High Performance Model Predictive Control For Induction Motors At Low Switching Frequency

Medium voltage ac machines fed by high-power inverters operate at lowswitching frequency to increase output power. A reduction in switching frequencyincreases harmonic distortion. To ensure accepted waveform distortions and allowfast dynamic responses at low switching frequencies is a challenge problem.Supported by the project of National Natural Science Foundation of China(51377102) and Grants from the Power Electronics Science and EducationDevelopment Program of the Delta Environmental&Educational Foundation(DREG2013009)，this dissertation has done thorough and comprehensive researchon the high performance control for induction motor at low switching frequency of200~300Hz. The main contributions of the dissertation are as follows:Model predictive control is introduced into inverters driving motor control field,a novel model predictive direct torque control (MPDTC) is presented. A costfunction evaluates the square sum of torque and stator flux magnitude errors toremedy the drawback with large pulsating torque and stator flux output of thetraditional DTC. The number of switching step is added in cost function to reduceswitching frequency properly. The performance of presented MPDTC scheme iscompared with traditional DTC.Based the one-step MPTDC, a novel model predictive direct torqure control(MPDTC) methodology with long prediction horizon for low switching frequencycontrol of induction motors fed by a NPC three-level inverter is proposed. Accordingto a receding horizon policy of the cost function which evaluates the averageswitching frequency, the optimal switch vector is seleted. The infeasibility problemof the optimal control is solved by update rule of the cost function. By utilizing thisalgorithm, the switching frequency can be dropped around300Hz and a lowharmonic distortion can be achieved, meanwhile, the fast transient response is alsoinherited. Simulations based on4kW and1.6MW drives verified its effectiveness.On the basis of the presented MPTDC method, a low switching frequencymodel predictive direct current control (MPDCC) is developed. Compared to theestablished one step predictive current control, the proposed MPDCC scheme canreduce the switching frequency significantly and keep the similar harmoniccharacteristic. Simulation results demonstrate that the switching frequency can be dropped below300Hz and a low harmonic distortion rate has been achieved.Lower harmonic distortion at low switching frequency calls for adoptingoptimal pulse patterns, but it cannot be directly used for high performance system,thus a novel stator flux trajectory tracking control (FTTC) system is studied in thisdissertation. According to the proposed PWM correction method, the closed-loophigh performance control of optimal PWM based on FTTC with the self controlledmachine model can be realized. The simulations demonstrated that both fast dynamicresponse and low harmonics characteristic can be achieved at the low switchingfrequency of200~300Hz, its harmonic characteristic is more outstanding than thepresented MPDTC and MPDCC methods.The state of art method that combines the optimal PWM with the very fastdynamics of trajectory tracking control is presented. A constrained optimal controlproblem with a receding horizon policy is formulated and solved. Two modelpredictive flux trajectory tracking control schemes, which are based on DB and QP,are discussed. Fast dynamic control is to be achieved while performing the minimumpossible modification of the offline calculated PWM by the QP method. This systemis simpler than the FTTC with the self-controlled machine model and can achieve thesimilar dynamic response performance and low current total harmonic distortionwithin5%.