Optimized Artificial Neural Network-Based Control Strategy For Boost Converters

Abstract

Due to the rapid development of adaptive control and industrial automation science, the necessity of the boost converters is perceiving more and more, accordingly, designing a controller for these components which is dynamically characterized could be advantageous both theoretically and practically. Consequently, A machine learning-based controller using an artificial neural network (ANN) is designed, which is able to regulate the voltage of DC-DC Boost converters. In this paper, the primary controller is a linear quadratic regulator (LQR) based controller which is replaced by an ANN controller after generating training and testing data. After training the neural network, the Genetic Algorithm (GA) and an integral control action are used to minimize the system's overshoot and steady-state error, respectively. All in all, for the performance validation and making comparison accurately, the output voltage of a boost converter controlled by the optimized ANN model is simulated in the MATLAB/Simulink, which is conclusive that the new ANN controller can track the reference voltage properly.