Performance assessment of two classification schemes for cutting tool degradation monitoring

The ability to actively predict the failure and degradation of critical machine components is a pressing concern in modern manufacturing. The potential cost savings stimulated the introduction of numerous artificial intelligent techniques into the manufacturing arena. The need to explore and understand the capacities of these new methods is extremely important to define their suitable application. This paper offers a global comparison of hidden Markov model (HMM) and rough sets theory (RST) based classifiers, using traditional statistical measures as well as different key criteria to the manufacturing community. The results showed very close performances over several criteria and explored the possible combination of the two methods in a hybrid model.