【摘要】 配电网结构复杂、中性点接地方式多样，存在故障选线方法可靠性不高的问题。针对该问题提出了一种基于构造特征量的长短期记忆（long short-term memory,LSTM）网络单相接地故障选线方法。首先，通过搭建单相接地故障仿真模型，分析故障线路与健全线路接地前后电流、有功功率以及无功功率的变化特点，并基于此构造新特征；然后，结合多种特征选择算法选取特征量；最后，建立LSTM故障选线模型并采用实际故障数据进行算例验证，结果表明该方法具有故障选线准确度高的特点。更多还原

【Abstract】 The distribution network has a complex structure, diverse neutral grounding methods, and the problem that the fault line selection method is not reliable exists. This paper proposes a long short-term memory(LSTM)network single-phase grounding fault line selection method based on structural feature quantities.Firstly, a single-phase grounding fault simulation model is built to analyze the variation characteristics of the faulted line and the sound line before and after grounding, based on which new features are constructed.Secondly, multiple feature selection algorithms are combined to select feature quantities. Finally, LSTM fault line selection model is established and verified by actual fault data examples, and the results show that the method has the characteristics of high fault line selection accuracy.更多还原