面向文本非结构化数据的输变电系统故障诊断方法

摘要

Effective diagnosis faults in transmission system is important for the precise determination of the equipment operating status. The fault diagnosis method of transmission system based on deep learning network was proposed. Aiming at the problem of unstructured data processing, a series of deep learning methods were presented in this paper, such as unstructured data feature extraction, deep neural network construction, deep neural network training and fault diagnosis. The recurrent neural network (RNN) was improved, and the long-short term memory model (LSTM) was added to the memory unit in the neural network. The corresponding neural network training strategy was thus proposed. The simulation results verified the effectiveness of the proposed method.%从非结构化数据中提取信息,有效诊断输电系统故障对于精确确定设备的运行状态具有重要意义.提出基于深度学习网络的输变电系统故障诊断方法;面向非结构化数据处理问题,提出非结构化数据特征提取、深度神经网络构建、深度神经网络训练、故障诊断等一系列深度学习处理方式;构建并改进循环神经网络RNN,将长短时记忆模型(LSTM)添加到神经网络中的记忆单元,并提出相应的神经网络训练策略.以南方电网故障巡检报告作为数据源,仿真分析结果验证了该方法的有效性.