Research on Intelligent Mining Algorithm for Distribution Network Transformer Fault Early Warning

• Published in: IOP conference series. Earth and environmental science 2021-03, Vol.701 (1), p.12014
• Main Authors: Luo, J P, Yuan, L L, Han, L Q, Huang, W T
• Format: Article
• Language: English
• Subjects: Algorithms; Correlation analysis; Data analysis; Data collection; Data mining; Electric power grids; Failure analysis; Fault detection; Internet of Things; Mathematical analysis; Network reliability; Power supply; Transformers; Vibration
• ISSN: 1755-1307; 1755-1315
• DOI: 10.1088/1755-1315/701/1/012014

The distribution network is the end link of the power grid. The level of management directly affects the power supply capacity and power quality. It is related to the operation level and social image of the power grid enterprise. The transformer is a very important component in the distribution network. The transformer is continuously safe and stable. Grid power supply reliability. In this context, this paper proposes an intelligent gateway intelligent fault detection algorithm for transformer operation failure, which collects data of transformers in multiple dimensions and multiple source channels, such as electrical variables and non-electrical variables. Correlation analysis and causality analysis of the data to achieve an all-round warning of transformer operation failure. The traditional method of transformer fault warning is solved by relying on the special gas concentration or transformer vibration signal generated by the fault of the transformer fault to solve the problem of fault identification and early warning singleness and low reliability, and fully exert the technical advantages of the Internet of Things and data mining. The evaluation value of the five “ancestor“ elements is 0.2, 0.25, 0.15, 0.3, 0.1. Transformer operation fault trend early warning coefficient is set to 0.80. Once the early warning coefficient is found to exceed 0.80 during data collection and mining analysis, a transformer failure early warning will be issued, which can facilitate grid operators’ intervention and control in advance. The actual calculation case also demonstrates the correctness and effectiveness of the intelligent mining algorithm. This paper provides an effective support for the development and application of big data technology in the distribution network.