Current Optimizing-Based Torque Control For Switched Reluctance Motor

Switched Reluctance Motor(SRM)is a new type of special motor.Compared with traditional AC motor,SRM has advantages of simple structure,high reliability,large starting torque,no risk of breakdown,low cost,wide speed range and strong robustness.Based on these characteristics,SRM is not only widely used in traditional industries,such as textiles industries and household appliances,but also got attention by novel industries,for instance,electric vehicles(EV).However,due to its own double salient pole structure,the motor has serious magnetic saturation,which leads to strong nonlinearity between torque,current and rotor position.As a result,it is difficult to obtain an accurate motor model.Yet SRMs have advantages over traditional AC motors,severe noise and large torque ripple at low and medium speeds hinder the further application and development of SRMs in EVs and other industries.In this thesis,SRM is taken as the research objective.The torque ripple generated during operation of SRM is studied,and the torque ripple suppression control strategies are realized.This thesis proposes the following two control strategies:(1)Aiming at the problem of large torque ripple of SRM,a torque ripple suppression control algorithm based on Lagrange multiplier current optimization strategy and improved iterative learning control compensation is proposed.Under the condition of inductance constraint,the Lagrange optimal multiplier method is used to segment and optimize the current with constant set-up target torque.The iterative learning controller with error preprocessing is used to output the compensation current and superimpose with the optimized current.The reference current of the current hysteresis constant torque control realizes the closed-loop control of the torque and achieves the purpose of suppressing the torque ripple control.The results based on simulation show that the proposed integrated control strategy combining current optimization and iterative learning improves the convergence speed of the iteration and effectively reduces the torque ripple.(2)From the perspective of strong nonlinear characteristic model determined by the strong coupling of parameters caused by internal magnetic saturation of SRM,a current compensation control strategy is proposed based on Inductance-RBF double-hidden-layer neural network modeling using inductance nonlinear expression and segmented low pass filter based on rotor position.a)current-related parameters are introduced to the nonlinear model of flux linkage,as a result,the inductive nonlinear model of the current is derived,and the Inductance-RBF double-hidden-layer neural network current compensation controller is constructed: introducing the inductive base hidden layer to the inductive nonlinear model.The hidden layers of inductance,which taken nonlinear inductance formulation as activation function is introduced.Combined with the RBF hidden layer,this constructed neural network can learn the nonlinearity of SRM,and the three-phase currents of Torque Sharing Function method are compensated.b)under the constant target of torque,the first-order segmented low-pass filter based on the rotor position is proposed.The fluctuation of the superimposed current after the compensation of neural network is filtered out,and the torque ripple of the system can be further suppressed.The results based on simulation show that the proposed integrated control strategy combining current optimization compensation and first-order segmented low-pass filter effectively reduces the torque ripple.Based on the SRM experimental platform of TMS320F2812 DSP system,the experiment based on RBF double-hidden-layer neural network torque control is carried out,and compared with the torque control strategy based on conventional torque distribution function,the results proof the feasibility and effectiveness of the network torque control strategy that based on RBF double-hidden-layer neural network.