Research On Sensorless Control Strategy Of Permanent Magnet Motor Based On Pseudo-random Pulse Signal Injection

High frequency signal injection(HFSI)based sensorless vector control is broadly used in low speed control of the internal permanent magnet synchronous motor(IPMSM)which is of significant saliency.Under low speed driving condition,HFSI has good observation accuracy and dynamic response performance.However,HFSI method causes the vibration of the rotor and leads to an extra audio noise,which limits the practice of the injection method in a varieties of aspect.To study this subject,this paper studies the structure and parameters design of the position observer for pseudo-random signal input,analyses the noise suppression strategy based on pseudo-random pulse signal injection by power spectral density analyze.The proposed control system is verified by simulation and experiments.First,a sensorless vector control system based on the pulse injection signal in stationary reference frame is studied.According to the discrete high-frequency equivalent model of PMSM,the expression of the current differential term in the voltage equation is simplified.Based on the study of the impact on the current sampling results of different injected pulse signals,the selected pulse signal is less affected by the system delay,the system torque ripple is relatively small,and the rotor position demodulation process is easy to achieve.The estimated position information demodulation with quadrature signal input is studied to accomplish sensorless vector control of PMSM at low speed.According to analysis of the power spectral density(PSD),the motor noise suppression scheme based on pseudo-random pulse signal injection is studied.The power of the response current is concentrated at the injection frequency,which performs as peaks in PSD.This paper analyzes and designs the noise suppression schemes of the pseudo-random pulse signal injection method.By studying the relationship between PSD and the injection frequency,an improved method of random frequency injection to adjust the pulse width is proposed to suppress the discrete spectral peak components in PSD.This noise suppression method is theoretically adaptive to traditional position demodulation process.This paper compares the random frequency switching method,the random starting point method and the total random method to achieve the theoretical noise reduction goal of reducing the number of PSD peaks and attenuate the amplitude of peaks.For the problem that the updating frequency of orthogonal signal is relatively low,the position observation accuracy improvement method based on filtering algorithm is studied.The spatial Fourier series filter compensates the high frequency components of spatial position vectors.The quadratic interpolation filter uses quadratic function to approximate sinusoidal function to fit discrete orthogonal signals and improve the continuity of signals.The comb filter achieves the suppression of the specific frequency signal to filter out the harmonic components by certain order.The filtering methods is compared by the bandwidth and precision of the position observation results.The accuracy of rotor position estimation information is improved by designed filter.The sensorless control system of PMSM based on pseudo – random pulse signal injection is applied on both Simulink software and the STM32F103 based converter.The experimental verification is carried out on the 2.2kw IPMSM platform.Simulation and experimental results show the practicable of the above methods.The sensorless control strategy based on pseudo-random pulse signal injection method has good performance and good loading capacity.The accuracy of the estimated position signal is improved compared with other low-frequency signal injection methods.