Abstract:We propose an intelligent fault identification method for the VP-type tiltmeter. Using empirical mode decomposition(EMD), we decompose the normalized fault signal into six intrinsic mode functions(IMF), and calculate the approximate entropy respectively to construct the EMD multiscale approximate entropy input matrix. Combined with the grasshopper optimization algorithm(GOA), we optimize the parameters of the self-organizing feature map(SOM) neural network. Then, we embed the obtained GOA optimal value, and form the GOA-SOM identification model. We use the identification test set to obtain the cluster label value of the target, and compare it with the cluster label of the training set and the real fault type, to obtain the fault identification result. The experiments show that the GOA-SOM model has a mean and standard deviation of the identification accuracy under 100 random sampling conditions of 99.329 7% and 1.218 8. These are better than traditional models.
PANG Cong,MA Wugang,LI Chawei et al. An Intelligent Fault Identification Method for VP Tiltmeter Using GOA-Optimized SOM Neural Network[J]. jgg, 2023, 43(3): 322-326.