| Switched Reluctance Motor (SRM) Drives (SRD) is an advanced electromechanical device. But the extensive applications of SRD have been limited due to its drawbacks of high vibration noise and torque ripple. This dissertation is devoted to fundamental and experimental research centering on the reduction of vibration noise and torque ripple in SRD. This dissertation is composed of following aspects. Since the stator radial vibration caused by radial force is the main origin of acoustic noise in SRD, the analysis and computation of radial force are the basis of vibration and acoustic noise research of SRD. In this dissertation magnetic tube method is adopted to deduce the analytic formula of radial force. Then the relationship between radial force and structural parameters of the motor is qualitatively analyzed. According to the principle of virtual-work, the formula of electromagnetic force based on magnetic vector potential is deduced. Only one time computation of magnetic field is needed when electromagnetic force is computed using this formula. Not only the amount of computations reduces, but also the accuracy of computation is improved by comparing the new method with the traditional virtual-work method. The precise numeric results of torque and radial force are achieved by using the new method. To resolve the problems of time consumption and the requirement of large quantity of finite element, which present in the performance simulation of SRD, this dissertation utilizes the powerful identification capability of non-linear model of Artificial Neural Network to implement the training of flux linkage inversion and torque computation of SRM. A precise analytic mathematic model of SRM is constructed in the end, based on Artificial Neural Network. There is a close relationship between the formal selection of SRM body and the level of vibration noise. From acoustic noise radiation and vibrative amplitude point of view, the determination of main dimensions of SRM is discussed. Then the selections of phase number, winding connection, pole number and number of branches in parallel are optimized according to the SRM's symmetry and order of force wave. Some common measures to reduce mechanical acoustic noise are summarized. The study of vibrative features of the system is important for reducing vibration noise. From motion equations, the transfer function from exciting force to vibrative acceleration and the free vibration results are derived. The analytic formulae of stator natural frequencies and vibrative amplitude of SRM are derived from electromechanical analogy method. Furthermore, various factors that influence vibrative amplitude are analyzed qualitatively. Finally, the conclusion is drawn that, based on finite element solution of energy method, by constructing SRM three-dimension finite element models using different structural forms, heights, numbers and shapes, the optimal radiating rib structure contributing to lowering the acoustic noise level of SRM and eliminating the heat is the radiating rib with the top-narrow bottom-broad trapezoid section of much higher and larger amount. Another conclusion is drawn that, by constructing SRM three-dimension finite element models of different winding assembly crafts, solidifying the windings makes the low order natural frequencies of the system ascend. Modal analysis is proved to be effective by comparing the modal analysis results with the measure results of the test prototype motor. Simulation is the effective way to compute the performance of SRD and assess the vibration of SRM. Static performance simulation, dynamic performance simulation and load jump simulation of SRD are implemented, based on the non-linear dynamic simulation model of SRD that is constructed by MATLAB. Then the switching angles are optimized, generally considering two optimizing targets of maximum average torque and efficiency, by using static performance simulations. At last, combining radial force table derived from finite element computation of magnetic field with static performance simulation, the radial force waveform is acquired by non-linear interpolation. The main harmonic components of radial force waveform are derived by the spectrum analysis of radial force waveform. At the stage of motor design, it is the primary condition of reducing acoustic noise to avoid the matching of main harmonics of radial force with the natural frequencies of SRM. Appropriate control strategy is necessary to reduce vibration noise of SRM. The formulae of time parameters of three-step commutation method are derived. Simulation results prove that the effect of reducing vibration by the new formulae is better than the effect of old formulae reported by previous literatures at small damping ratio. In the situation of more than one mode shapes being activated during the SRM runs, time parameters of three-step commutation method are optimized by numeric optimization method. As a result, the effect of reducing vibration is optimal overall. The method mentioned here is also effective to two-step commutation method. To overcome the deficiencies of the previous direct instantaneous torque control, some schemes are presented here, such as fixed frequency torque control instead of inner hysteretic loop control, torque control method at the beginning of overlap area, the second optimization method of the switching angles and the time parameter optimized three-step commutation method. These schemes are proved to be effective and have acquired the expected results by dynamic simulation. At last, three-step commutation method is adopted at each time of torque chopping. According to the real phase voltage, either two-step or three-step commutation method is adopted at the time of phase windings commutation. Furthermore, a switching frequency design method is presented, considering reducing vibration requirement. All these methods contribute to the control strategy which can reduce vibration noise and torque ripple simultaneously. Hardware implementation of a SRM controller has been achieved using Texas Instrument's TMS320LF2407 Digital Signal Processor. According to the requirements of control strategy, the system uses the circuit topology of asymmetric half-bridge. The scheme of MOSFET double-branch parallel connection is adopted for low cost and reliability. The control strategy presented in this dissertation is implemented within thecontrol software, which can reduce vibration, acoustic noise and torque ripple simultaneously. The static torque measure is done in the end in this dissertation. The results of magnetic field finite element computation are proved to be effective, by comparing the computative torque with the measure torque. A series of running tests of the prototype motor are conducted, according to the run condition of no-load and load, current control and torque control, low speed chopping and high speed single-pulse, and using two-step/three-step commutation method or not. By comparing the various results of these running tests, the control strategy presented in this dissertation is thus proved to be effective, which can reduce vibration noise and torque ripple simultaneously. Our research team has undertaken the electric vehicle fundamental special project titled "EQ6110HEV Hybrid Power City Bus Electric Motor and Controller"(2001AA501421) of the 863 Program of the "Tenth Five-Year Plan"of the country. The research work of this dissertation is sponsored by the project. Furthermore, the conclusions drawn in this dissertation on the structural forms of motor body, radiating rib structures, and measures contributing to reduce mechanical acoustic noise are validated by the 60kW prototype motors of the project.
|
PDF Full Text Request