Untitled

A review on lubricant condition monitoring information analysis for maintenance decision support

“…DL has been proved to yield more accurate results than human beings but limited by factors such as overfitting, big data requirement and difficult to work with the reasoning for instance application of scientific methods and programming. It was used in texture entropy detection of transformer oil [147], wear particle analysis and classification by various authors [148][149][150][151][152]. Deep learning is fast, usually more accurate and capable of building the features without supervision from either unlabeled or unstructured data.…”

“…ES utilize domain expert knowledge employing to emulate decision-making ability of a human, perform and exhibit reasoning for problem-solving, hence advancing maintenance support. ES has been used for LCM data interpretation as developed by [149,183] , to automate intelligently the classification process of wear particle [160] , estimate the quality of oil [184], to evolve EXCARE system for maintenance of lubricants in service [77] and in a lubricant monitoring and diagnostics case study using computer-aided wear particle analysis software [185]. In a study in conjunction with a different condition monitoring technique, ES was applied using data from wear debris from oil and vibrational analysis by [186].…”