Research On Current Sliding Mode Control Of Permanent Magnet Synchronous Machine For Hybrid Electric Vehicle

The motor drive system is one of the most critical components for hybridelectric vehicles (HEV). The permanent magnet synchronous motor (PMSM)drive system for HEV should meet the specific requirements such as high torqueoutput, wide speed range, high efficiency and good reliability in the whole speedrange. PMSM is a multivariable, nonlinear and strong-coupled system which isextremely sensitive to parameters and disturbances, so traditional linear controlmethods can not accurately describe its static and dynamic processes andguarantee its high operation performances in a wide speed range. Sliding modevariable structure control (SMVSC) is a special non-linear control method withadvantages such as fast response, strong robustness and simple realization.However, the sliding mode control will bring in inevitable system chattering,which makes the present-day research on the sliding mode control applied to theactual motor drive system focused on how to alleviate the chattering. In order toimprove the system robustness of PMSM driver for HEV against parametervariations and external disturbances, this dissertation researches on theapplication of current sliding mode control to PMSM maximum torque perampere (MTPA) control system, and lays emphasis on how to alleviate thesystem chattering and improve the system characteritics of anti-disturbance byimproved reaching law, sliding mode load observer and gain-scheduled basedsliding mode control.The reaching law method can improve the dynamic quality of sliding motion.This dissertation analyzes and compares the influence of several typical reachinglaws on system characteristics. In view of disadvantages of the regular reaching laws such as higher chattering near the sliding band and non-convergence to theorigin, an improved exponent reaching law method with time-varying switchinggain is proposed. In this method, a time-varying factor is introduced to theswitching gain which is modified to remain smaller near the sliding band and toapproach zero with the system state in the stage of sliding motion, meanwhile,the integral sliding surface with integral separation is adopted and the boundarylayer method is applied. Thus, the designed reaching law method can improve thesystem dynamic performances and effectively alleviate the system chattering.In the sliding mode control system, the minimum switching gains onlyincreases with the amplitude of load disturbances can it meet the sliding modeexistence and accessibility condition and thus effectively resist the disturbances.However, the increasing of switching gain will intensify the system chattering.To solve the problem above, this dissertation proposes a load observation andcompensation method with sliding mode observer. In the load disturbanceobserver, the integral sliding surface with respect to the speed observation erroris selected, and the observer switching gain is adaptively adjusted according tothe load disturbance observation error, which not only ensures the observerrobustness, but also weakens the load observation chattering. Meanwhile, theload disturbance observed value is applied as feed-forward compensation to theoutput terminals of the current sliding mode controller, so much smallerswitching gain is used to meet the demand of anti-disturbance. Thus, this methodimproves the current tracking responses under load disturbances and significantlyweakens the system chattering.In the real system, the structure and parameters of controller are required tochange with the operating conditions and external environment in order to obtainsatisfactory control performances in the total operating range. Gain-scheduledcontrol is an effective method to solve the control problems for nonlinearsystems. This dissertation presents an adaptive and gain-scheduled hybridmethod to set the parameters of the current sliding mode controller. In thismethod, the boundaries of the switching gain are adjusted with the adaptivemethod, and taking the integral sliding surfaces with regard to d-axis current andq-axis current as scheduling variables, the controller coefficients of the switchinggains, the sliding surfaces and the boundary layers are set with gain-scheduled method. Thus, this method can overcome the disadvantages of regular gain-scheduled controller which requires the determination of typical operation pointsand the optimization of a large number of parameters, which not only ensures thesystem robustness, but also improves the system control performances andefficiency.In this dissertation, simulation and experimental studies on PMSM currentsliding mode control system have been carried out. Simulation and experimentalresults show that the system has strong robustness against load disturbances, andobtains good steady characteristics and dynamic tracking performances. Thedesigned methods and controllers are simple in structure and practical inengineering, so they are applicable for motor drive system for HEV.