Study On The Assisted Torque Control Algorithm Of Electric Power Steering System Using Induction Motor

Electric power steering system is a new type power steering system developed rapidlyin recent years. This new system can solve the contradiction between handiness andsensitivity, and has many advantages such as energy saving, environmental protection andeasy assemblity. Recently, electric power steering system is gradually replacing the hydraulicand electro-hydraulic power steering system, and becoming automotive technology researchfocus all over the world.This paper treated the electric power steering system using induction motor as theresearch object, studying on the control algorithm of induction motor under the EPSapplication conditions.EPS system as a servo system, the first requirement is fast andaccurate output torque response of motor. Due to the special nature of the structure ofinduction motor, it can not like the DC brush motors and permanent magnet synchronouspermanent magnet motors that you can more easily achieve high-performance torque control,in order to ensure control performance of induction motor EPS system, this paper focus onthe study of how to improve the performance of the induction motor torque control above. Inthis paper, off-line parameter identification, parameter sensitivity of torque control accuracyunder rotor field oriented control conditions, current control algorithm, EPS hardware in theloop test rig, hardware and software development are discussed in detail.1) Off-line parameter identification for induction motorIn the induction motor vector control algorithm, the parameters are needed by the motorrotor flux position observer. Through the analysis of conventional off-line induction motorparameter identification method, explained the reasons for the lower accuracy of theidentification results. In order to improve the accuracy of parameter identification,parameter identification can improve from two parts: Keep sample data clean and usingappropriate method, the internal resistance of the power devices is equivalent to the statorresistance of the motor, through the voltage correction and work conditions selecting toeliminate the voltage distortion and error part, getting a reasonable sample data; Treatedtwo-phase reconstruction voltage and speed as input, two-phase current and electromagnetictorque as output. Establish a parameter identification model in Simulink/Matlab. Design thefitness function from the current match and torque match, using a genetic algorithm toidentify the equivalent electrical parameters of induction motor. In order to verify theaccuracy of the identification results, design parameter validation experiment in stationarycoordinates and rotor field oriented coordinates, the validation results showed that: motorparameter identification was able to accurately describe the steady-state and transientresponse of inverter-fed induction motor condition; rotor flux orientation using identificated parameters can achieve precise torque control.2) Analysis of parameter sensitivity of torque control accuracy under rotor field orientedcontrol conditions.In the induction motor rotor field oriented control, torque control accuracy is heavilydependent on the accuracy of the rotor field observations. Although off-line parameteridentification algorithm can obtain accurate parameters, but motor parameters will changewith temperature and operating conditions change. This paper presents an analysis of torquecontrol accuracy error caused by parameter mismatch. The impact of motor parametersmismatch in torque control when using different type observer such as stationary coordinatescurrent-speed model, voltage-current models, synchronous coordinates current-speed modeland full-order observer was got. Taking into account the combined results and the continualmotor speed fluctuations in EPS, the current-speed observer model was selected because itstorque control accuracy has no relation with motor speed. In order to eliminate the controlerror, on-line identification algorithms called model reference adaptive system for rotorresistance was introduced, the parameters sensitivity of the algorithm were analyzed.3) Current regulator in rotor field oriented coordinateWhen rotor flux position was estimated correctly, the response speed of torque dependson the performance of the current controller. By the theory of complex vector, two typescontrol method under synchronous rotating coordinate: traditional PI control and PI+feedback decoupling control were analyzed. Conventional PI current regulator’sperformance will weaken with synchronous speed performance increases, while theconventional PI+feedback decoupling current regulator’s performance depends on the PIparameters designing using motor parameters. In this paper, M-axis current is set to aconstant value control, T-axis current is follow-up control, in order to improve the currentregulator performance, taking into account the M/T axis current difference, the M-axiscurrent is regulated by a fuzzy PI+feedback decoupling control method, focusing on theelimination of interference caused by T-axis current changes, ensuring the accuracy of theM-axis current control. The T-axis current is regulated by adaptive feedforward+feedbackdecoupling control method, by reducing the proportion of feedforward amount in the totalamount to improve the response speed of the system; In order to improve feedforwardperformance when motor parameters change, feedforward parameters online identificationalgorithm based on T-axis feedback control was proposed, robustness of feedforward wasimproved.4) Optimized design of assisted characteristic curveAssisted characteristics determine the distribution ratio of the steering wheel torqueand power steering motor output torque, the paper through three type assist characteristiccurve analysis. The four-part assist characteristic curve with parameters restricted wasdesigned, treat the ideal torque characteristics as the design target, using genetic algorithm tooptimize the design parameters, so that the actual steering torque characteristics close to theideal value.5) Hardware system and experimental environment of EPS system used induction motor.Motor test rig was established for motor control algorithm design and debug. In order tosimulate the automotive environment, the EPS hardware in loop bench was build, CARSimsoftware was used to solve road surface reaction force in real-time and EPS system is loadedby motor. This chapter describes the EPS controller hardware architecture and the softwareprocess. And at last, the EPS system control strategy was validated in EPS hardware in looptest bench.Through analyse and experiment validation for induction motor used in EPS，the mainconclusions are as follows:(1) Conventional induction motor parameter identification methods have someproblems, and greatly influenced by error sources. This paper analyze the error terms inparameter identification, through reasonable compensation and avoidance, error term wereeliminated, so more accurate motor parameters can be obtained.(2) Under the conditions of the rotor field oriented control, When different fluxobserver model was used, the status of torque control accuracy affected by error of the motorparameters、motor speed and torque command is not different.(3) When the synchronous speed increases, the current regulation performance ofconventional PID algorithm under synchronous rotation coordinates became weakened.Conventional PID+feedback decoupling control’s performance is better than conventionalPID, but its performance is dependent on the PID parameter tuning using motor parameters.(4) Feedforward algorithm can effectively improve the response speed of T-axiscurrent, but its performance is dependent on the accuracy of the feedforward parameters,adding parameters on-line identification can improve the performance of T-axis currentcontrol robustness.This paper has been innovative in the following areas:(1) A analysis method of induction motor torque control accuracy under the conditionsof the rotor field oriented control was proposed. The parameter sensitivity of torque controlaccuracy used different flux observers was analyzed. Quantized affect caused by resistancemismatch、inductance mismatch、motor speed、torque command quantify is obtained.(2) Consideration of the differences in control goal of M-axis and T-axis, proposedfuzzy PID+feedback decoupling method for the M-axis current regulation, adaptivefeedforward+feedback decoupling method for the T-axis current regulation. Parametersensitivity and current regulation performance of the proposed method was validated throughexperiment.（3）Proposed assisted characteristic curve optimization based on ideal steering torquecharacteristic. The four-part assist characteristic curve with parameters restricted wasdesigned, treat the ideal torque characteristics as the design target, using genetic algorithm tooptimize the design parameters, so that the actual steering torque characteristics close to theideal value.