Fault Diagnosis Of High-voltage Pulse Track Circuit Based On IHHO-KELM

Aiming at the problem of low fault diagnosis accuracy of high-voltage pulse track circuit, a track circuit fault diagnosis method based on kernel extreme learning machine (KELM) optimized by improved Harris hawks optimization (IHHO) is proposed. Firstly, in order to improve the optimization performance, the logarithmic convergence factor is used in combination with the symbiotic organisms search (SOS) algorithm to improve the basic Harris hawks optimization (HHO) algorithm. The benchmark test function is used for the experiment, which proves that the IHHO algorithm performs better in convergence accuracy and convergence speed. Secondly, the IHHO algorithm is used to optimize the kernel function parameter and penalty coefficient of the KELM model and then improves the fault diagnosis accuracy of the KELM model. Finally, the IHHO-KELM model is used to diagnose fault types of the high-voltage pulse track circuit. The experimental results show that the diagnostic accuracy of the proposed IHHO-KELM model is 93.3%, which is 6.6%, 5%, and 3.3% higher than that of the KELM model, GA (Genetic algorithm)-KELM model, and HHO-KELM model respectively. Further experiments verify that the IHHO-KELM model is superior to BP neural network, deep confidence network (DBN), long short-term memory (LSTM) network, and gated recurrent unit (GRU) network in terms of diagnostic accuracy, training time, and diagnostic mean square error, which proves the rapidity and stability of the IHHO-KELM model in track circuit fault diagnosis.