Research On Nonlinear Modeling And Intelligent Control Strategy Of Switched Reluctance Motors

The Switched Reluctance Motor (SRM) has strong nonlinear feature due to its structure and operation mode, and the performances and control methods of this new kind of motors are significantly different from those traditional ones. As a result, the accurate modeling and high performance control of the SRM proves to be very important and has gained widely research.An experimental set-up based on TMS320F2812 DSP is built up to measure flux linkage characteristic of the SRM in this paper. Real time voltage and current signals are sampled during the conduction of the phase winding. The data are sent to an upper PC by serial communication. Flux linkage characteristic, inductance characteristic and torque characteristic of the prototype are obtained according to the analysis of these data, which lays a foundation for the modeling research of the SRM in the following study.A Wavelet Neural Network (WNN) model of the SRM is established based on the measured magnetization curves. The validity of this model is verified by simulation and experimental results. Based on the model, a dynamic simulation method for the Switched Reluctance Drive system (SRD) is proposed. Simulation results under variant control modes are given. This method provides some new approach for the SRD simulation.The Torrey model of the SRM has high accuracy, but its parameters identification is difficult to do. An improved Hybrid Genetic Algorithm (HGA) is proposed in this paper to solve the problem. It is proved that the improved HGA is more capable on solving this problem than ordinary Genetic Algorithm (GA), by comparison between optimization results obtaining from them respectively.Simulations and experiments of the prototype SRM under different work conditions are performed, from which the validity and accuracy of this method can be verified. The identified model of the SRM can be used to motor performances predicting and optimized control strategy development, and the application range of the Torrey model is consequently extended by the work in this paper.A modeling method for the SRM based on its geometry and material properties is proposed. Only geometry-based requirements are taken as input needed. Since complete magnetic characteristics are not necessary, a lot of preliminary measurements and calculations are eliminated. The dynamic performances under two operating conditions are predicted using the proposed method, and the measured results are obtained by an experimental laboratory setup. Furthermore, predictions using other two traditional models are presented for the purpose of comparison. Experimental and simulation results verify the accuracy and rapidity of the proposed method. ?The modeling method is meaningful to optimum design of SRM. It also is a powerful tool for performance analysis and high performance control of SRM.The SRM is difficult to control because of its strong nonlinear characteristic. In this paper, a self-adaptive PID control for the SRM based on neural network is presented. The parameters of this PID controller can be optimized real-time using BP neural network. Moreover, WNN is used to achieve on-line identification of the nonlinear system for the PID controller. Simulation results show that, comparing with the traditional PID control, the control strategy proposed in this paper can trace the given speed more quickly and stably, and can achieve self-adaptation better to the disturbance of system parameters.A speed regulating experimental system for the SRM is built up based on TMS320F2812 DSP. Experiments under different operation conditions are preformed. Results and analysis to them are given.