain values. The traditional optimization methods are based on algorithm-based that has poor flexibility and scalability. It is because algorithm-based is predicted according to the conditions. This work focuses on implementing a deep learning(DL) model into the PID controller that predicts the combination of the gain values. This proposed method aims to improve the response and adaptation of the DC motor system. The hybrid system of PID controller and DL model is introduced. DL model was trained with the input signal of the DC motor, it was able to learn the characteristics of the DC motor. The DL model is developed in the DNN structure, the model. The function of the DL model is to compute a suitable combination of PID gain values to control the DC motor system in different conditions. The proposed model has obtained accuracy around 97% ± 1%, the deep learning can apply to the PID controller. The expectation of the method can be applied into different kinds of motors to control the PID gain values.