I. INTRODUCTION
The work presented here examines the feasibility of employing artificial–neural–network (ANN) models that are constructed by using genetic algorithms (GAs) to model the performance characteristics of switched–reluctance–motor (SRM) drive systems during normal and abnormal operating conditions. The modeling of these conditions requires the use of coupled magnetic and state space/lumped parameter circuit models of the machine and the associated converter [1]. However, a drawback with this approach is that one needs to fully repeat the analysis for any changes in the system topology, loading, or fault conditions to characterize the motor drive system. On the other hand, ANN–built models, due to their interpolation property can avail that.