Control of a DC–DC buck converter using adaptive neural network

Due to the DC–DC buck converter’s complexity and nonlinear characteristics with many time-varying uncertain parameters, and sensitivity to load current disturbances. Therefore, designing a robust control to achieve stable and satisfactory performance across multiple operating points is a challenging problem. This paper introduces an adaptive neural network sliding mode control (ANN-SMC) combined with a proportional-integral (PI) control for a DC–DC buck converter powering a resistive load to address this challenge. In this approach, the voltage control loop is managed by a PI control, while the current control loop is controlled by an adaptive neural sliding mode control. The radial basis function neural network (RBFN) is employed to quickly identify the gradient of the sliding surface to reduce chattering phenomena, which is trained online to ensure the system’s stability and adaptability despite parameter uncertainties and external disturbances. The proposed control algorithm’s stability is verified using Lyapunov stability theory. To validate the effectiveness of the proposed method, the DC–DC buck converter is simulated using MATLAB/Simulink under varying parameters and disturbances. Moreover, experimental validation is carried out on the DSP 320F28379 development board. Results from both simulations and experiments show that the proposed control has a faster response, reduces settling time, and eliminates overshoot compared to the traditional PI control, even when parameters such as the input voltage and the load change. In particular, the proposed control maintains robustness even when a large amplitude disturbance enters the input voltage.