Research On Indirect Detection Of Flux-Linkage And Rotor Position For Switched Reluctance Motor

Switched reluctance motor (SRM) has been applied in many fields due to its simple and strong structure, large starting torque, high operation efficiency, flexible controllability, strong fault-tolerance ability and wide speed range. Its doubly-salient structure and high saturation of magnetic circuit in operation results in the extremely complex magnetic field distribution and the high nonlinear electromagnetic characteristics, so it is difficult to establish the accurate analytical model through theoretical derivation. The existence of the position sensors brings about defects and inconvenience to the application of SRM. Thus, this dissertation focuses on the nonlinear modeling of the magnetic properties and the disposal of the position sensors, and lays emphasis on how to indirectly detect the rotor position with the magnetic properties model and the measurable information of motor.The flux-linkage characteristic is one of the significant characteristics of switched reluctance motor, whose magnetization curves family is the basis of the motor modeling, position detection and performance analysis. A combined method of the Finite Element Analysis and the experimental test was adopted in this dissertation, where the flux-linkage curves are obtained through the principle of magnetic co-energy correction. This method could accurately obtain the flux-linkage curves family without locking the rotor position, which lays the foundation of the research of modeling, simulation and the position detection algorithm for SRM.The nonlinear analytical model of the flux-linkage and the least squares support vector machine (SVM) model of SRM are established based on the measured data of flux characteristic. The method of the Fourier series transformation and exponential function fitting are used for the modeling of flux- linkage. In this method, the flux-linkage curve in the special position is modeled with the exponential function, and the flux-linkage in the middle position is described with the Fourier series with variable coefficient, so the analytic function model is obtained through four magnetization curves in the special position, which can satisfy the precision requirements, have fast calculating speed, and be available for the engineering application. In view of the advantages of support vector machine (SVM) in the small sample modeling, the SVM modeling method is studied, and the least squares SVM model of flux linkage-current-position angle is established, in order to improve its precision and generalization ability. Thus, a novel overall simulation model of SRM drive system is established based on the above two models, and finally the correctness and the validity of the established system model had been are verified through the comparison of simulation results and experimental results in the same condition.The flux-linkage characteristic is a nonlinear function of the phase winding current and the rotor position; it contains the rotor position information. The traditional flux-linkage detection method can obtain continuous rotor position information, but its algorithm is complex, computing time is long and memory is large. Firstly, a rotor position detection scheme based on the analytical model was proposed according to the established nonlinear analytical model of flux linkage in this dissertation which was practical by the simulation analysis. Then the position detection scheme is simplified and improved based on the principle of the simplified flux method, and a position detection scheme was put forward with synchronous operation of the double reference position analytical method and the standard position detection method. Finally, the simulation and experiments proved that the improved solution was a good way to estimate the rotor position angle of SRM in operation and adjusted the turn-on angle.The initial information of rotor position is a key problem for the starting operation of motor. The pulse injection method can solve the judgment problem of the initial conduction phase, but it could not realize the accurate positioning of the initial position angle, so it is difficult to achieve the best starting effect. This dissertation presents a combined method of the pulse injection and inductance model to estimate the rotor position angle. In the method, the high frequency pulse is injected into the phase winding, the phase inductance is indirectly calculated through the response current, then the accurate positioning of the initial position angle is realized through the phase inductance model according to the inherent relations between phase inductance and rotor position, and estimate the continuous rotor position information of low speed operation. The theoretical analysis and experimental research showed that the method is feasible and can effectively improve the starting effect of SRM.The sensorless drive system of SRM is developed with the core chip TMS320LF2812, where the related experiments are carried out, and the experimental results have verified the correctness of the proposed method in this dissertation.