融合故障波形及故障诱因信息的输电线路故障原因智能识别方法OA

Current research on transmission line fault identification faces several challenges,including the scarcity of real-world fault samples,insufficient extraction of fault cause information,and a lack of effective methods for integrating multi-source fault data.These issues limit the accuracy of practical fault identification.To address these problems,this paper proposes an intelligent comprehensive fault recognition framework that integrates both fault waveform data and fault cause information.First,based on existing fault mechanisms and real fault waveforms,a simulation scheme is developed to generate diverse waveform data for six common types of transmission line faults:lightning strikes,foreign object intrusion,wind deviation,icing,pollution flashover,and wildfires.Then,a fault waveform feature extraction algorithm based on improved polynomial features is proposed,along with enhancements to residual neural networks and Bayesian models for recognizing both waveform features and fault causes.Finally,a KL-divergence-based fault cause information fusion algorithm is introduced to combine the outputs from the two recognition models,forming a comprehensive fault identification solution.The proposed framework is validated using 200 sets of real fault data,demonstrating its optimal performance and the effectiveness of incorporating fault cause information.