Adaptive Neuro-Wavelet System for the Robust Control of Switching Power Supplies

In this study, a new method for designing an adaptive controller based on Wavelet Neural Networks, is represented. The proposed controlling method is based on a Neuro-Wavelet controller and a robust controller. The Neuro-Wavelet controller is designed to emulate an ideal controller and a robust controller is designed to recover the residual approximation for ensuring the stable control performance. The adaptive law is derived on the basis of Lyapunov stability theorem, so, the stability of the under controlled system is guaranteed, when no exact condition or no prior knowledge is available. Moreover, to relax the requirement for a known bound on aggregated uncertainty, which comprises a minimum approximation error, optimal network parameters and higher order terms in a Taylor series expansion of the wavelet functions, a system with adaptive bound estimation was investigated for the control of a forward switch mode power supply. In addition, numerical simulation results show that the dynamic behaviors of the proposed systems, due to periodic commands, are robust with regard to parameter variations and external load resistance disturbance.