Condition monitoring and fault prediction via an adaptive neural network

This paper describes the application of an adaptive neural network, called fuzzy ARTMAP (FAM), to handle fault prediction and condition monitoring problems in a power generation station. The FAM network, which is supplemented with a pruning algorithm, is used as a classifier to predict different machine conditions, in an offline learning mode. The process under scrutiny in the power plant is the circulating water (CW) system, with prime attention to monitoring the heat transfer efficiency of the condensers. Several phases of experiments were conducted to investigate the "optimum" setting of a set of parameters of the FAM classifier for monitoring heat transfer conditions in the power plant.