Modeling faulted switched reluctance motors using evolutionary neural networks

This paper examines the feasibility of using artificial neural networks (ANNs) and evolutionary algorithms (EAs) to develop discrete time dynamic models for fault-free and faulted switched reluctance motor (SRM) drive systems. SRMs are capable of functioning despite the presence of faults. Faults impart transient changes to machine inductances that are difficult to model analytically. After this transient period, SRMs are capable of functioning at a reduced level of performance. ANNs are applied for their well-known interpolation capabilities for highly nonlinear systems. A dynamical model for an SRM is constructed by feeding values for state variables back to ANN inputs. EAs are employed for their ability to search complex structural and parameter spaces to find good ANN solutions. Furthermore, the ANN structure and training regimen parameters are searched for using EAs. Finally, an analysis of the search is performed, and the resulting model is presented. The results of using the ANN-EA-based model to predict the performance characteristics of prototype SRM drive motion under normal and abnormal operating conditions are presented and verified by comparison to test data.