Reserch On Sensorless Control Techniques Of Permanent Magnet Synchronous Motor For Metro Applications

As one of the promising candidate of next-generation urban rail transit,permanent magnet synchronous motor(PMSM)traction drive system attracts much popularity due its advantages,e.g.,high power to weight ratio,high efficiency,low noise and so on.In the IPMSM drives,the speed and position information should be fed back to achieve high performance.This task,generally,is carried out by mechanical sensors which is connected to the motor shift.However,the use of sensors increases system cost and degrades system reliability.Moreover,the performance of sensors is obviously deteriorated in a harsh environment.Therefore,it is a trend to improve system reliability and reduce system cost is to use the sensorlss techniques and accordingly remove the sensors in IPMSM drives.In this paper,the sensorless techniques for IPMSM drives are focused on,and the main works are introduced as follows:In Section Ⅱ,based on the mathematical model of IPMSM,an introduction of the vector control of IPMSM drives with the design of proportional-integral(PI)controller-based current loop is first presented.Considering the adverse effect of the control delay under the low switching frequency,the inverter-motor model in the discrete domain is established.With this model,a complex-vector current controller is designed to replace the conventional PI-type current loop.A performance comparison between the designed current loop and the conventional current loop is made by using the pole-zero plot in the discrete domain.According to the results,it is indicated that the decoupling capability of the designed current loop is better than that of the conventional current loop under low switching frequency and with parameter mismatches.In Section Ⅲ,an extended back electromotive-force(EMF)sliding mode observer in stationary reference frame is used to obtain the extended back-EMF signals.Considering the conventional phase-locked loop(PLL)-based estimation scheme provides position with an unsatisfactory accuracy during acceleration/deceleration process.To tackle this,a simple yet effective estimation scheme based on the high-order PLL is proposed for sensorless control of IPMSM drives.In this scheme,the accurate estimation under acceleration and deceleration conditions can be obtained by increasing the system order.Moroever,the estimated back EMF signals may be distorted by various disturbances,e.g.,amplitude mismatch and abundant harmonics.Due to this,the pre-filter and an amplitude normalization(AN)unit are employed into the implementation of the proposed estimation scheme to guarantee the estimation performance.In Section Ⅳ,the issue of parameter variations adversely affects the performance of the extended back EMF observer.The relationship between the rotor position errors and the parameter variations,e.g.,inductance and stator resistance variations,is elaborated.Meanwhile,the effect of the parameter mismatches on position estimation errors under different operating conditions is detailed.On this basis,according to the operating characteristics of metro traction drive system,the parameter identification scheme combining the inductance off-line identification with stator resistance on-line identification is assisted in the estimation scheme to ensure the estimation performance without reducing the system bandwidth.Finally,based on the experimental set-up including an induction motor and an IPMSM,the effectiveness of the proposed methods in this paper is experimentally investigated.