Data-Driven Fault Detection of Power Distribution Network Based on Temporal Convolutional Network and Long Short-Term Memory
Ming Lu, Guang Feng, Suhui Huang, Shanfeng Liu, Ling Xiang, Hao Su · 2023
The stable operation of the power distribution network plays an important role in ensuring the reliability and continuity of power transmission. However, the distribution network often experiences faults such as equipment failures and line short circuits during its operation. These faults not only affect the normal operation of the distribution network but also result in significant social and economic losses. Therefore, a new power distribution network fault detection method which is temporal convolutional network (TCN) cascaded with long short-term memory (LSTM) parallel network (TLPN) is proposed. The feeder current is adopted as input and output in this model. The proposed model highlights the feature extraction capabilities, and the temporal characteristics adaptive data are strengthened in established network model. Statistical analysis is performed on the residuals of the output, and an adaptive threshold is set to detect the trend changes of the feeder current. Through a case study of a certain substation, the proposed method can detect distribution network faults in advance, which can ensure the safety and reliability of power distribution networks.