| Electric Multiple Units(EMU)system employing permanent magnet synchronous machine(PMSM)serves as the electrical traction core in the next generation train and the fault of position sensor during operation results in potential risk,which needs to be solved.Therefore,sensorless control method is essential for its safety operation.Due to the low switching frequency and multi-mode PWM modulation of the traction inverter,the phenomenons that rapid change of harmonics,modulation modes switching,and long system delay all increase the design difficuly of sensorless control algorithm.In order to solve these problems,this dissertation studies the sensorless technology in low-switching-frequency-based multi-mode PWM modulation,including high-frequency injection method in asynchronous modulation,back-EMF-based method in multi-mode synchronous modulation,design of position observer and closed-loop regulators and delay performance optimization in hybrid-position-observer-based sensorless control.The research results provide support for the sensorless control in EMU train.Firstly,asymmetric space vector modulation(ASVM)-based high frequency(HF)square-wave injection is proposed in this paper.Considering that the extraction process of fundamental current and high-frequency current in the traditional signal demodulation requires the use of low-pass or band-pass filters,the dynamic of sensorless closed-loop system is reduced.Hence,in the signal processing process,the filtering effect is realized by a simplified mathematical operation.Based on this,the ASVM is used to optimize the harmonic performance of the indcued HF current under low switching frequency,and harmonic performance comparison is given in the corresponding speed range.The effectiveness of the sensorless control is verified on a 3.7k W interior PMSM(IPMSM)platfom.Secondly,a non-singular sliding mode observer(NTSMO)-based smooth sensorless drive method is designed in the multi-mode synchronous modulation region.Based on the harmonic characteristics of 21-pulse regular synchronous modulation and selective harmonic elimination PWM(SHEPWM),in the switching process of different SHE modes,the influence mechanism of harmonic current shocks on the estimated back-EMF is analyzed.By establishing a harmonic equivalent circuit,when the phase-angle of the fundamental voltage is located at the middle60~°,the harmonic shock current amplitude is zero.According to the equivalent principle,the three-phase fast transition can be completed in the half fundamental period of the A-phase voltage,then the estimation error caused by the harmonic current shock can be eliminated in the transition process.On the other hand,a slight fluctuation inevitably occurs in the estimated speed,which cause SHE modes repeatedly switch,and the smooth transition method is invalidated.Therefore,a frequency hysteresis zone with load torque disturbance compenstion is set near the switching speed,which improves the robustness of the sensorless smooth drives.Thirdly,under low switching frequency,the dynamic optimization method of sensorless closed-loop system based on the design of regulators and position observer is studied.Using the traditional non-singular sliding mode surface,a variable gain sliding mode acceleration factor is introduced to increase the convergence rate when the system states are far away from the equilibrium point.A universal design of the speed regulator,current regulator and position observer using the improved sliding mode structure is presented.In the case that the convergence of sliding mode control law is guaranteed,the sensorless closed-loop control can be implemented with only one set of design parameters.In order to verify the effectiveness of proposed method,the dynamics of the estimated speed,current and estimated position are selected as comparison objects.Simulation and experimental comparisons are made with traditional non-singular sliding mode and fuzzy PI adjustments as counterparts.Finally,this dissertation studies the system delay optimization strategy in full-speed range for hybrid position observer.The influence of inverter nonlinear delay and inherent calculation delay in the square wave injection method,the back-EMF model method and the hybrid position observer are analyzed.Based on the theoretical analysis of the system delay,two delay compensation strategies are proposed.In the first strategy,taking the inverter nonlinearity delay as the starting point,a compensation time calculation method based on q-axis voltage error is designed,and the compensation for the on-time of three-phase IGBT is completed.Considering the total time delay,the Euler prediction theory is introduced in the calculation of the voltage error,and the inverter nonlinearity delay compensation is converted into the total delay compensation,which is suitable for carrier-based asynchronous modulation and regular synchronous modulation mode.After the inverter enters the SHEPWM mode,according to the pulse reconstruction theory,it is only necessary to predict the rotor position angle and the voltage vector angle to eliminate the influence of the system delay.In the second strategy,the current lag-phase-angle caused by the system delay is estimated as the target.Based on the basic theory derivation,the q-axis current error can be used as the input of the PI compensator to obtain the estimated phase lag angle.The phase compensation of the estimated rotor position angle is carried out,and the amplitude error correction of the feedback dq-axis currents is performed to realize the global delay compensation of the sensorless closed-loop system.By comparing the power spectral density(PSD)performance of the two delay compensation strategies in the sensorless control,the advantages and disadvantages of the two methods are obtained,and the correctness of the theoretical analysis is verified by experiments. |
PDF Full Text Request