Fault classification using genetic programming

Genetic Programming (GP) is a stochastic process for automatically generating computer programs, which has seen relatively little use in Condition Monitoring (CM). In this paper, GP is developed to classify different normal/faulty conditions in rotating machinery for the purpose of both dual-class and multi-class classification. In our investigations, all classes, after initial training, can be recognised with 100% accuracy. Besides this good performance in classification, GP has many other advantages, such as fast training, easily understood results and inherent feature selection. These properties show that GP could be a very beneficial tool in condition monitoring.