A Power System Harmonic Problem Based on the BP Neural Network Learning Algorithm

Abstract

At present, due to the large-scale use of different kinds of power electronic devices in the power system, the problem of harmonic pollution in the power grid is becoming more and more serious, which will lead to a serious decline in the production, transmission, and utilization rate of electric energy, overheat electrical devices, generate vibration and interference, and then affect the aging and service life of the lines. In order to effectively reduce the harmonic problems caused by different levels of the power system, it is necessary to analyze the harmonic components. In this paper, the BP neural network learning algorithm is introduced into the harmonic problems of the power system. The mapping relationship between input and output signals is obtained by using the BP neural network algorithm, and the harmonic frequency, amplitude, and phase contained in the obtained data are analyzed. According to the type of equipment with problems in the operation of the power system and the rapid diagnosis of existing defects, the problems are quickly located and the causes are analyzed. The practical results show that the BP neural network learning algorithm proposed in this paper has higher detection accuracy and analysis speed for the difficult problems in the power system.

Research on Subway Harmonic Compensation Control Method Based on SSA Optimized Adaline Algorithm

“…Compared with traditional methods such as Fast Fourier Transform (FFT) [1], Wavelet Transform [2], and Prony algorithm [3], combining intelligent algorithms with harmonic detection can achieve more accurate results. Studies by Li et al [4] and Zhao et al [5] apply an adaptive interference cancellation technique to harmonic detection, which can accurately distinguish the harmonic content, but the anti-interference ability is poor, and the dynamic response is slow; Zhu et al's work [6] utilizes the advantages of full-phase FFT phase invariance, combined with the artificial neural network, which can detect the harmonic number and phase with high accuracy, yet the detection of harmonic amplitude and frequency is not accurate; Wang et al [7] and Yue et al [8] use BP neural network for harmonic analysis, which directly outputs the indexes required by the user with less calculation, although BP is easy to fall into the local optimal solution; Liu and Fei [9] use RBF in their research, a method that can detect all the harmonic components in only half a cycle with high accuracy. However, when the amount of data is large, the detection results are not accurate.…”