WO2023138258A1 - 一种主动转向和横摆力矩自学习协同控制方法 - Google Patents

一种主动转向和横摆力矩自学习协同控制方法 Download PDF

Info

Publication number
WO2023138258A1
WO2023138258A1 PCT/CN2022/138265 CN2022138265W WO2023138258A1 WO 2023138258 A1 WO2023138258 A1 WO 2023138258A1 CN 2022138265 W CN2022138265 W CN 2022138265W WO 2023138258 A1 WO2023138258 A1 WO 2023138258A1
Authority
WO
WIPO (PCT)
Prior art keywords
formula
vehicle
equation
mass
ecu
Prior art date
Application number
PCT/CN2022/138265
Other languages
English (en)
French (fr)
Inventor
付志军
郭耀华
赵登峰
丁金全
刘朝辉
何文斌
杨文超
姚雷
周放
王辉
明五一
Original Assignee
郑州轻工业大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 郑州轻工业大学 filed Critical 郑州轻工业大学
Publication of WO2023138258A1 publication Critical patent/WO2023138258A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B62—LAND VEHICLES FOR TRAVELLING OTHERWISE THAN ON RAILS
    • B62D—MOTOR VEHICLES; TRAILERS
    • B62D7/00—Steering linkage; Stub axles or their mountings
    • B62D7/06—Steering linkage; Stub axles or their mountings for individually-pivoted wheels, e.g. on king-pins
    • B62D7/14—Steering linkage; Stub axles or their mountings for individually-pivoted wheels, e.g. on king-pins the pivotal axes being situated in more than one plane transverse to the longitudinal centre line of the vehicle, e.g. all-wheel steering
    • B62D7/15—Steering linkage; Stub axles or their mountings for individually-pivoted wheels, e.g. on king-pins the pivotal axes being situated in more than one plane transverse to the longitudinal centre line of the vehicle, e.g. all-wheel steering characterised by means varying the ratio between the steering angles of the steered wheels
    • B62D7/159—Steering linkage; Stub axles or their mountings for individually-pivoted wheels, e.g. on king-pins the pivotal axes being situated in more than one plane transverse to the longitudinal centre line of the vehicle, e.g. all-wheel steering characterised by means varying the ratio between the steering angles of the steered wheels characterised by computing methods or stabilisation processes or systems, e.g. responding to yaw rate, lateral wind, load, road condition
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W10/00—Conjoint control of vehicle sub-units of different type or different function
    • B60W10/18—Conjoint control of vehicle sub-units of different type or different function including control of braking systems
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W10/00—Conjoint control of vehicle sub-units of different type or different function
    • B60W10/20—Conjoint control of vehicle sub-units of different type or different function including control of steering systems
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units, or advanced driver assistance systems for ensuring comfort, stability and safety or drive control systems for propelling or retarding the vehicle
    • B60W30/02—Control of vehicle driving stability
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/72—Electric energy management in electromobility

