Intelligent Modeling And Nonlinear H_∞ Control For Switched Reluctance Motor

The construction of model for switched reluctance motor (SRM) greatly affectsthe motor design, the analysis of dynamic character and the evaluation of SRM.Meanwhile, the exact motor model also benefits the high level control of SRM. Thispaper applies soft computing techniques to the nonlinear modeling of switchedreluctance motor (SRM). From the measured data of stator current, windinginductance and rotor angle position in experiment, the nonlinear characteristic ofSRM is obtained and fuzzy rules are further developed. Continually, because of thedisadvantage of fuzzy modeling, neuro-networks with the self_learning ability iscombined with fuzzy method to reduce the modeling error. Finally, due to theradomness of the choice of fuzzy membership function the genetic algorithm (GA) isemployed in the paper to optimize nonlinear modeling. Soft computing techniquebased nonlinear modeling could avoid the complexity of analytical modeling andgreatly simplified the procedure of modeling.This paper proposed the resolution of an intelligent H ∞ control problem for aclass of nonlinear systems in which unknown nonlinear functions are assumed to exist.The proposed scheme is applied to the position control of linear switched reluctancemotor drives system. In the procedure, RBF Neural Networks (NNs) are used tomodel the nonlinear functions and the approximation error is looked on as part of theexternal disturbance. Further, H ∞tracking controller is derived based on Lyapunovfunction and the notion of dissipativeness. Not only can the controller guarantee thestability of the overall control system, but it can also attenuate the effect of both theexternal disturbance and NNs approximation error to a prescribed level.