Power Grid Fault Diagnosis Method Based on LightGBM

Abstract

With access to distributed power generation and interconnection between regions, China’s power system structure is becoming increasingly complex, and the requirements for power grid dispatching capacity are getting higher and higher. When the power grid fails, a large amount of alarm information appears in the monitoring window, which needs to be processed and classified first to speed up the fault diagnosis and processing. Therefore, this paper proposes a power grid fault diagnosis method based on LightGBM. First, after pre-processing the power grid alarm information, Word2vec is used to vectorize the text, and then the LightGBM model is used to process the text, extract the information features and continue to train. Finally, realizing the three classifications of such simple faults as the transformer fault, bus fault, and line fault. Finally, through the test and verification of the actual power grid case data, it is verified that the method can effectively classify simple faults and meet the needs of the actual power grid online diagnosis.