• CSCD核心库收录期刊
  • 中文核心期刊
  • 中国科技核心期刊

ELECTRIC POWER CONSTRUCTION ›› 2023, Vol. 44 ›› Issue (3): 77-84.doi: 10.12204/j.issn.1000-7229.2023.03.008

• Research and Application of Key Technologies for Distribution Network Planning and Operation Optimization under New Energy Power Systems?Hosted by Professor WANG Shouxiang and Dr. ZHAO Qianyu? • Previous Articles     Next Articles

Distribution System External Fault Causes Identification based on Multi-Source Information Fusion

TAN Yuan(), ZHANG Wenhai(), WANG Yang()   

  1. College of Electrical Engineering, Sichuan University, Chengdu 610065, China
  • Received:2022-04-18 Online:2023-03-01 Published:2023-03-02
  • Supported by:
    National Key R&D Program of China(2020YFF0305800);Sichuan Natural Science Foundation(2022NSFSC0234)


Accurate identification of distribution network fault causes is of great significance to shorten the time of fault search and improve the speed of power supply recovery and power supply reliability. According to the attribution of responsibility, the causes of distribution network fault can be divided into internal causes and external causes. Internal causes refer to electrical related causes such as equipment insulation weakening and overvoltage, while external faults usually refer to weather, animal or human activities. Faults caused by external causes are the result of a combination of many factors. In this paper, a fault external cause identification method is proposed, which integrates the non-electrical information such as line parameters, weather and time, and the electrical information such as fault current and phase number. Firstly, the characteristics and related influencing factors of five typical external fault causes are analyzed to build the basis of identification model. Then unsupervised learning training is used to obtain the optimal parameters of each layer of deep belief network, and supervised learning is used to fine-tune the global parameters, and the classification model of external causes of distribution network faults based on deep belief network is obtained. Finally, the accuracy of the algorithm is verified by using the actual fault data of a certain area in western China. The result shows that the recognition accuracy can reach 94.82%, which proves the correctness of the method.

Key words: distribution system fault, external fault causes, fault cause identification, deep learning, multi-source information fusion

CLC Number: