Research On Position Sensorless Control For Ipmsm Drives

Interior permanent-magnet synchronous motors(IPMSMs)have been widely used in electrical drive fields such as industry,aerospace,transportation and domestic appliances due to the distinguished advantages of high efficiency,high power density,and easily-implemented field-weakening operations.Since the merits such as reduced system cost and enhanced reliability can be achieved through position sensorless(self-sensing)control,much focus has been put on the position sensorless IPMSM drives of higher accuracy,wider speed range and stronger robustness.However,several key issues and challenges remain to be addressed,such as the limited dynamic performance due to the filtering process for the high-frequency(HF)injection based method at low-speed range,the position error fluctuations for model-based method,torque or current ripple caused by the inverter nonlinearities,and the stability of the controller and the position observer for low frequency ratio operation.Therefore,it will make a big difference to broaden the applications of position sensorless drives through further exploring the position sensorless control and breaking through the key issues and challenges remained.Commonly,the filtering process is necessary to demodulate the position error signal and track the position and speed information for the conventional signal injection based methods applied to low-and zero-speed operation for position sensorless drives.The challenges of limited bandwidth and dynamic performance along with the increased resource utilization due to the filtering process are analyzed.Consequently,to eliminate the filtering process and further increase the dynamic bandwidth,a promoted filterless square-wave injection scheme is proposed to implement carrier separation and demodulation through analyzing the timing sequence of the injected square-wave voltage vector and the HF current response.And meanwhile,a modified position tracking observer is adopted.With regard to the initial position detection for position sensorless IPMSM drives,existing saliency-tracking-based methods suffer from the implementation problem in determining the amplitude and width of the pulses for the short pulses injection method and from the low signal-noise ratio for the position-dependent secondary-harmonics-based method.Based on the magnetic saturation effect,the magnetic polarity can be identified without interrupting the HF square-wave injection from the amplitudes of the induced d-axis HF current against two d-axis current offsets which are equal and opposite in direction.The proposed initial position detection is well applied to the standstill and free-running rotor applications with a fast convergence speed.The(1±6k)th harmonic exists in the stator current and the back electromotive force(EMF)estimates due to the influence of the inverter nonlinearities and the flux spatial harmonics,which gives rise to the(6k)th harmonic ripple in the rotor position estimate.To accomplish accurate decoupling and further improve the performance for the position sensorless IPMSM drives,an improv ed adaptive linear neural(ADALINE)filtering rotor position observer is presented based the active flux model considering the EMF harmonics.The selective ripple harmonic components of the equivalent EMF can be eliminated through the proposed ADALINE filter based on the adaptive cancelling principle,whereby,the accuracy of the rotor position obtained from the software quadrature phase-locked loop(PLL)can be improved.On the basis of the estimated harmonic information,the filter coefficients can be updated continuously online through the least mean square or recursive least square algorithms which guarantee the fast convergence rate of the rotor position observer.With regard to voltage error caused by the inverter nonlinearities,the d-q-axis currents suffer from the(6k)th harmonic ripple for the position sensorless drives.A dual adaptive vectorial filtering(AVF)harmonic decoupling network(HDN)based compensation scheme is proposed with the analysis of the inverter nonlinearity effect considering the dead time.A set of AVF-HDN is adopted to track the rotor position and speed information,eliminating the position error fluctuations;and meanwhile,another set of AVF-HDN is utilized to detect the voltage error in time,which is added to the output as a feedforward term for compensating the inverter nonlinearities,eliminating the d-q-axis current ripples and improving the dynamic performance.It is not necessary to detect the stator current polarity in the proposed Dual-AVF-HDN based inverter nonlinearity compensation,and therefore,the zero current clamping effect can be well compensated.Plus,the proposed AVF-HDN is distinguished by the discrimination capacity of positive-and negative-sequence harmonic contents and the easy-to-implement structure,which makes it ideal for implementation on low cost digital chips.To increase the output power and decrease the switching losses,a low pulsewidth modulation(PWM)to operating fundamental frequency ratio is necessary for middle-and high-power applications as well as high-speed PMSM drives.However,the dynamic decoupling control and the stability of the position observer suffer from the digital control delay caused by the low frequency ratio.To deal with this problem,a complex-vector based discrete-time domain zero-order hold(ZOH)equivalent model of IPMSM is developed explicitly incorporating with the stationary frame voltage latch effect of a PWM voltage source inverter.To improve the dynamic decoupling control and robustness,the direct design form of the discrete-time current regulator is proposed based on the pole-zero cancellation.Whereby,a full-order observer is designed and implemented using direct pole assignment in the discrete-time domain to guarantee the stability and dynamic performance with low frequency ratios.