Research on Power Grid Fault Diagnosis Method Based on PMU Data and Convolutional Neural Network

• Published in: 2020 IEEE 4th Conference on Energy Internet and Energy System Integration (EI2) pp. 2987 - 2993
• Main Authors: Ding, Xuanwen, Zhu, Xin, Li, Xiaopeng, Zhang, Chun
• Format: Conference Proceeding
• Language: English
• Published: IEEE 30.10.2020
• Subjects: Convolutional neural networks; Current measurement; deep learning; Fault diagnosis; Feature extraction; one-dimensional convolutional neural network; Phasor measurement units; PMU measurement data; Power grids; Power system stability
• DOI: 10.1109/EI250167.2020.9347132

With the rapid development of the power system, it becomes more and more complex, and the safety and stability performance is also facing severe tests. Therefore, it is urgent to explore a fast and accurate method for power grid fault diagnosis. The wide application of synchronous phasor measurement technology has laid the foundation for accurate fault diagnosis in the power grid. Therefore, based on the measurement data of the synchronous phasor measurement unit (PMU) simulated by digsilent simulation software, a sample set with a time series of 8 seconds and characteristic parameters of three-phase current and zero sequence current amplitude was constructed. Proposed a power grid fault diagnosis method using one-dimensional convolutional neural network (1D-CNN), which realized the end-to-end "timing feature extraction + fault type classification". The verification of simulation and measured data shows that compared with the traditional power grid fault diagnosis method, the proposed method can make more accurate fault type diagnosis with shorter response time.