Research and Experimental Verification of Bearing Failure Methods in Elevator Traction Systems

In order to improve the diagnosis effect of elevator traction system bearing faults, the Kernel Extreme Learning Machine (KELM) is obtained by improving the Extreme Learning Machine, and the ISSA algorithm is obtained by improving the Sparrow Search Algorithm. The population initialization of LHS and the discoverer position update strategy of adaptive weights are introduced. The comparison shows that the ISSA algorithm can find the optimal solution of the function after 32 iterations, and the overall performance is good. Finally, the KELM is introduced into the ISSA algorithm to obtain the fault recognition model of ISSA-KELM. The experimental verification shows that the single bearing fault recognition rate is 95%, and the comprehensive bearing fault recognition rate is 93.4%, with high recognition rate and good diagnostic results.