Research On Several Adaptive Control Strategies For PMSM Drive System

Permanent magnet synchronous motors(PMSMs)are most competitive driving motors and widely used in high-performance servo applications owing to their high efficiency,superior power density,and large torque-to-inertia ratio.Meanwhile,the rich rare earth materials in China also provide a solid material basis for the development of PMSM.Nowadays,PMSM driving has become the trend in electric drive control systems with the advanced control strategies.PMSM driving has been highly valued at home and abroad,and is also an important means to carry out the implementation of motor energy conservation project.However,PMSMs are nonlinear multivariable dynamic systems,and the uncertainties induced by parameter variations,external disturbances,and unmodeled dynamics have become a common problem,which reduces the stability and robustness of the controlled system.To suppress these uncertainties and achieve the desired servo control performance,advanced strategies must be developed.In view of these "uncertainties",this thesis studies a class of common adaptive control method-Model Reference Adaptive Control(MRAC).Based on the analysis of the existing problems and challenges of MRAC,a class of nonparametric model adaptive control method is introduced,and a series of new methods of adaptive control and compound control are presented.The main work of this thesis mainly includes the following aspects:(1)The primary task of the parametric model control method is to establish the mathematical model of the controlled plant.However,the traditional linear model of PMSM,which ignoring the effect of iron loss and magnetic saturation,has not been able to accurately describe the complex behavior of the PMSM.The design of the control system based on this imprecise mathematical model will lead to poor dynamic performance or even the instability of the closed loop system.In order to solve this problem,this thesis proposed a new method for the establishment of PMSM precise mathematical model,which considers all loss and magnetic circuit saturation.The method connected an equivalent iron loss resistance across the air gap voltage to represent the iron loss by considering that the total iron loss may be caused by magnetizing flux.Furthermore,the effect of mechanical loss and stray loss has been taken into account to modify the model.Based on the fact that the motor parameters do not keep constant with the saturation degree of magnetic circuit,the nonlinear relationship between motor parameters and the axis excitation current was obtained by using finite element analysis method,so as to achieve the construction of accurate mathematical model.The simulation analysis showed that the proposed method is effective to describe the dynamic behavior of the motor,and has better dynamic response performance comparing with the traditional mathematical models.Based on this precise model,this thesis designed a vector control system of PMSM.Firstly,in the realization of SVPWM technology,a simple algorithm for dividing sector and calculating the action time of basic vector was proposed.The basic vector of two phase coordinate system was decomposed into three phase coordinate system(perpendicular to the coordinate system).The sector was determined by linear combination of vector symbols in the coordinate system,and the action time of the basic vector was the component of vector in the coordinate system,which can simplify the computation effectively.Secondly,in the realization of speed sensorless technology,the estimation method of velocity and rotor position was designed based on the inverse tangent function with the motor exact model.According to the voltage equation and the flux equation in the two phase coordinate system,the back EMF,which contains the information of velocity and rotor position,was calculated to realize the speed sensorless control.Finally,the simulation of vector control system was carried out,and the dynamic response performance of the system and the speed sensorless technology were analyzed and evaluated.(2)PMSMs are high-order nonlinear uncertain systems,and the uncertainties induced by parameter variations,external disturbances,and unmodeled dynamics have become a common problem,which causes the conventional feedback control unable to meet the complex production process and high precision control.For such problems,MRAC is an effective solution and also the point of the beginning research.Firstly,a double closed loop PMSM system was designed based on the adjustable gain Lyapunov-MRAC method,which derived the adaptive control law using the output generalized error and the input variables.Meanwhile,the control performance of the system and the selection of adjustable gain factor are analyzed.Secondly,the state feedback Lyapunov-MRAC system was designed to overcome the dependence on state space description and known parameters matrix of the adjustable gain Lyapunov-MRAC method.Furthermore,the similarities between the state feedback Lyapunov-MRAC and signal synthesis Narendra control were discussed.Simulation results and performance analysis showed that the Lyapunov-MRAC systems offered low computational burden with only partial known structure or model,easy implementation adaptive control law with the input,output,state,and error of the controlled plant or reference model,higher dynamic performance,and strong stability.Thirdly,the continuous system MRAC method was extended and a PMSM discrete time MRAC control system was designed.This design was based on Popov's super stability theory,and derived the adaptive control law by solving Popov integral inequality.The discrete time Popov-MRAC overcome the problem of Lyapunov function selection,and reduced the function influences on system performance and complexity,and showed strong robustness to external disturbances and parameter changes.Finally,the MRAC method was used to achieve the PMSM speed sensorless technology.Unlike adaptive control,parameter identification of MRAC took PMSM as a reference model,and used generalized error to update the adjustable system including velocity information.The asymptotic convergence of speed estimation was guaranteed by Popov super stability theory to achieve the observation of velocity and rotor position.The simulation analysis showed that the velocity and rotor position observation method based on Popov-MRAC presented lower estimation precision and larger error at low speed,which caused by the limitation of mathematical model,parameter change,current detection,and the uncertainties.Contrarily,there was higher estimation precision and smaller error at middle or high speed,which can meet the needs of the motor control performance.