Genetic Based Optimization of Fuzzy Classifier for Diagnostics of Synchronous Motor Inter-Turn Faults

Paper describes creation of fuzzy classifier with structure and parameters optimized using genetic algorithm. Classifier is used for detection of synchronous motor inter-turn faults with gradually increasing level of damage. For each of the three investigated states - healthy and two grades of fault, a 200 samples were recorded and divided into two groups. The samples from first group were used for optimization of the classifier, and the samples included in second group were used to carry out the validation process. During the optimization process parameters such as: types and parameters of output functions, implication and aggregation methos, rules in rule base were determined using genetic algorithm. The optimization problem was defined as searching for such a classifier structure and set of its parameters, for which the classifier would generate the minimum classification error.