Transmission Line Fault Diagnosis Based On Variational Mode Decomposition And Convolutional Neural Networ

As the power system enters a new era of intelligence,transmission lines,as a core component of the power system,play an important role in long-distance power transmission.Short-circuit faults are extremely common and can cause significant damage to equipment and systems.Due to the special working environment of transmission lines,they are subject to many interference factors,such as weather,load changes,noise,and more.These factors can affect the transmission and reception of signals,making it difficult to extract the characteristics of fault signals completely.Therefore,it is necessary to combine modern scientific and technological methods to develop effective techniques to improve the accuracy of short-circuit fault detection.In this paper,a feature extraction method combining Variational Mode Decomposition(VMD)and Multiscale Permutation Entropy(MPE)is used,and a onedimensional convolutional neural network(1DCNN)is employed for the recognition and classification of faults in transmission lines.Considering the special environment of transmission lines and the challenges in data collection,a MATLAB/Simulink module is utilized to construct a single-machine power system with voltage up and down transmission,and various types of short-circuit faults are simulated by changing component parameters.The transient voltage,current,and zero-sequence components during fault occurrence are analyzed,and the waveform data of voltage and current during faults are collected,providing data support for subsequent fault feature extraction.To address the issue of poor fault feature extraction,this paper proposes a feature extraction method for transmission line faults based on VMD and MPE.Furthermore,to overcome the limitations of the VMD algorithm,the Firefly Optimization Algorithm is employed to improve the parameter search efficiency of VMD.Experimental comparisons with other feature extraction methods demonstrate that the proposed method effectively extracts fault features from short-circuit voltage signals and improves the accuracy of fault diagnosis.Finally,this paper explores the application of one-dimensional convolutional neural networks in the identification of fault types in transmission lines and establishes corresponding models.Based on the characteristics of convolutional neural networks,this paper uses directly collected fault signals as inputs and employs a one-dimensional convolutional neural network for classification.Additionally,feature extraction using VMD-MPE is performed,and experimental comparisons with other classifiers are conducted regarding the accuracy of fault diagnosis.The experimental results demonstrate that the one-dimensional convolutional neural network achieves a fault recognition rate of over 97% for different types of faults,confirming the accuracy and feasibility of the proposed method.