Voltage Regulation of Buck Converter Based on the Improved Radial Basis Function Neural Network PID Control Algorithm

To maximize the output efficiency of Switch Mode Power Supply, the stability and robustness of the output voltage are of crucial importance in practical work. Aiming at the problems of low steady-state accuracy and slow response speed of voltage output in the switching power supply system under traditional Proportional Integral Derivative (PID) control, this paper designs an algorithm for tuning PID parameters based on the Radial Basis Function (RBF) neural network, improves the RBF neural network, combines the Adaptive Gradient Algorithm (Adagrad) with the L2 regularization strategy, improves the output weights of the neural network, and conducts simulation through MATLAB software to compare the traditional PID control with the improved RBF-PID control. Simulation results indicate that the improved RBF-PID controller is highly adaptable and refines the buck converter’s output voltage waveform. The voltage overshoot is limited to 0.5%, representing a 99% reduction compared to the traditional PID controller. The voltage regulation time is 0.12 s, representing a 33% reduction compared to the traditional PID controller. All performance metrics of the improved RBF-PID algorithm exceed those of the traditional PID algorithm. It has better robustness and adaptability, and can ensure that the switching power supply system has good static and dynamic characterristics.