Research On Active Disturbance Rejection Control Of Permanent Magnet Synchronous Motor System

Permanent magnet synchronous motor(PMSM),which owns the advantages of high torque density,high efficiency,and good speed regulation performance,is an important component for supporting the transformation of the manufacturing industry to high-end.In the meantime,its driving equipment is facing stringent performance requirements.The high-performance permanent magnet synchronous motor drive system should own excellent performances such as fast response,no overshoot,and zero steady-state error.In the face of diverse disturbances under complex operating conditions,in order to maintain the above performance,the system should also have strong robustness to disturbances.Aiming at the problem of disturbance suppression,the active disturbance rejection control(ADRC),which integrates the essence of classical control theory and modern control theory,has broad application prospects.Its unique "disturbance estimation & feedforward compensation" structure has outstanding advantages in suppressing disturbances,which has already attracted scholars’ attention from home and abroad.However,the unforeseen multi-type disturbances in complex operating conditions bring challenges to the popularization and application of ADRC in PMSM drive system.Relying on ADRC,this article focuses on the common problems faced by sensored control system and sensorless control system in suppressing load disturbance,parameter perturbation and harmonic disturbance,and the specific problem faced by sensorless control system in suppressing load disturbance.The research on the ADRC strategy of PMSM drive system is carried out from multiple dimensions to lay the foundation for meeting the application requirements of anti-multitype disturbance.The conventional linear active disturbance rejection controller(LADRC)has a simple structure and convenient parameter adjustment,but its error feedback function has insufficient convergence efficiency,which makes it difficult for the extended state observer(ESO)to accurately estimate the rapidly changing load disturbance,thus limiting the system’s anti-load disturbance performance.In this regard,a linear/nonlinear switching error feedback function is proposed in this article.The characteristic of "large error and large gain" in the nonlinear region forces the error to converge rapidly,and the characteristic of "small error and constant gain" in the linear region ensures low steady-state noise.In order to reduce the number of parameters to be tuned,the switching function is normalized.In order to ensure the smooth switching process,the principle of constant equivalent gain is proposed.Aiming at the problem of stability analysis caused by the existence of nonlinear function,the system is simplified through the equivalent gain transformation by using the characteristics of the switching function.Then,the stability condition is deduced.Owing to the adaptive change characteristics of the equivalent gain of the linear/non-linear switching feedback function,the improved ADRC exhibits smaller speed fluctuations than the conventional LADRC when encountering load disturbances,while the steady state performance is not affected.Problems such as motor temperature rise,magnetic circuit saturation,and load mechanical characteristics changes in complex operating environments will inevitably lead to changes in motor parameters.The additional disturbance caused by the perturbation of motor parameters will increase the observation burden of ESO,prolong the convergence time,and affect the dynamic performance of the system.In this regard,this article uses the disturbance estimation characteristics of ESO to design the parameter identification equation,thereby improving the parameter robustness of ADRC.Firstly,considering the perturbation of mechanical parameters,the perturbation model of speed loop ESO is reconstructed,and the identification equations of the damping viscosity coefficient and the moment of inertia are deduced.Secondly,considering the perturbation of electrical parameters,the perturbation model of the dq-axis current loop ESP is reconstructed,and the identification equations of the dq-axis inductances and the stator resistance are deduced.Aiming at the problem that the identification value cannot be converged due to the lack of rank of the electrical parameter identification equation set,the concept of dual time scale is introduced,and the estimation results are converged through time division multiplexing and cross update.Aiming at the problem that the inverter nonlinearity will affect the steadystate identification accuracy of electrical parameters,an error compensation strategy is proposed.The proposed identification method is integrated with ADRC,and the disturbance information estimated by ESO is used in the identification process to calculate the identification result,thus the system has low structural redundancy.Harmonic disturbances caused by the design defects of the motor or various non-ideal factors in the drive hardware will cause the motor speed and torque to fluctuate,and the steady-state operation performance will decrease.The structural characteristics of "disturbance estimation & feedforward compensation" determine that the disturbance rejection performance of ADRC depends on the disturbance estimation accuracy to a considerable extent.However,due to the limitation of bandwidth,it is difficult for conventional ESO to accurately estimate harmonic disturbances.In this regard,based on the internal model principle,this article introduces a complex coefficient filter(CCF)with the characteristics of frequency selection and phase sequence distinguishing to transform the integral element of the conventional ESO,and proposes a complex coefficient extended state observer(CCF-ESO)to improve the harmonic disturbances estimation accuracy.Also,the design criteria of the observer in frequency domain and in time domain are given,and the stability is verified.Owing to the peculiar harmonic disturbance estimation ability of CCF-ESO,the complex coefficient active disturbance rejection controller(CCF-ADRC)designed on the basis of it has the ability to suppress the specified harmonic disturbance.Furthermore,considering the complexity of the harmonic disturbances in the actual system,multiple center frequencies are designed for CCF,so that the CCF-ADRC controller is extended to a multi-center frequency CCF-ADRC controller to achieve comprehensive suppression of multi-frequency positive and negative phase sequence harmonic disturbances.In order to improve the reliability of the system,the high-performance PMSM drive should have the driving ability when the position sensor fails.However,limited by the dynamic performance of the speed/position estimator,there is an obvious weakness in the anti-load disturbance performance of the sensorless control system,which limits its competitiveness in mid-to-high-end applications.In this regard,this article integrates the idea of ADRC into the speed/position estimation link to improve its dynamic estimation performance.Firstly,on the basis of the back-EMF model,a second-order ESO with adaptive bandwidth is designed to estimate the back-EMF.Secondly,on the basis of the motion equation,a novel quadrature phase-locked loop(QPLL)based on a third-order ESO is designed to extract the speed/position information from the estimated back EMF.In order to further improve the speed estimation accuracy during load disturbance,the small-signal analysis of the motion equation revealing the quantitative relationship between speed fluctuation and torque differential during load disturbance is conducted.Based on the smallsignal analysis,a speed estimation compensation strategy is designed.The amount of compensation generated by this strategy is proportional to the torque derivative,so compensation only occurs in dynamic processes.Therefore,the anti-load disturbance performance of the sensorless control system is further enhanced,while the steady-state operation performance is not affected.