Study On Parameter Identification Of Switched Reluctance Motor Base On Direct Torque Control

As switched reluctance motor(SRM)has features of simple structure,low cost,high reliability and superior performance etc.,which make it has attracted general concern.In recent years,Many scholars apply direct torque control(DTC)method into switched reluctance motor(SRM).The whole direct torque control system's accuracy has close relation with accuracy of timely feedback quantity of motor.In fact,in standard direct torque control system,there is only voltage and electric current sensor.With only voltage and electric current in the control process,flux linkage from them is the most important factor of control accuracy.In traditional calculation of flux linkage according integrating exist many issues,then accuracy of relevant factors in flux linkage's calculation is the guarantee of distinguish the flux linkage's accuracy.First in integral model,resistor is the key of flux linkage,especially in low speed,its influence is more obvious.Stator resistor is influenced by temperature mainly.In order to avoid the complexity of the control system,temperature sensor is not allow.This article through optimizing BP neural network,study the relation between electric current and resistance in two different resistance models.In basis of the traditional BP neural network,optimizing select the number of neuron in each layer,optimizing select learning rate etc.Then proposing a method of stator resistor identification based on electric current error.The experiment results show that the resistance recognizer has a simple algorithm,good real-time performance.It can satisfy the control need.Second,against the issue about integral model's error increasing when the integral increases.Torrey proposed another SRM flux linkage model.The model has high precision matching feature,can describe the feature of flux linkage accurately.But the parameter identification is difficult.Against to solving this issue,this study proposes a modified genetic algorithm and applying it into identifying parameter of SRM motor model.Modifying genetic algorithm can avoid precocity.The experiment results show this genetic algorithm can better solve this issue then basic genetic algorithm.Under the Matlab2014a/Simulink simulation environment,this article builds SRM motor direct torque control system simulation model which add resistance recognizer and Torrey flux linkage model.The simulation results show two algorithms can optimize traditional direst torque control.At the same time,comparing performance of these two algorithms,which makes better basis to expand applying these two algorithms in the future.