Research On Control Method Of Switched Reluctance Motor For Vehicle

It is an inevitable trend for the development of the automobile industry that electric vehicles replace fuel vehicles.Countries all over the world are committed to promoting the development of energy vehicles.The performance of the motor system directly affects the performance of the whole vehicle.The unique characteristics of SRM,such as high reliability,strong fault tolerance and high overload multiple,are well matched with automobile performance requirements.The special core magnetic circuit structure and the torque ripple caused by nonlinear electromagnetic characteristics limit its application.In addition,the control of switched reluctance motor requires high position,and the cost and complexity of position sensor affect the promotion.Based on this,this paper mainly studies torque ripple suppression technology and sensorless technology of switched reluctance motor.The main contents of this paper are as follows,Switched reluctance motor has large torque ripple due to its special core magnetic circuit structure and nonlinear electromagnetic characteristics,and the design of high precision torque observer is the key to solve the problem.A torque observer based on stochastic dropout neural network is proposed in this paper,the stochastic dropout algorithm is introduced into the deep network model,The deep neural network has a computation model provided by a multi-layer non-linear hierarchical structure,and the neural network with different parameters is trained by using its powerful learning and expressing ability of complex relation between input and output data,by optimizing reconfiguration of network structure,the network model with stronger generalization ability is obtained,and the approximation and convergence ability are improved to avoid the network relying too much on some neuron nodes,and the problem of the overfitting fitting with weak generalization ability due to the model too fitting the training data.Through off-line model training,the nonlinear mapping characteristics between current-angle-torque discrete data is obtained,and the torque-sharing strategy is used to make the actual torque quickly track the given torque,avoid the storage space problem of look-up table method and complex operation problem of analytic method.The simulation and experimental results show that the proposed control method has obvious inhibitory effect on the torque ripple of the switched reluctance Motor(SRM).A large variation in current will generated when using the traditional current hysteresis current control method,consequencely,leading to torque ripples.In order to improve the current dynamic tracking ability and reduce the torque ripples,this paper proposes a novel method to enhanced tracking capability based on the torque distribution function strategy.The current tracking ability is improved without increasing the switching frequency or changing the hysteresis width.The PWM modulation mode such as "positive voltage-zero voltage" and "negative voltage-zero voltage" are selected flexibly to determine the freewheeling mode,demagnetization mode of the power converter.The saturation of the magnetic circuit are considered and the proposed approach is verified to be not affected when the motor operates under different load saturation condition.Aiming at the problem of the peak current and high torque ripple during the operation of switched reluctance motor,the inductance model of switched reluctance motor and torque distribution function based on the indirect torque control are introduced in this paper.Meanwhile,the relation between torque distribution function and waveform of output current is analyzed.The quadratic compensation function of linear torque distribution function is proposed.The current rate of change and the copper consumption are selected as the optimization goal.The optimal torque distribution function is achieved by using genetic algorithm to optimize model parameters of the torque distribution function.The speed and position estimation estimated are inaccura cy due to the change of the saturation inductance with the change of the current of the switched reluctance motor,a method of estimating the rotor position of the switched reluctance motor based on the six special inductance intersection is proposed.Full-cycle incremental inductance is calculated and the inductance curve is acquired based on injection voltage pulse which is exerted on the non-conducting phase and conduction phase current.The output intersection angle is determined by the adjacent two-phase inductance.The intersection of adjacent two inductance at high inductance area is influenced by magnetic saturation,and the intersection polynomial fitting method is used to obtain the relationship between the intersection angle and current function.Intersection position is obtained by the actual sampling current and the speed and position is estimated at any time.Finally,the estimation method is verified at inertial operation,light load,heavy load,load mutation and high speed by simulation and experiment.