Research On Fault Diagnosis Of Intelligent Mine Transmission Line Based On Multi Information Fusion

Due to the complex and changeable working conditions in the coal mine,the line fault of the power supply system can not be completely avoided.When the fault occurs,the fault must be handled accurately and quickly to ensure the smooth production of the mine and the personal safety of personnel.Correctly identifying the fault type is the key to the safe and rapid repair of the line.Aiming at the problem of low accuracy of existing fault diagnosis algorithms,an intelligent mine transmission line fault diagnosis method based on multi information fusion is proposed.Firstly,the obtained fault signal is decomposed by improved complete ensemble EMD with adaptive noise(ICEEMDAN),the correlation coefficient is introduced to screen the modal components,and the feature information of the screened modal components is extracted from three angles by using energy entropy,arrangement entropy and fuzzy entropy to form a multi information fusion feature data set;In order to reduce the dimension of the high-dimensional fault feature data set and reduce the sparsity and multicollinearity of the high-dimensional data,multiple dimensional scaling(MDS),principal component analysis(PCA),linear discriminant analysis(LDA)and locally linear embedding(LLE)are used to reduce the dimension respectively,the dimensionality reduction performance of the four algorithms is evaluated by introducing three performance indexes: feature retention objective function,category separation parameter and dimensionality reduction time.The results show that MDS has advantages in performance.Then the salp swarm algorithm(SSA)is introduced and improved.By introducing the population trust mechanism and mutation mechanism into the population location update mode,the convergence speed and convergence ability of the algorithm are enhanced.By using the test function to compare the optimization speed and optimization ability of the improved SSA,SSA,Particle Swarm Optimization(PSO),Grey Wolf Optimizer(GWO)and β-GWO,the superiority of the improved SSA is verified.Finally,the improved SSA is used to optimize the optimal dimensionality reduction parameters of MDS and the C、g parameters of support vector machine(SVM),and then the fault diagnosis model is constructed.The accuracy of fault type identification between this model and SVM fault diagnosis model optimized by common algorithms,BP neural network(BPNN),k-nearest neighbor(KNN)and random forest(RF)fault diagnosis model is analyzed and compared.The results show that the proposed fault diagnosis model has strong generalization ability and the accuracy of fault diagnosis is higher than other algorithm models.This thesis has 42 figures,11 tables and 81 references.