Fuzzy-clustering as a tool for magnetic losses analysis in induction machines

In the present work a new application of fuzzy clustering techniques is developed. The authors use this classification technique to study the magnetic losses in the induction machine stators, dividing this part of the motor in different clusters. There is a substantial difference between this fuzzy technique and the classical techniques of classification, while in the classical techniques a point can belong or not to a cluster in the fuzzy techniques a point can belong at the same time to different clusters with different membership degrees. These methods are strongly recommended for pattern recognition, classification and dimensionality reduction. In this case of study the methods are applied for classifying the points of the stator in the induction machines according to its magnetic losses. This classification allows us to separate the stator in different regions with different features. This region division is very helpful in the design step for many reasons. For example, the optimal shape of these regions in order to minimize the total amount of magnetic losses could be extracted.