2022 , Volume 27, № 1, p.70-87

Condition-based maintenance (CBM) utilizes the real conditions of the components to make a decision when to replace and/or maintain the components, thus maximizing the life span of the machineries, whilst minimizing the count of service interferences. Particularly, CBM system identifies the noteworthy variations and fluctuations of signals and variables on the basis of sensor information so as to avoid or prevent the breakdown in machines. Thereby, this work develops a new prediction model on CBM in class II machines, where vibration velocity and average time are considered as input parameters and accordingly, the availability and reliability are predicted. Here, the proposed scheme consists of 2 chief phases like (1) feature extraction and (2) prediction. At first, feature extraction is performed, wherein statistical and higher-order statistical features are extracted. Subsequent to this, the extracted features are given to the Neural Network (NN) classifier that predicts the final output (availability and reliability of machines). To enhance the prediction accuracy of the classifier, the weights of NN are fine-tuned via Levy Flight Adopted Grey Wolf Optimization (LF-GWO). At last, the supremacy the presented approach is proved with respect to varied measures.