Research On Modeling Method And Control Strategy For Switched Reluctance Motor

As a new motor drive system, switched reluctance motor (SRM) drive system has been extensively used from the1980s. Both the characteristics and control mode of SRM are significantly different from the traditional motor by virtue of its structure and dynamically characteristics. Building the accurately model of SRM is favourable to the optimization design, the dynamic performance analysis as well as the high performance control of the motor.Both the modeling methods and the control strategies are studied in this thesis, the main contents are as follows.Firstly, the mathematical model and basic operational principle of SRM are analysed in this thesis, at the same time, the support vector machine (SVM) regression theory including the ε-SVM, the υ-SVM and the least square support vector machine (LSSVM) is also studied. Moreover, two optimization methods of the kernel function’s parameters are given.Secondly, the nonlinear characteristics of SRM were studied, according to the observed data of the SRM phasecurrent, rotor angle and magnetic flux, The static model of SRM is built based on SVM method including the ε-SVM, the u-SVM and the LSSVM with different kernel functions. Aiming at the static nonlinear characteristics of SRM, compared to the neural network methods, the experimental results show that SVM and LSSVM methods give well performance in modeling aspect. In addition, the corresponding LSSVM will give the best performance once the parameters of LSSVM with RBF kernel function are further selected by ant colony optimization algorithm.Finally, for the speed control of SRM, taking a three-phase6/4SRM as an example, a neural network PID hybrid control strategy and its realization method are presented, the experimental results verify the feasibility of this proposed method. Simultaneously, for the torque control of SRM, in order to effectively restrain the torque ripple in operation process, the direct torque control (DTC) method of SRM has been studied on the basis of the DTC of the asynclironous motor. The performance of the above control strategies are further analysed from three aspects, namely, the torque ripple, speed regulation as well as the antijamming capability. Compared to neural network PID hybrid control method, experimental result indicate that the DTC method can well adjust the torque because of the torque feedback is introduced to DTC. Moreover, it can solve the problem of the torque ripple effectively.