(3)The traditional MRAC method designs the closed loop system based on the system model hypothesis and the deterministic equivalence principle,and suffers from the uncertainties of mathematical models,external disturbances,and unmodeled dynamic.For such problems,the "model free" property of nonparametric model control method is an effective solution.Based on this consideration,this thesis introduced the model free adaptive control(MFAC)into the PMSM speed control,and proposed a higher-order model free adaptive control(HMFAC)method to improve the tracking performance of PMSM.Firstly,the method established an equivalent differential expression linear model for the discrete-time nonlinear description of PMSM and designed a higher-order controller using a model-free adaptive technique.The controller involved more input and output(I/O)data in a fixed length sliding time window and improved the freedom and flexibility of the controller.This design depended on the unique bounded parameter referred as pseudo-partial-derivative(PPD)that was a slowly time-varying parameter deriving online from the I/O data.Secondly,the design guaranteed the stability of the bounded input and output and ensured tracking error monotonic convergence under a restricted set of parameters.Simulation results showed that the HMFAC method was essentially the integral behavior of tracking error,and similar to PID in the structure of control law.However,the HMFAC method eliminated the modeling process of the parameterized model,and designed the data-driven controller without any model structure or parameter information.This method can effectively reduce the influence of external disturbances and unmodeled dynamics,is easy to implement and has strong robustness.This method is especially useful for nonlinear systems with vague dynamics.To make further efforts on the convergence accuracy of HMFAC,a 2-D type high-order model free adaptive iterative learning control(HMFAILC)method is proposed in this thesis.The method analyzed the essential similarity relation between HMFAC and ILC methods,and the ability to deal with unknown uncertainty.Based on the repetitive operation characteristics in the finite time interval of PMSM,the method improved the system tracking accuracy by using the iterative learning law along the moving direction of the iterative axis,and the stability and disturbance rejection performance with the motion of the time axis under the control input.The HMFAILC method was a hybrid data driving technique of the HMFAC and ILC,and not only preserves the convergence and robustness of HMFAC along the moving direction of the time axis,but also improves the convergence precision of the tracking error in the finite time interval.The simulation results showed that the HMFAILC method not only solved the defect without learning ability of HMFAC,but also effectively overcome the dependence on the system prior knowledge and the blind selection of learning gain.Moreover,the HMFAILC method,combining advantages of both,was far superior to the simple HMFAC or ILC control methods in convergence and robustness.(4)Combining advantages of HMFAC and MRAC,two hybrid control methods have been put forward,namely series HMFAC-MRAC and parallel HMFAC-MRAC.The MRAC method depends on the parameterized model structure information of the controlled plant,and the uncertainties induced by external disturbances and unmodeled dynamics have become a common problem,which reduces the stability and robustness of the controlled system.While the HMFAC method depends on the I/O data of the controlled plant,and the robustness is also a problem in the case of data measurement noise and data loss(or incomplete).Therefore,it is a reasonable choice to study the hybrid adaptive control method with complementary advantages.Based on the idea of modular design,series HMFAC-MRAC hybrid method and parallel HMFAC-MRAC hybrid method have been put forward.The main idea of proposed serial HMFAC-MRAC method was to apply HMFAC to design controller,established system equivalent linear data model and deduced higher-order adaptive control algorithm,and apply MRAC to identify the controller's adjustable parameter PPD based on the super stability theory.This method was essentially a data-driven MRAC,and had no need the support of the model structure or parameter information.Comparing with the traditional MRAC,the uncertainties induced by parameter changes,the external disturbances and the unmodeled dynamics were effectively suppressed in the HMFAC control loop of the serial HMFAC-MRAC,and strong robustness was guaranteed even if the data were lost,which verified by simulation analysis.The main idea of the parallel HMFAC-MRAC method was to add HMFAC as an outer compensation loop to the original MRAC control system,and estimated the uncertainties to modify the control input for compensating the system's unmodeled dynamics and parameters estimation error.When the HMFAC loop was opened and the system was degraded to the original MRAC system.Therefore,the parallel HMFAC-MRAC method was essentially a MRAC method with parameter compensation,and belonged to the parameter model control method.The comparison simulations among the hybrid method,MRAC and PID showed that the proposed hybrid method had similar characteristics in convergence and robustness,and had the good ability to deal with uncertainties.Based on the modularization idea,the control performance of the hybrid control method is far superior to the simple HMFAC or ILC control method in convergence and robustness,which is especially useful for nonlinear motor systems with uncertainties,and provides a valuable reference for theoretical or practical research.