To determine the defect level of distribution network equipment,a method for automatic defect analysis of distribution network equipment based on unstructured data and CNN-BiLSTM is proposed.Firstly,propose the overall idea of automatic classifi-cation of equipment defects in this distribution network;Then,unstructured data is processed using word2Vec and a CNN-BiLSTM defect level analysis model based on attention mechanism is constructed for data analysis;Finally,the feasibility of the above method was verified through simulation.The results showed that compared with a single CNN and CNN BiLSTM model,the proposed method in this study increased the weight of important information in the semantic features of power equipment defect description text by utili-zing attention mechanism,thereby improving the performance indicators of the analysis model.Among them,the accuracy index of Acc was improved by 4％～5％,the MF1 index was improved by 5％～6％,and the WF1 index was improved by 3％～4％.From this,it can be concluded that this defect analysis model can accurately analyze the defect level of power defect description texts,and has certain feasibility and effectiveness.
关键词
配电网/电力设备缺陷/CNN-BiLSTM/注意力机制
Key words
unstructured data/defects in power equipment/CNN-BiLSTM/attention mechanism
维普
万方数据