In recent years, with increasing urbanization,  The HV cable lines has more than 30000 kilometers in SGCC. However, the long-term operation will cause insulation aging, fatigue to make the cable insulation performance gradually decline, equipment operation reliability gradually reduce. How to effectively analyze and manage large cable lines is a difficult problem to be solved urgently.
At present, monitoring of part of the high voltage cable lines has been realized, including the partial discharge, grounding current, temperature of cable line and  temperature, humidity of channels. In this paper, the research adopts the method of deep learning, puts forward an operation state prediction and comprehensive evaluation method Based on LSTM, to achieve Multi-state joint inference of single node for continuous time series, enhance the accuracy of state prediction, then using the analytic hierarchy process to predict the state weighted score, Finally, the quantitative score of high-voltage cable line operation status is obtained, which provides technical support for high-voltage cable maintenance decision.
The 100 groups of continuous monitoring status data has been used to verified the proposed model. The analysis results show that 90% of the 30 groups of fault line data are evaluated as abnormal or above conclusions, and 97% of the 70 groups of faultless line data are evaluated as normal conclusions, and all of them have no deviation conclusions. Therefore, it can be considered that the multi-operation state deduction and comprehensive evaluation method of high-voltage cable line based on LSTM neural network proposed in this paper can provide technical support for realizing big data analysis of high-voltage cable state quantity and improving cable state deduction and evaluation, and its application can effectively guarantee the operation reliability of high-voltage transmission cable.
 

peration State Prediction,Comprehensive Evaluation,HV Cable Lines