Motor bearing fault diagnosis by combining vibration feature optimization and GWOA-XGBoost

In order to solve the problem that the fault state of motor bearings is difficult to identify, which results in low diagnosis accuracy, a motor bearing fault diagnosis model was proposed by combining the signal feature extraction and XGBoost (extreme gradient boosting algorithm). The optimized variational mode decomposition was used to obtain the IMF (intrinsic mode function) components of the vibration signal, and the multi scale entropy value of each IMF component was calculated on the basis of the multi-scale entropy theory for feature reconstruction. The WOA (whale optimization algorithm) was improved by introducing the selection, crossover and mutation operation of genetic algorithm. The improved WOA algorithm was used to optimize the hyperparameters of XGBoost, and obtain the hyperparameter combination that helped the XGBoost achieve the best classification effect. The vibration signals of 7 different types of faults were reconstructed and input into the optimized XGBoost model for fault diagnosis. Experimental results show that the motor bearing fault diagnosis accuracy of the proposed GWOA-XGBoost model can reach 97.14%, which significantly improves the performance compared with the traditional diagnosis method.