Research On Model Predictive Current Control Of Inverter Supplied Induction Machine

In industry, control techniques of induction motor drives aiming for energy saving andsuperior performance have been gaining significant research attention. The state-of-the-artField Oriented Control (FOC) and Direct Torque Control (DTC) have various coverage inhigh performance industrial application despite of its disadvantages. In the recent ten years，Model Predictive Control (MPC) has emerged as a promising control technique in thecurrent control of power converter and AC motor drives due to its advantages such asstraightforward simplicity, good dynamic performances, accurate control capability, easyinclusion of system nonlinearities and restrictions, flexibility to control different variables.Based on the complementary strengths of the FOC theory and the powerful MPC method,this paper investigated the model predictive current control of induction motor drives,mainly including the basic ideas and characteristics of MPC, the MPC controller design andits implementation considerations needed to enhance the performance of the variable speedinduction system.The basic ideas and control issues of the model predictive control is firstly introduced.The two main parts of this control technique are then studied in details, i.e., thediscrete-time state-space model of the converter system and the current-oriented costfunction. The key elements affecting the overall current control performance are analyzedto provide a guidance of the MPC controller design for motor drives application.In order to deal with the computational challenge of the general MPC，especially forthe multilevel inverters and long predictions applications, two methods capable of onlineimplementation with considerably reduced amount of calculations are proposed. Onereduces the calculations taking advantage of the sector information of the reference voltagespace vector and the other works in a rules of adopting adjacent subset of voltage vectors. Acomparison between the general MPC and the proposed MPC is underlined in terms of thecomputational effort, the performance of the current control and the reduction of switchingfrequency is conducted.The basic theory and control structure of the model predictive current control methodare examined with the combined strengths of Vector Control and MPC. A discrete-time state-space induction model is adopted which features taking the stator current and rotorflux in the two-phase stationary coordination as the state variables. The model predictivecurrent controller is carried out in details, and the speed PI is designed in an optimalengineering way and a flux PI controller as well. A hybrid voltage-current model of a rotorflux observer aimed for effectiveness in whole speed range is analyzed. Besides, the delayof this predictive control method is compensated by using a two steps prediction. Acomparison among three different ways of discretization of the induction model isevaluated to investigate the influence of the accuracy of the prediction model on systemperformance, and the impacts of the typical errors in the stator and rotor parameters arestudied as well.The motor drives system is designed, including the design of a general three-phasetwo-level voltage-source inverter circuit and the programming of digital control based onthe DSP platform. The key characteristics of model predictive current control and theproposed modifications of MPC method, the current control and speed control performanceare validated with simulation and experimental investigations on the inverter circuit withthe resistive-inductive load and the induction machine, respectively.