Research On Position Sensorless Control Strategy For Interior Permanent Magnet Synchronous Motor Drives

Permanent magnet synchronous motors(PMSMs)have been given due attention due to the very nature of high torque to weight ratio,high power factor,high reliability,less noise,and high efficiency for railway applications.Besides,PMSM removes the brushes,slip-rings,and field winding copper loss.Meanwhile,vector control of PMSM requires an exact rotor position.Thus,mechanical sensors have been used to collect the rotor position information.However,due to the challenges of the mechanical sensors related to hardware complexity,cost,noise effect,sensor failure,and others,sensorless control methods have been devised.The sensorless control procedure requires estimating the motor states and extracting the motor position information from the states.This thesis thoroughly studied the sensorless control strategy from low-speed to the rated speed for interior PMSMs(IPMSMs)in light of rail transit applications.In Chapter II,IPMSM modeling in different reference frames is detailed.The extended back electromotive force(EEMF)model of IPMSM to ease the sensorless control performance in a stationary and estimated reference frame is given.The EEMF dependence on the q-axis current makes the EEMF estimation strategy difficult.Thus,an active flux-based model of the IPMSM is briefly introduced.The active flux model converts the salient motor into a fictitious nonsalient motor.Thus,the sensorless control methods applicable for surface-mounted PMSMs(SPMSM)can be easily tailored.Furthermore,a vector control strategy summary is included.Finally,the proportional-integral(PI)controller with a time delay compensator is implemented.Integrating a controller with a disturbance estimation algorithm usually enhances the dynamic performance of the controller.Different disturbance estimation algorithms have been reported in the literature for various applications,including the motor drive system.The motor parameter variation and frame transformation due to position error cause a certain impact on motor state estimation.Thus,Chapter III presents an extended sliding mode disturbance observer-based(ESMDO)sensorless control of IPMSM.The effect of parameter variation and coordinate transformation is considered a lumped disturbance,and the ESMDO is designed.A Lyapunov stability theory is adopted to guarantee the stability of the proposed ESMDObased sensorless control method.The ESMDO has an inbuilt low-pass filter(LPF)and does not introduce a phased delay.A very large load change/reference speed variation may result in system instability or require higher controller gains that lead to system chattering.Therefore,a phase-locked-loop(PLL)-like speed-position identification method is designed considering the very nature of the dq-axis estimated disturbances.The speed from the Lyapunov stability is added with a PI compensator to enhance the speed estimation performance.Two compensation strategies are developed,taking that the estimated disturbances are a function of rotor position under no parameter variations.The first method uses the d-axis disturbance and is robust against permanent flux linkage and d-axis inductance variation.However,motor resistance may affect the system’s performance,especially in the low-speed region.Meanwhile,the q-axis inductance effect is independent of electrical speed and varies linearly with external load.The second method combines the dq-axis disturbances to design the compensator.A small-signal approximation and certain manipulations of the dq-axis disturbances have been adopted to obtain the input for the compensator.Thus,the method is robust against resistance variation.However,the permanent flux linkage variation sustains irrespective of the speed.To this end,a novel motor resistance variation voltage drop compensator for the first method and flux linkage variation back EMF compensator for the second method has been designed.The proposed motor resistance variation voltage drop compensator is enabled in the low-speed area and disabled for medium to the high-speed range to maintain the advantage of the first method’s robustness against flux linkage variation.Similarly,the proposed flux linkage variation back EMF compensator is enabled only in the medium to the high-speed range.Thus,the advantage of the proposed method for the lowspeed area for motor resistance variation is intact.The ESMDO-based current estimator with a PLL-like speed-position identification has a good dynamic and steady-state performance under numerous operating conditions.Simulation and experimental verification of the ESMDO are provided.The ESMDO has excellent sensorless control performance up to 5% of the rated speed.The conventional EEMF-based approach experiences all motor parameter dependency and external load change,especially for a high load system.Thus,an active flux-based model has been reported in the literature.The back electromotive force(EMF)can be estimated using the sliding mode observer(SMO),where a delay compensator is necessary.Meanwhile,the nonlinear sign function contributes to computational time.Thus,an advanced disturbance observer(ADO)-based back EMF estimation is introduced in Chapter IV.The back EMF observer only requires one observer parameter than the conventional SMO that requires two observer gains design.However,the phase delay demand is not discarded due to the inbuilt LPF.Therefore,further improvement of the ADO is proposed.A frequency adaptive secondorder disturbance observer(FASODO)is designed for back EMF and active flux estimation.FASODO is a frequency adaptive approach and eliminates the phase delay compensator demand of the traditional SMO and ADO-based back EMF estimator.Besides,FASODO estimates the active flux and the back EMF simultaneously.As a result,a quadrature-PLL(QPLL)using either the back EMF or the active flux and a tan inverse approach can be used for speed-position identification.Harmonic elimination and dc-off set analysis of the FASODO have been presented.Meanwhile,the conventional Q-PLL has a position error during ramp frequency tracking.Thus,two methods are proposed.The first method is called an integrally compensated quadrature-phase-locked-loop(IC-Q-PLL).The second method adopts the d-axis component for accurate position error compensation.Thus,Maclaurine expansion has been used to obtain the speed estimation error where the position error is obtained through integration.Consequently,the position error during ramp frequency tracking has been greatly improved.Experimental and simulation studies for various operating conditions showed that the proposed methods are effective for sensorless control of IPMSM.FASODO delivers good sensorless control performance up to 10% of the rated speed.A low-speed sensorless control strategy using a model-based approach is mostly affected by invertor nonlinearity and motor resistance variation.Although motor resistance estimation could improve the sensorless control performance for the model-based approach,sensorless control is impossible for ultra-low speed due to the difficulty of observing the back EMF or flux.Thus,the saliency approach has been used.A pulsating high-frequency(HF)signal injection-based low-speed sensorless control of IPMSM using a conventional stationary reference frame LPF may experience a phase delay.Thus,the phase delay will contribute to dynamic and steady-state performance degradation unless fully compensated.Similarly,the heterodyning process may exhibit time delay and bandwidth restriction.Thus,an improved approach with a single high-pass filter(HPF)in a stationary reference frame is proposed.The rotor position error is extracted from the HF current using the sign of sine of the injected signal frequency considering the phase shift.Consequently,the estimated reference frame LPF in the heterodyning process and the delay compensator for a stationary reference frame LPF-based approach are discarded.A phase-shift tracking observer(PSTO)independent of position error is included.An online daxis HF current amplitude estimation is adopted to normalize the effect of d-axis inductance variation.The comparative study with a stationary reference frame LPF and the heterodyning process has shown that the proposed self-sensing strategy delivers an excellent sensorless control performance.Thanks to the online d-axis current amplitude estimation,the impact of d-axis inductance variation on speed-position identification has been improved.Finally,a hybrid method is introduced for the whole speed range.The proposed modelbased approaches,ESMDO and FASODO,are integrated with the proposed HF-based selfsensing to ensure a smooth transition from low-speed to high-speed.The proposed methods are validated with a hardware-in-the-loop(HIL)experimental setup.