A rolling bearing fault diagnosis technique based on PLO-FMD and SSA-LSSVM-Adaboost

In order to further improve the accuracy and efficiency of fault diagnosis for rotating machinery such as rolling bearings, A fault diagnosis method for rolling bearings based on Polar Lights Optimization (abbreviated as PLO) for optimizing Feature Mode Decomposition (abbreviated as PLO-FMD), Sparrow Search Algorithm (abbreviated as SSA) and Adaptive Boosting (Adaboost) for Least Squares Support Vector Machine (abbreviated as LSSVM) is proposed. Firstly, in order to reduce the noise interference in the signal, PLO-FMD is used to filter the signal and the filter size and the number of mode components in the feature mode decomposition are optimized by the polar lights optimization algorithm. Afterwards, in order to improve the comprehensiveness and completeness of the feature parameters, the composite multi-scale attention entropy (abbreviated as CMATE) of the reconstructed signal is extracted and used as the feature vector. Again, the sparrow search algorithm with high convergence accuracy and fast speed is introduced to optimize the regularization parameters and kernel function parameters of the least squares support vector machine, and combine the adaptive boosting algorithm to construct the optimal SSA-LSSVM-Adaboost model. The rolling bearing dataset of Case Western Reserve University (CWRU) and Jiangnan University (JN) was introduced for case analysis, the results show that the accuracy rate of this method can reach 98.6111% and 97.9167%, respectively, and even after adding additional noise, it still maintains a high diagnostic accuracy. Compared with different features and learning models (MATE, MFE, SVM, etc.), its computing speed is very fast, it has strong noise resistance, and the diagnostic efficiency is also high. It has certain engineering application value.