Recognising patterns in electromechanical systems

Pattern recognition using fuzzy logic and neural network techniques is usually related with image and speech processing. In a different approach, this paper shows results concerning the use of these techniques to extract models of electromechanical system's. This approach is important because such systems always contain internal non-linear relations that are difficult to model and, in most cases, there are unknown internal phenomena dominating their dynamics. So, these techniques enable to recognise system patterns and to include them in a more complete system's model. (C) 1997 Elsevier Science B.V.