A novel hybrid predictive maintenance model based on clustering, smote and multi-layer perceptron neural network optimised with grey wolf algorithm

Abstract Considering the complexities and challenges in classification of multiclass imbalanced fault conditions, this study explores systematic combination unsupervised supervised learning by hybridising clustering (CLUST) optimised multi-layer perceptron neural network with grey wolf algorithm (GWO-MLP). The hybrid technique was meticulously examined on a historical hydraulic system dataset first, extracting selecting most significant statistical time-domain features. selected features were then grouped into distinct clusters allowing for reduced computational complexity through comparative four different frequently used categories algorithms classification. Synthetic Minority Over Sampling Technique (SMOTE) employed to balance classes training samples from various which served as inputs GWO-MLP. To validate proposed (CLUST-SMOTE-GWO-MLP), it compared its modifications (variants). superiority CLUST-SMOTE-GWO-MLP is demonstrated outperforming all terms test accuracy seven other performance evaluation metrics (error rate, sensitivity, specificity, precision, F score, Mathews Correlation Coefficient geometric mean). overall analysis indicates that efficient can be classify conditions. Article Highlights issue class outputs addressed improving predictive maintenance. A classifier based proposed. robustness feasibility validated complex dataset.