Sensorless Control Of Switched Reluctance Motor Based On Neural Network

Since Switched Reluctance Motor(SRM)has come out,SRM have received extensive attention due to its simple and robust structure,reliable operation and high efficiency.Moreover,the Switched Reluctance Motor Drive(SRD)can operate over a wide range of speeds which compared to the conventional AC speed control systems and the DC motor systems,and the SRD's performance and economic indicators show significant advantages.Therefore,in summary,SRM has a good development prospect in practical applications.However,due to the special double salient pole structure inside the SRM and the flux linkage characteristic of SRM is the nonlinear relationship between rotor position angle and current,which makes it difficult to establish accurate nonlinear mathematical models.Moreover,the existence of a conventional mechanical position sensor makes the internal structure of the SRM more complicated and increases the cost.Thereby,the mechanical sensors reduce the reliability of SRD system operation so that it limits the application of SRM.Therefore,this paper explores the way in which the rotor position angle is obtained by indirect detection based on the operating parameters of SRM.Therefore,the method can eliminate the traditional mechanical position sensor inside the SRM,and realize SRM's position sensorless control and optimize the SRM's structure.In this paper,on the premise of in-depth analysis of the ontology structure and nonlinear mathematical model of SRM,the flux linkage characteristics of SRM simulation and experiment were obtained by finite element analysis and blocking experiment respectively.Since the rotor position angle,flux linkage and current have a nonlinear mapping relationship,it is difficult to establish an accurate mathematical model to express the functional relationship between the three.However,artificial neural networks(ANN)have strong nonlinear mapping capabilities and are widely used in nonlinear control systems.Therefore,this paper uses Back Propagation Neural Network(BPNN)to establish the nonlinear mapping relationship of the three.Moreover,using the improved Particle Swarm Optimization(PSO)to overcome the shortcomings of BPNN,such as slow convergence rate and easy to fall into local minimum value.And it overcomes the shortcomings of the traditional PSO algorithm,such as early convergence speed and low precision.In the MATLAB/Simulink environment,the simulation results show that the SRM's rotor position angle obtained by the BPNN model has a large error in the commutation interval.In view of the above problems,based on the traditional flux linkage and current as the double input signal of BPNN model,the inductor of curve fitting is proposed as the third input signal,this paper proposes the inductance of the curve fitting as the third input signal.The simulation results show that the error of SRM's rotor position angle obtained by the improved BPNN model is smaller.Thus the method implements position sensorless operation Finally,this paper builds the SRM hardware experiment platform with DSP TMS320F2812 as the main controller and designs related software programs.And using DSP chip's powerful signal processing capability and real-time control function to realize the position sensorless control strategy of the motor and laid the foundation for further work.