Artificial Intelligence for the Prediction of Bearing Failures in Industrial Machines Based on Data Entry: A Systematic Review

Abstract

Artificial intelligence in the field of predictive maintenance is becoming increasingly prominent, especially for identifying problems in industrial machinery bearings. The objective of this RSL is to show the most effective artificial intelligence-based methods for the analysis of various bearing failures. In the methodology, a manual search of original articles was carried out in Scopus, obtaining 295 in total, of which 55 studies met the established inclusion criteria. The application of artificial intelligence for problem identification reduces the costs associated with bearing failures, and greatly increases the effectiveness in fault detection. Finally, the methods with the highest efficiency included in this RSL were: the neural network method based on an initial block capsule network (ICN) and the refined composite multiscale dispersion entropy entropy method (RCMDE.) with an efficiency of 100 % respectively.