Research On The Position Estimation Of SRM Based On Few Flux-Linkage Characteristic Data

Switched reluctance machine(SRM),with its excellent performance of simple structure,flexible control,wide speed range,and adapting to the harsh environment,has broad application prospects and huge potential for development in both the civil and the military field,such as more/all electric aircraft,electric vehicles and wind power.It is necessary for the high performance control of SRM to get the exact position information.However,the position sensor not only increases the cost,but also greatly reduces the reliability of the system and limitsthe scope of its application.Therefore,it has been a hot spot to find a practical sensorless control method for SRM.Accurate static electromagnetic characteristics are essential basic data in the modeling of SRM for performance analysis and sensorless control purposes.This thesis presents a rotor-clamping method to accurately measure the flux-linkage and static torque characteristics of SRM,and its accuracy is preliminarily verified by three ways: coenergy method,finite element method(FEM)and inductance measurement method,and further verified by the comparison of dynamic performance.To solve the problem of complex measurement process and high cost for the rotor-clamping method,this thesis proposes a new torque-balanced method to measure the flux-linkage characteristics.Without any position sensor,the proposed method takes full advantage of the symmetry of SRM structure,and the flux-linkage characteristics of SRM are measured at four specific rotor positions by different excitation forms to keep the rotor still.To improve the accuracy of the measurement,the effects of magnetic coupling among phases are evaluated in detail.Compared with the results from the rotor clamping method,the accuracy of proposed method is verified.This thesis proposes a combined modeling method to deal with the problem that few flux-linkage characteristic data is not enough to build the model directly.The proposed method mainly consists of two steps,namely data reconstruction and characteristic description.In data reconstruction step,the entire flux-linkage characteristics are obtained by training support vector machine(SVM)with the measured few samples.In characteristic description step,back-propagation neural network(BPNN)is adopted to describe the reconstructed flux-linkage characteristics and calculated static torque characteristics.On this basis,the simulation model of the SRM prototype is built in Matlab.The results from simulation under different conditions are compared with those from experiments,and good agreements can be found,which prove the effectiveness and accuracy of the proposed modeling method.The applicability of the proposed method to different SRM topologies is discussed as well.This thesis proposes a novel accurate position estimation method based on few flux-linkage characteristic data.In this method,flux-linkage characteristics are devided into two regions,and linear flux-linkage model and monadic linear regression analysis are used to estimate the positions in two regions,respectively.To further improve the accuracy of the position estimation,a multiphase estimation method is considered,which can select the optimal phase to estimate the position according to the different conditions automatically.The feasibility and accuracy of the proposed method is verified by detailed simulation and experiment under different operating conditions.The proposed method can effectively avoid the influence of mutual inductance coupling and magnetic saturation on the accuracy of position estimation,and has the advantages of strong anti-interference capability and extensive applicability.Furthermore,the effects of open-circuit fault and static eccentricity fault are also discussed in this thesis,which indicates the good fault tolerance of proposed method.