Sensorless Control Of IPMSM With Low Audible Noise And Full Speed Range Operation

The sensorless technology of internal permanent magnet synchronous motor(IPMSM)can effectively reduce the system cost and improve the system reliability.It is of great significance to study the high performance IPMSM sensorless control system with low audible noise,high control accuracy and wide speed range.At present,the IPMSM sensorless control technology still has the following key technical difficulties in low audible noise and full speed range operation:(1)High frequency injection method uses the salient pole characteristics of the IPMSM to accurately estimate the rotor position at zero and low speeds.However,the injected high frequency voltage will cause harsh audible noise.(2)Domestic famous inverter brands usually use flux observer in medium and high speed to estimate the rotor position,which is easy to implement and has good versatility.However,flux observer has poor performance in suppressing DC bias and high order harmonics.(3)The speed and position observation method of medium and high speed closed loop based on extended state observer(ESO)does not depend on the precise mathematical model of the IPMSM.However,the ESO has low pass filtering characteristics,so it can not accurately estimate the rapidly changing back electromotive force.(4)The low speed estimation method,medium and high speed estimation method are interact with each other and difficult to switch smoothly.Breaking through the above core technical difficulties has important theoretical and practical significance for broadening the application of the IPMSM sensorless control.The fixed frequency high frequency voltage injection method injects the fixed frequency high frequency voltage,and the power spectral density(PSD)of the excited high frequency current produces a high peak at the injected signal frequency,resulting in the harsh audible noise.Aiming at the problem of the audible noise produced by the traditional high frequency voltage injection method,based on the Markov chain random process,the IPMSM sensorless control method based on the Markov chain pseudo random high frequency square wave voltage injection under fixed switching frequency is proposed,and the audible noise produced by the high frequency voltage injection method is reduced by expanding the PSD of the high frequency current.Firstly,the corresponding high frequency current demodulation strategy is designed to ensure the rotor position observation accuracy of the pseudo random high frequency voltage injection method based on the Markov chain under the premise of reducing audible noise.Then,the probability model and PSD of the four states Markov chain scheduling high frequency voltage injection method are studied.The noise reduction mechanism of the pseudo random high frequency square wave voltage injection method based on the Markov chain is revealed theoretically.Finally,the effectiveness of the pseudo random high frequency square wave voltage injection based on the Markov chain to reduce audible noise is verified by the experiments.The discrete PSD of the high frequency current excited by the pseudo random high frequency square wave voltage injection method will appear at the least common multiple of the injected signal frequency,which is not conducive to further reducing the audible noise of the high frequency voltage injection method,and the sensorless control of the high frequency voltage injection method has not been extended to the random switching frequency operation condition In order to further reduce the audible noise of the high frequency voltage injection method and expand the application scope of the high frequency signal injection method,an IPMSM sensorless control method based on the random high frequency square wave voltage injection with random switching frequency is proposed.Firstly,a high frequency current demodulation strategy without high frequency voltage demodulation is designed to suppress the position observation error caused by the phase mismatch between high frequency voltage and high-frequency current.Then,the PSD of the random high frequency square wave voltage injection method with random switching frequency is analyzed,and the noise reduction mechanism of the random high frequency square wave voltage injection method with random switching frequency is revealed theoretically.Next,the rotor position observation error caused by the inductance cross coupling and A/D converter quantization error in high frequency voltage injection method is analyzed.Finally,the effectiveness of the random high frequency square wave voltage injection method based on random switching frequency to reduce audible noise is verified by experiments.Aiming at the problem of rotor flux estimation inaccuracy of the traditional flux observers caused by the non-ideal factors such as motor parameter mismatch,motor magnetic field spatial harmonics,detection error,inverter nonlinearity,sampling noise and so on,a fifth order flux observer(FOFO)is proposed based on the mathematical model of the IPMSM considering current harmonics and back electromotive force harmonics.Firstly,the principle of estimating the speed and position of the FOFO is analyzed,and the parameters of the FOFO are designed Then,the effectiveness of the FOFO against DC bias and high order harmonics is analyzed and verified.Next,the discrete FOFO model is constructed.Finally,the effectiveness of the FOFO to suppress DC bias and high order harmonics is verified by the experiments,so as to suppress the position error ripple and improve the rotor position observation accuracy.The traditional ESO has the characteristics of low pass filter,so it can not estimate the rapidly changing back electromotive force.To solve this problem,an IPMSM sensorless control method based on the generalized integral ESO(ESO-GI)is proposed.Firstly,the ESO-GI is constructed.A fast-varying sinusoidal disturbance estimator is added to the unknown disturbance estimation loop of the ESO to estimate the fast-varying back electromotive force The frequency of the fast-varying sinusoidal disturbance estimator changes adaptively with the operating frequency of the motor.The ESO-GI disturbance loop is composed of pure integral and fast changing sinusoidal interference estimator.Then,the performance of the ESO-GI for back EMF observation is analyzed.The results show that the ESO-GI can accurately estimate the rapidly changing back electromotive force at a relatively low bandwidth,thus solving the tradeoff between bandwidth and high frequency noise filtering.Finally,the effectiveness of the ESO-GI to observe the fast changing back EMF is verified by experiments.In the transition region,low speed estimation method and medium high speed estimation method influence each other and are difficult to smooth handoff.Firstly,the position observation error estimated by the high frequency voltage injection method is analyzed when considering stator resistance and extended electromotive force,which provides a theoretical basis for determining the transition interval of composite sensorless full speed control.Then,the influence of the injected high frequency voltage signal on the observation performance of the FOFO is analyzed in the transition zone.The FOFO has a strong ability to suppress high frequency current and does not need to add any filter at the input of the FOFO.The rotor position observation error of the ESO-GI caused by high frequency current in the transition region is analyzed,and the ESO-GI rotor position observation method based on the adaptive complex coefficient filter in the transition region is proposed to improve the position observation accuracy of the ESO-GI.Finally,the unified phase locked loop composite sensorless method is studied to realize smooth transition in transition region.