Research On The Vector Control System Based On The Back-to-Back Three-Level PWM Converters

The dissertation took the back-to-back three-level vector control converters as research models.Firstly, a comparison was made for three fast SVPWM algorithms.New general multilevel SVPWM methods were provided, which shows the absolute advantages of the established general multilevel SVPWM algorithms in briefness and FPGA resources use.A new hybrid SVPWM algorithm was proposed based on the single side diode neutral-point clamped (NPC)three-level converter. Different vector on-times factors were selected for different short vectors according to the diviation of capacitors voltages and the polar of three phase currents.Modulation was switched between the conventional SVPWM and the virtual vector based SVPWM according to the situation of three phase currents.Thus, a finer control to the neutral-point voltage balancing was obtained compared to the established hybrid SVPWM algorithm.Aiming at the neutral-point voltage balancing control issue of back-to-back diode NPC three-level PWM converters,a comprehensive predictive strategy for the grid side and motor side was proposed.The phase currents at both sides and the dc-link capacitor voltages were measured for the prediction of the neutral-point current.A quality function was found to balance the neutral-point, and a metabolic on-times distribution factor was used as a predicator to minimize the quality function at each switching state.Based on the conventional three-level rectifier direct power control (DPC),a new predictive three-level DPC algorithm was presented with an output of double non-zero vectors which distinguishes it from the conventional three-level DPC.The active power and reactive power were predicted to select the on-times distribution factor which enables the accurate control of the power.On the basis of analyzing the induction motor parameter identification of the two-level inverters,a mathematic model based on the recursive least square (RLS)algorithm for the static motor parameter identification of the three-level inverters was deduced.Simulation results were abtained under the unbalanced and balanced neutral-point voltage condition.A new model for the dynamic on-load parameter identification was built, for which a constant rotor flux was assumed.A control loop for the rotor flux was implanted into the model,which eliminates the influence of the flux coupling.A new random SVPWM method with fixed switching and sampling frequency was proposed.In this method, all vectors action time was randomly assigned between the ascent and descent stage of each triangle carrier on the basis of the unchanging voltage-second balance law, and the random effect dose not rely on the zero vectors.To insure better random effect of the SVPWM,a trebling randomized maximal-length shift register sequences (M-sequences) was designed.Simulation and experiment results show the proposed scheme leads to a well-distributed spectrum of the line-to-line voltage and phase current, and smaller harmonic amplitude at the integral multiple of the switching frequency compared with the conventional SVPWM.