循环神经网络输电线路双端故障测距方法

Two-terminal fault location algorithm for transmission lines based on recurrent neural network

针对传统人工智能算法用于输电线路故障测距时泛化能力不足导致测距精度不高的问题,提出一种基于循环神经网络(recurrent neural network,RNN)的输电线路双端故障测距方法。首先通过仿真模拟不同故障模式下线路两端的电压、电流信号作为输入特征,对数据进行预处理后构造样本集,构建RNN模型用样本集进行训练和测试,使模型充分学习输入数据的深层特征,进而对故障距离进行较好的学习预测,最终获取高精度的故障测距模型。本文通过对IEEE10机39节点模型进行仿真试验及对某供电局管辖范围220 kV线路故障数据的测试,结果表明该模型较浅层模型具有更高的测距精度和更好的抗噪性能。

Aiming at the problem of low accuracy of fault location caused by insufficient generalization ability of traditional artificial intelligence algorithm in fault location of transmission lines.A method based on the recurrent neural network(RNN)is proposed for transmission line double terminal fault location.Firstly,the voltage and current signals at both ends of the line in different fault modes are simulated,and the sample set is constructed after preprocessing the data.The sample set will be used as the input to train and test the RNN model.The model can fully learn the deep features of the input data,and then perform better learning and prediction of fault location.Finally,a high precision fault location model is obtained.In this paper,the simulation tests of the standard model of 39 nodes system for IEEE 10 machines and the test results of 220 kV transmission line fault data within the jurisdiction of a power supply bureau show that the model has higher accuracy of fault location and better anti-noise performance than shallow model.