Research On Parameter Identification Method And Vector Control Of Three-Level Inverter System

Three level vector inverter has been widely used in AC speed control system, and field oriented control of three phases asynchronous motor is one of the most mature control technology, F.O.C is short for the field oriented control. With the widely use of the F.O.C method, AC variable speed drive system has greatly improved its control performances, such as dynamic response and high accuracy. Though the performance of vector control can compared to the DC variable speed drive system, it also influenced by the motor parameters, for vector control based on dynamic mathematical model of asynchronous motor. In fact, the motor parameters vary with the temperature and frequency, which result in the bad control performance. In case the time-varying parameters deteriorate the control effect, a robust rotor flux observer is needed, as well as a better parameter identification system.Frist, the frontier theory and method of field oriented control, coordinates systems and transformations, mathematical model under different coordinates are involved in this thesis. The simulation by the MATLAB/Simulink, proving that the field oriented control has very high speed accuracy, and can well adapted to the changes of load disturbance. Besides, the performance of different flux observer models is analyzed in the frequency domain, including the convergence and parameter sensitivity. Combine the advantages of both voltage model and current model, a new rotor flux observer was designed, and it can adjust the proportion of the two models automatically according to different speeds.With inherent hardware resources of three-level inverter, by injecting voltage and current test signal to the motor, the parameters can be identified before the normal work of the motor. The offline parameter identification can be finished conveniently with a zero rotor speed, and can also provide a relatively accurate initial value for the online vector control.Finally, a kind of online parameter identification method based on full-order state observer was introduced, which focus on overcoming the time-varying parameter problems. Rational design of the feedback matrix gain and adaptive law, can ensure the global stability of the full-order state observer system, as well as enough convergence speed and recognition accuracy. Then real-time motor parameters can be obtained from the full-order state observer system, to ensure the precisely decoupling control of vector system. At last, the simulation result can verify the correctness of the theory and design.