Self-sensing And Inverse-control For Five Degree Of Freedom Bearingless Switched Reluctance Motor Using Least Squares Support Vector Machines

Bearingless motor combines the functions of a motor and a magnetic bearing together within the same stator frame. It can produce the driving torque and suspension force on the rotor simultaneously so that there is no mechanical contact between the stator and rotor. Therefore, it pioneers a new field in the study of high-speed motor. Bearingless switched reluctance motor (BSRM) is a novel bearingless motor, which integrates the bearingless technology to the switched reluctance motor (SRM). It not only inherits the advantages of bearingless motor, but also enhances the high-speed performance and adaptability to atrocious surroundings of the SRM. Moreover, by the actively control of rotor radial displacement, it also provides a new approach to solve the problem of vibration and noises caused by asymmetric magnetic pull in the SRM. Therefore, BSRM becomes a worldwide research hotspot in the study of bearingless motor.The dissertation focuses mainly on the theory and realization of a five degree of freedom BSRM (5-DOF-BSRM), including the working principle and experimental prototype, the accurate analytical expression of torque and suspension force, the self-sensing algorithm of rotor position and displacement, the nonlinear decoupling control method and the digital control system design. The main researches and the corresponding results are as follows:1) The topological structure of5-DOF-BSRM is introduced, and a novel AC-DC3-DOF radial-axial hybrid magnetic bearing (HMB) is designed and the working principle of radial-axial HMB is introduced. Then, the stracture of three phase12/8pole double windings BSRM is introduced and the principle of torque and radial force production in BSRM is analyzed. Besides, the major electromagnetic and structure parameters of5-DOF-BSRM are designed and the experimental prototype of integrated5-DOF-BSRM is presented.2) The rotor dynamics model of5-DOF-BSRM is built in translational and rotational coordinate system and the rotor support characteristics are analyzed. The nonlinear suspension force model of radial-axial HMB in large air gap is deduced by equivalent magnetic circuit. Meanwhile, the torque and radial force model of BSRM in the case of eccentric and coupling is deduced based equivalent magnetic circuit, finite element analysis and virtual displacement principl. In addition, the coupling and nonlinear features are analyzed qualitatively and quantitatively in Matlab and Ansoft.3) To realize the self-sensing control for BSRM, a designing method of rotor displacement and position observers using least squares support vector machine (LS-SVM) is proposed. The state space model of BSRM is built and the design principle of LS-SVM observer is described. the stability of LS-SVM observer is proved by Lyapunov stability theory. Through offline training and online learning, the observers of BSRM are obtained. The findings show that the proposed method can observe the actual rotor displacement and position accurately, independent of mathematical model and specific parameters.4) To realize the rotor displacement self-sensing for radial-axial HMB, a predictive modeling method of rotor displacement for radial-axial HMB using LS-SVM optimized by particle swarm optimization (PSO) is presented. Through the collection of representative input/output data based on the nonlinear force model, the prediction model of radial-axial HMB is obtained by training LS-SVM. PSO is used to optimize LS-SVM parameters to improve the performance of the prediction model. The findings show that the prediction model can accurately predict the rotor displacement.5) In view of the coupling and nonlinear characteristic in model of BSRM, LS-SVM inverse model identification and decoupling control approach is proposed. The reversibility of the BSRM model is analyzed, and the adaptive genetic algorithm (AGA) is adopted to optimize LS-SVM to build the inverse model of BSRM. Then through the combining of LS-SVM inverse model and BSRM, BSRM is realized linearization and decoupling. Besides, the internal model controller is designed to achieve strong robustness and anti-interference ability for the decoupling control system.6) LS-SVM inverse-control method is employed to large air gap nonlinear control of radial-axial HMB. The reversibility of radial-axial HMB model is proved, and identification procedures of LS-SVM inverse model are elaborated, and the identification performance of LS-SVM inverse model is tested. Then the linearization and decoupling of radial-axial HMB is realized to series the LS-SVM inverse model with radial-axial HMB. On this basis, an improved PID controller is designed to achieve closed-loop feedback control, which improves the dynamic and static performance of LS-SVM inverse-control method.7) Two experimental systems for5-DOF-BSRM are presented based on digital control platform using dSPACE and DSP respectively. The control platform structure, power converter, drive and buffer circuit, displacement and position detection circuit, current and voltage detection circuit are introduced detailly, which establishes a solid foundation for developing the high performance control of the5-DOF-BSRM further.