Detection and diagnosis of bearing and cutting tool faults using hidden Markov models

摘要: Abstract Over the last few decades, research for new fault detection and diagnosis techniques in machining processes rotating machinery has attracted increasing interest worldwide. This development was mainly stimulated by rapid advance industrial technologies increase complexity of systems. In this study, discrete hidden Markov model (HMM) is applied to detect diagnose mechanical faults. The technique tested validated successfully using two scenarios: tool wear/fracture bearing first case correctly detected state (i.e., sharp, worn, or broken) whereas second application, classified severity seeded different engine bearings. success rate obtained our tests classification above 95%. addition severity, a location index developed determine location. been (inner race, ball, outer race) with an average 96%. training time required develop HMMs less than 5 s both monitoring cases.