基于小波包能量谱算法和RBF神经网络的铁路牵引网故障识别研究  

The railway traction network fault identification based on wavelet packet energy spectrum algorithm and RBF neural network

作　　者：李伟 Li Wei(China Railway International Co.,Ltd.,Beijing,100038,China)

机构地区：[1]中国铁路国际有限公司,北京100038

出　　处：《机械设计与制造工程》2024年第12期87-91,共5页Machine Design and Manufacturing Engineering

摘　　要：针对铁路牵引网故障类型识别效率低下的问题,构建了基于小波包能量谱算法和径向基函数神经网络的电气化铁路牵引网线路仿真模型。该模型通过生成故障瞬态信号,利用小波包能量谱算法提取故障特征,并通过径向基函数神经网络进行训练和识别。结果显示,在故障瞬态电流测试中,雷击故障的瞬态电流达到了5400 A。在仿真测试中,所提模型对不同类型故障进行识别的平均准确率均稳定在88%以上,同时在迭代次数为59时误差值仅为0.0001。通过该模型能够有效识别多种类型故障,可为电气化铁路的安全运行与持续发展提供可靠的技术支持。Aiming at the low efficiency in identifying fault types in railway traction networks,a simulation model of electrified railway traction network line based on wavelet packet energy spectrum algorithm and radial basis function neural network is constructed.The model generates fault transient signals through the traction network simulation model,extracts fault features using wavelet packet energy spectrum algorithm,and trains and identifies them through radial basis function neural network.The results show that in the fault transient current test,the transient current of lightning fault reaches 5400 A.In the simulation test,the average accuracy of the proposed model in identifying different types of faults is stable at more than 88%,and the error value is only 0.0001 when the number of iterations is 59.The model can effectively identify various types of faults and provide reliable technical support for the safe operation and sustainable development of electrified railways.

关 键 词：牵引网 故障识别 RBF神经网络 仿真模型 小波包能量谱算法 

分 类 号：U22[交通运输工程—道路与铁道工程]