Research On Nonlinear Model Of 6/3 SRM With Flux-reversal-free Stator

Switched reluctance motor(SRM)has the advantage of simple structure,low cost,high mechanical strength and suitable for high-speed operation.SRM has a huge application market in the field of electric transmission.My research group puts forward the structure of 6/3 SRM to solve the traditional two-phase SRM problems of large torque ripple,non-self-starting,flux-inversal.Due to the double salient pole structure of switched reluctance motor,the magnetic circuit of the motor is seriously nonlinear.In order to accurately calculate and simulate the motor characteristics and improve the speed regulation performance of the motor drive system,this paper in view of the researches on nonlinear model of 6/3 SRM.The main research work of this paper is as follows:In order to improve the accuracy of magnetic linkage data fitting,the exponential term of Gauss function is improved,and Gaussian of decreasing power and convergent is proposed to fit the magnetic linkage data of different rotor positions.This new function eliminates the problem that the traditional Gauss function model could not be integrated when calculating electromagnetic torque.In addition,the arctangent function is used to fit the flux data to verify the accuracy of the fitting method of the Gaussian of decreasing power and convergent,and the coefficients of different fitting function expressions were solved through the flux data.The fitting effects of two functions are compared.In view of the inductance characteristic curve of 6/3 SRM has no symmetry in a cycle because of the rotor structure of non-uniform air-gap,this paper has studied Fourier series of divided domain orthogonal nonlinear modeling method.The rising and falling ranges of inductance are modeled respectively.With the help of the sinusoidal component of Fourier series,the complete nonlinear modeling of flux and torque of the motor were finished.In addition,aiming at the errors of the nonlinear model of torque in the zero region,a binary function of rotor position and phase winding current is constructed,and the error surface between the nonlinear torque model and the simulation data is fitted to correct the torque model.In view of complexed modeling process of Fourier series of divided domain orthogonal method,the least squares support vector machine(LSSVM)global nonlinear modeling method is studied.The parameters of the optimal model are obtained through continuous machine learning of the training data of the entire inductance cycle,and the nonlinear modeling is completed by this method.On the basis of nonlinear modeling,a prototype is developed for static experiments to verify the accuracy of the least squares support vector machine modeling method.To verify the rationality of nonlinear model established by least squares support vector machine in two-phase 6/3 SRM system and simulation analysis,the SRM system simulation based on the nonlinear model of inductance and torque is built.The simulation results of the SRM system simulation model constructed by look-up table method and the LSSVM global nonlinear modeling method are compared to verify the rationality of the LSSVM model and realize the application of motor nonlinear model in system simulation.