Research On Sensorless Control For PMSM Over Wide Speed Range

Permanent magnet synchronous motor(PMSM)has been widely used in many drive fields such as household appliances,industry,transportation and aerospace,due to its advantages of high power density,high efficiency,high reliability and easilyimplemented field-weakening operations.The high performance control of PMSM is based on accurate rotor position information.But traditional mechanical position sensors are limited in some occasions because of their vulnerability to interference,difficulty in installation and maintenance,poor adaptability to the environment,and increasing system volume and cost.For these reasons,sensorless control technology was proposed and developed into a hotspot in the filed of electric drive.The contribution of this paper is to achieve sensorless control for PMSM with wide speed range and easy industrial application,which combines strong robustness control theories including signal injection,active disturbance rejection control,sliding mode control,and parameter identification.The main research work of this paper as follows:By deducing the mathematical model of PMSM,analyzing the PWM control of power converter and the optimal trajectory of PMSM in full speed domain,the frame of PMSM vector control system is established.In the framework,the systematic comparison between these two systems,which based on PI controllers and linear active disturbance rejection controllers(LADRC)respectively,illustrates that the system based on LADRC has not overshoot in the inner and outer loop and it is insensitive to parameter changes.Moreover,this system does not need integration process,and it has the better dynamic performance and easier parameter tuning because of the extended state observer estimating and compensating disturbances in real time.But the dynamic response is slightly slower.For equal switching frequency square-wave pulsating injection method,the transient and steady state performance differences of filterless method,lowpass filter(LPF)method and double-delay method in extracting fundamental wave and high frequency response current are systematically compared.The results indicate that in the case of extracting control current,the torque ripple in LPF method is slightly larger,but it is totally acceptable.Considering the time delay,the filterless method has less time delay for small inertia motors,but more time delay for large inertia motors or heavy load operation.When extracting high frequency response current,filterless method and double delay method have the same extraction accuracy,but the double delay method has stronger attenuation effect on harmonics,less delay of filter for estimated speed and better dynamic performance of the system.For comparison,the sinusoidal pulsating injection method is briefly described and simulated.In addition,based on the stability analysis of phase locked loop,a set of parameters selection principles based on the concept of "bandwidth" is derived.For the sensorless strategies based on first-order sliding mode observer on stationary ?? coordinate and estimated ?? coordinate,a better parameters selection method about sliding mode gain and boundary layer is proposed.And systematic comparative analysis is made for the performance differences on these two coordinates.The results show that on the ?? frame,the incomplete compensation of the phase shift makes the position error increase with the decrease of the speed,and the loading capacity at medium and low speed is limited.On the ?? frame,the speed range of the system is narrower than that on the ?? frame owing to the facts that sliding mode observation contains the speed error term and the observation valus deviates from the actual value when the speed is lower;but the constant position error makes it have stronger loading capacity.To resolve the problem that the zero theoretical error of the ?? frame has not been fully utilized,the parameter adaptive second-order super-twisting sliding mode observer is adopted to reduce the chattering of the system effectively,eliminate the low pass filter and compensation module,and then improve the loading capacity and speed range of estimation algorithm on ?? frame.In addition,for the inherent chattering problem of sliding mode observer,a sensorless control strategy based on LADRC is adopted.The influence of the variation of motor parameters on the position error is analyzed.A strategy of off-line identification of resistance and quadrature inductance by recursive least square method with forgetting factor is proposed,which effectively reduces the position estimation error.Based on the amplitude of virtual dq axis currents,the initial rotor position is detected.The method has high signal-to-noise ratio,of which current regulation and pole judgment are not needed,and positioning error is controllable.The switching algorithm with speed-weighted and normalized position estimation error achieves smooth switching between zero low speed range and medium high speed range.In view of the switching process from zero speed to available estimated speed based on model method,an open speed loop with stabilizing loop and regulated current loop transition strategy is proposed.The introduction of speed stabilization loop accelerates the natural stabilization process of the motor and makes the switching process more stable and reliable.The PMSM towing experiment platform is built and perfected,and the feasibility of the proposed algorithm and the correctness of theoretical analysis are verified by experiments.