Research On Robust Model Predictive Control Of Induction Motor Driven By Three-level Inverter

In the field of power transmission,induction motors are widely used in practical engineering such as metallurgical steel rolling,locomotive traction,and ship propulsion and other important national economic occasions,due to their simple structure,stable performance,and high cost-performance ratio.At present,the power core of our country’s high-speed rail EMUs is a high-power AC induction motor.CRH380 A,the most advanced EMU in my country,is a three-phase AC induction motor.In these application areas,high-performance motor control technologies such as high steady-state speed accuracy,fast torque response,and large power output are required.Neutral Point Clamped(NPC)type three-level inverter has the advantages of low waveform harmonic content,high efficiency,high power output,etc.,and has been widely concerned and applied.In the induction motor control system,the traditional control methods mainly include vector control,field-oriented control and direct torque control.For the fast-developing industrial field,traditional control strategies cannot meet the ever-evolving industrial needs,so some new modern control methods have been proposed by scholars.Finite-Control-Set Model Predictive Control(FCS-MPC)has achieved great success in complex industrial processes since its inception.It has developed from the original heuristic control algorithm to a new branch of the industrial field.Therefore,this thesis adopts an improved FCS-MPC strategy for the induction motor system driven by an NPC type three-level inverter,which effectively reduces the amount of prediction and calculation of the three-level inverter model while suppressing midpoint voltage fluctuations,and combines Sliding mode control enhances the parameter robustness of the improved FCS-MPC.The control performance of the induction motor system directly depends on the speed measurement feedback.In terms of speed measurement,speed sensors have many shortcomings in terms of hardware,cost,and size applicability.Therefore,the design of speed sensorless is a very important part.The specific research contents of the thesis are as follows:(1)Establish mathematical models of induction motors driven by three-level inverters in different coordinate systems,and improve the original algorithm for the three-level model predictive control with a large amount of calculation and the problem of midpoint voltage balance control.Improve the working principle of the model predictive control,and compare and analyze the calculation amount with the existing literature.The simulation experiment shows that the improved FCS-MPC designed and the conventional FCS-MPC algorithm are compared.The two systems have similar control performance and can Effectively suppress midpoint potential fluctuations.(2)In order to improve the parameter robustness of model predictive torque control,sliding mode control is integrated into the improved model predictive algorithm.Combining the integral sliding mode and the super-twisting sliding mode,the arrival stage is eliminated by selecting the integral sliding mode surface,and the second-order super-twisted sliding mode can effectively reduce chattering.According to the designed integral super-twisted sliding mode control law,the torque controller,flux linkage controller and stator flux linkage observer are designed,and a proof of Lyapunov stability is given.Through Matlab simulation and StarSim real-time simulation platform,the experiment results show that the model-predictive torque control system based on the integral super-twisted sliding mode designed in this paper show good robustness.(3)Designing a sliding mode speed controller with integral super-twisted sliding mode as a substitute for PI speed controller.The speed observer consists of two current observers and a rotor flux observer.The two sliding mode current observers are used to compensate the influence of parameter changes on the rotor flux estimation.The rotor speed is obtained through an adaptive law.The experimental results of Matlab simulation and StarSim real-time simulation platform show the effectiveness of the designed algorithm.