HARMONY THEORY YIELDS ROBUST MACHINE FAULT-DIAGNOSTIC SYSTEMS BASED ON LEARNING VECTOR QUANTIZATION CLASSIFIERS

This contribution describes an algorithm to improve the ability of a l earning vector quantization (LVQ) classifier in machine fault diagnosi s. By adding a harmony model to the LVQ classifier the proposed method can construct an input-output mapping based on human knowledge and st ipulated input-output vector pairs. Knowledge atoms from harmony theor y are used to encode the knowledge of various machine fault patterns b y capturing the probability distributions of input features during the training process. Therefore, the class boundaries of various fault pa tterns are made more distinguishable, and the capability of classifica tion is enhanced. Moreover the summation of all the deviations generat ed from the input vectors and weights during the classification proces s can be better discriminated; therefore, the chance of misclassificat ion caused by a few dominant distorted features is reduced. This propo sed approach has been tested an classifying various faults obtained fr om a tapping machine, against other popular neural-network-based class ifiers. The results from a series of experiments have demonstrated tha t this hybrid approach is promising, and particularly useful in classi fying input features inherent with overlapping distributions and high uncertainty in the class boundaries. Copyright (C) 1996 Elsevier Scien ce Ltd