University of Isfahan Computational Intelligence in Electrical Engineering 2821-0689 13 1 2022 03 21 Decentralized Adaptive Voltage Control and Equal Current Sharing of Parallel- Connected Buck Converters via Wavelet Neural Network Approximators Decentralized Adaptive Voltage Control and Equal Current Sharing of Parallel- Connected Buck Converters via Wavelet Neural Network Approximators 84 102 25666 10.22108/isee.2020.118428.1262 FA Sajjad Shoja-Majidabad Assistant Professor, Department of Electrical Engineering, University of Bonab, Bonab, Iran Fatemeh Farjami MSc Student, Department of Electrical Engineering, University of Semnan, Semnan, Iran Journal Article 2019 07 27 A parallel connection of Buck converters improves system reliability and efficiency. However, the open circuit fault, load, and supply voltage uncertainties, and interactions among the converters increase the complexity of output voltage control and balanced current sharing. Thus, in this paper, first, a decentralized backstepping sliding mode control strategy is designed to meet such challenges. However, this controller is quite conservative since the uncertainties and interaction bounds are not known. Moreover, the sliding mode based controllers suffer from chattering phenomena which limits the practical applications. Therefore, a decentralized adaptive backstepping control strategy with wavelet neural network approximators is proposed. This strategy reduces the chattering and approximates the uncertainties and interactions by replacing the switching terms with wavelet neural networks. To show the effectiveness of the proposed controller, different numerical simulations have been performed in the presence of reference voltage changes, load, supply voltage variations, and open circuit faults. A parallel connection of Buck converters improves system reliability and efficiency. However, the open circuit fault, load, and supply voltage uncertainties, and interactions among the converters increase the complexity of output voltage control and balanced current sharing. Thus, in this paper, first, a decentralized backstepping sliding mode control strategy is designed to meet such challenges. However, this controller is quite conservative since the uncertainties and interaction bounds are not known. Moreover, the sliding mode based controllers suffer from chattering phenomena which limits the practical applications. Therefore, a decentralized adaptive backstepping control strategy with wavelet neural network approximators is proposed. This strategy reduces the chattering and approximates the uncertainties and interactions by replacing the switching terms with wavelet neural networks. To show the effectiveness of the proposed controller, different numerical simulations have been performed in the presence of reference voltage changes, load, supply voltage variations, and open circuit faults. https://isee.ui.ac.ir/article_25666_489c8640c7747df77add47fc196d880f.pdf