An Improved Wavelet Neural Network Harmonic Detection Method
Weiwei Jing,Mingming Shi,Xian Zheng,Xinchen Wang,Xuan Jin,Shaorong Wang
DOI: https://doi.org/10.1109/ifeea60725.2023.10429536
2023-01-01
Abstract:With the escalating adoption of electronic devices, harmonic pollution in power systems has emerged as a significant concern, posing a threat to the secure and stable operation of equipment. The traditional harmonic detection methods are mainly based on Fourier Transform, but this approach often leads to phenomena like spectral leakage and the picket-fence effect in practical applications. This paper combines wavelet transform with neural network algorithms for harmonic detection. It employs a training algorithm with an added momentum term and an adaptive learning rate adjustment algorithm to enhance the accuracy and learning efficiency of the wavelet neural network. This approach improves network performance, demonstrating higher applicability. Finally, this paper conducts simulation comparisons between the proposed improvement method and traditional approaches, demonstrating its faster convergence speed and higher detection accuracy. These findings highlight the method’s engineering value.