Busbar fault diagnosis method based on multi-source information fusion

Against the backdrop of smart grid development, the electric power system demands higher accuracy and comprehensiveness in fault analysis. Establishing a digital twin platform for multiple equipment faults represents the future direction of power system development. Presently, while many researchers employ artificial intelligence algorithms to diagnose faults in key equipment such as transmission lines and transformers, intelligent diagnostic methods for busbar faults remain insufficient. Therefore, this paper proposes a busbar fault diagnosis method based on multi-source information fusion. Initially, the diagnostic method for busbar faults is explored, conducting both time-domain and frequency-domain analyses on simulated fault data. The data of this model are optimized using Dempster-Shafer evidence theory to enhance algorithm training speed. Subsequently, BP neural network training is implemented. Finally, validation testing of fault data demonstrates a fault recognition accuracy of 99.1% for this method. Experimental results illustrate the method's feasibility and low computational costs, thereby advancing the development of digital twin platforms for power system fault diagnosis.