Skip to Main Content
A new fault diagnosis method is proposed, which combines negative selection algorithm and conventional classification algorithm. All the available training samples, including normal samples and known anomaly, are treated as positive samples (self). The real-valued negative selection algorithm is adopted to generate the negative samples (non-self), which distribute among the rest of the detected space. Both the positive and negative samples are used to train a feed forward artificial neural network classifier, whose output neurons are assigned to fault indicators. The simulation result demonstrates that the performance of the proposed approach is satisfactory.