Font Size: a A A

The Research Of Neural Network Control Strategy Of Inhibition Of Switch Reluctance Motor Torque Ripple

Posted on:2015-10-26Degree:MasterType:Thesis
Country:ChinaCandidate:J X ZhangFull Text:PDF
GTID:2272330434465756Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Switched reluctance motor (abbreviation: SRM)is a kind of efficient integrationproducts,which has many advantages and has been widely used.During the1980s A newtype of speed control system of switched reluctance motor (SRD) has shown powerfulvitality,with its unique advantages: high efficiency, good reliability, low costadvantages.It has a wide range of speed regulation performance, caused the wideattention of scholars at home and abroad.Because the structure and operationcharacteristics of switched reluctance motor problem, it is difficult for the traditionalcontrol system using the ideal control effect on Switched Reluctance motor,Withmagnetic saturation, eddy current, hysteresis and other highly nonlinearcharacteristic.So we need from the accurate modeling and high performance controlstrategy of switched reluctance motor, in order to give full play to the advantages ofswitched reluctance motor.First of all, in the MATLAB/Simlink environment,We use thefour phase (8/6) prototype, dynamic simulation model of switched reluctance motordrive system.On the basis of the simulation model,we can see the switched reluctancemotor of the torque ripple of great,according to the simulation results.In order to reducethe torque ripple in this paper presents a method of neural networks for control methodfor switched reluctance motor torque ripple suppression.In this paper, the author usesBP neural network.But the BP network also has its own shortcomings and deficiencies,so choose a quantum particle to a double chain structure optimization algorithm of BPneural network weights and threshold optimization.Then with the neural networkoptimized for SRM dynamic simulation data were generated off-line training, currentwaveform through off-line training the learning optimization.Finally, the training of thenetwork is optimized for torque control of SRM, nonlinear mapping in different T, θ, Icase.Using the hysteresis current control method, change make the winding current tofollow the reference current changes, and ultimately the torque ripple of switchedreluctance motor control.
Keywords/Search Tags:Switched reluctance motor, neural network, torque ripple, particle swarm
PDF Full Text Request
Related items