Definitions

  • the invention relates to the fields of automatic control, information technology and advanced manufacturing, and specifically relates to the problem of cooperative control of active steering and yaw moment with an unknown model.
  • Advanced active chassis control systems play a key role in the pursuit of better handling and stability, and safer vehicles.
  • For the actual chassis system it should have higher intelligence to adapt to various road conditions.
  • Active steering control systems can apply additional steering angles to the driver's steering commands and directly affect the lateral dynamic behavior of the vehicle by modulating the lateral forces on the tires.
  • the active steering control system cannot generate sufficient tire lateral force, so its performance in the linear steering region is limited.
  • the yaw moment control system is very effective for vehicle stability in linear and non-linear maneuvering regions. It generates an appropriate amount of corrective yaw moment through differential braking on the left and right sides of the vehicle to maintain vehicle stability.
  • yaw moment control systems are only suitable for limited maneuvering.
  • a separate active steering control system and yaw moment control system cannot achieve comprehensive and optimal safety performance. Therefore, it is necessary to integrate the active steering control system and the yaw moment control system, and realize the tracking of the desired yaw rate and sideslip angle through the cooperative control method between the two, so as to obtain satisfactory vehicle stability performance under different driving actions.
  • the model uncertainty caused by parameters such as tire cornering stiffness, longitudinal vehicle speed, and non-linear characteristics of tires brings great difficulties to the design of the controller.
  • the existing common control method is based on the linear quadratic regulation method (LQR), and its main disadvantage is that the suspension system model must be known accurately in advance to find the optimal control law.
  • the feedback control gains are obtained by solving the Riccati equation off-line. Once the feedback gains of the controller are obtained, they cannot be changed with different road surface external inputs and related vehicle uncertain parameters. Therefore, a more effective control strategy is needed to adaptively deal with the intelligent chassis control problem with time-varying parameters under different driving conditions in real time.
  • the purpose of the present invention is to provide an active steering and yaw moment self-learning cooperative control method, which does not need to know the suspension system model in advance, to deal with the problem that the uncertainty of the model brings great difficulties to the design of the controller, and only needs sensor data to perform control, and is used for real-time and adaptive processing of intelligent chassis control problems in which parameters change with time under different driving conditions.
  • a kind of active steering and yaw moment self-learning cooperative control method of the present invention is used in a motor vehicle, the motor vehicle has a vehicle-mounted ECU, and the vehicle-mounted ECU is connected with a vehicle speed sensor for obtaining vehicle speed v x information, a rotation angle sensor for obtaining driver steering angle information ⁇ f , a yaw rate sensor for obtaining yaw rate ⁇ information, and a mass center slip angle sensor for obtaining center of mass side slip angle ⁇ information;
  • each sensor is conventional technologies, such as the angle sensor being arranged at the steering shaft of the steering wheel of the motor vehicle, and the yaw rate sensor and the center of mass side slip angle sensor being arranged at the center of mass of the motor vehicle;
  • the vehicle-mounted ECU stores a constant K 1 representing the identification gain, a constant ⁇ representing the online learning gain, a time constant ⁇ r , a time constant ⁇ ⁇ , a vehicle mass m, a distance l f from the center of mass to the front axle, a distance l r from the center of mass to the rear axle, and a constant K 2 representing the control gain;
  • the first step is to construct the basic equations stored in the vehicle ECU, including the identifier, control target reference model and controller;
  • Equation 1 The basic equations include Equation 1 to Equation 15;
  • the first step is to construct formula 1 and formula 2;
  • formula 1 is a parameterized neural network:
  • ⁇ is the sideslip angle of the center of mass, and its unit is radian rad
  • ⁇ is the yaw rate
  • its unit is radian per second rad/s
  • a ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 matrix in the real field R
  • ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 matrix in the real field R
  • ⁇ (x) is the sigmoidal activation function in the neural network
  • ⁇ (x) [ ⁇ (x 1 ), ⁇ (x 2 )] T
  • ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 matrix in the real field R
  • u [ ⁇ c , M c ] T
  • ⁇ R 2 is a 2-dimensional column vector in the real field R
  • ⁇ C is the active steering angle
  • its unit is radian rad
  • M c is the yaw moment
  • its unit is N m
  • the first step and the second sub-step are to design the identifier and online self-learning rate, and construct formula 3 and formula 4;
  • Equation 3 is the identifier used to achieve system identification:
  • Formula 4 is the online self-learning rate:
  • ⁇ is a constant obtained by trial and error and stored in the vehicle ECU, ⁇ >0 and represents the online learning gain;
  • the third sub-step of the first step is to construct the control target reference model expressed by formula 5, and generate the target tracking signal;
  • x r is the reference state
  • x r [ ⁇ r ⁇ r ] T
  • ⁇ r is the sideslip angle of the target center of mass
  • ⁇ r is the target yaw rate
  • ⁇ r and ⁇ r are the calculation results of formula 5 and serve as target tracking signals
  • I Z is the moment of inertia of the vehicle around the vertical direction, the unit is kg ⁇ m 2 , determined by the vehicle manufacturer and stored in the vehicle ECU;
  • the first step and the fourth sub-step are to construct the controller expressed by formula 6 based on the approximate dynamic programming theory, so as to realize the cooperative control of active steering and yaw moment self-learning;
  • Equation 6 The u 1 in Equation 6 is used to ensure that the steady-state error of the control converges to zero, and its expression is Equation 7:
  • K 2 is a constant indicating the control gain obtained by trial and error and stored in the on-board ECU, K 2 >0;
  • the u 2 in formula 6 is used to ensure the optimal control performance, and it is designed according to the approximate dynamic programming theory, specifically obtained through formula 8 to formula 15;
  • Equation 8 is the evaluation function V as follows:
  • Q ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 diagonal matrix in the real number field R, which represents the weight of the tracking error e 2 in the optimal control evaluation function, which is obtained by trial and error and stored in the vehicle ECU;
  • P ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 diagonal matrix in the real number field R, representing the weight of u 2 in the optimal control evaluation function, which is obtained by trial and error and stored in the vehicle ECU;
  • Equation 9 is the Hamiltonian function:
  • W V is the ideal weight vector
  • ⁇ (e 2 ) is the sigmoidal activation function in the neural network
  • Equation 11
  • Equation 12 Equation 12
  • Formula 12 is:
  • the second step is to calculate the values of active steering angle ⁇ C and yaw moment Mc online according to the following sub-steps when the vehicle is running, and control the running state of the vehicle according to the calculation results of ⁇ C and Mc .
  • the first sub-step of the second step is that the ECU collects original real-time parameter values, including the steering angle ⁇ f value measured by the steering angle sensor when the driver manipulates the steering wheel, the vehicle speed v x value from the vehicle speed sensor, the mass center side slip angle ⁇ value from the mass center slip angle sensor, and the yaw rate ⁇ value from the yaw rate sensor.
  • the second step The second sub-step is the calculation step of the identifier and the reference model of the control target;
  • the on-board ECU provides the center of mass sideslip angle ⁇ value and the yaw rate ⁇ value to the identifier described in formula 3, and calculates as the basis for calculating the value of u1 in Equation 6;
  • the on-board ECU provides the value of the steering angle ⁇ f and the value of the vehicle speed v x to the control target reference model described in formula 5, and calculates the value of the target center of mass side slip angle ⁇ r and the target yaw rate ⁇ r , which are used as the calculation basis for the parameters e 2 required by u 1 and u 2 in formula 6.
  • the second step and the third sub-step is that the on-board ECU converts ⁇ r , ⁇ r and Provided to the controller described in Formula 6, the active steering angle ⁇ C and the yaw moment M c are calculated through Formula 6 to Formula 15;
  • the vehicle-mounted ECU controls the steering action of the steering wheel of the motor vehicle according to the active steering angle ⁇ C calculated in real time, and controls the braking action of the brake of the motor vehicle according to the yaw moment M c calculated in real time;
  • the present invention gets rid of the restriction that the optimal control law can only be found by accurately knowing the suspension system model in advance, and can self-adaptively deal with the intelligent chassis control problem that the parameters change with time under different driving conditions in real time.
  • the vehicle-mounted ECU can realize the self-learning cooperative control of active steering and yaw moment, calculate the ideal active steering angle ⁇ C and yaw moment M c , adjust and control the braking and steering, correct the driver’s steering operation, overcome improper driving, and make the vehicle tend to neutral steering when turning; after the vehicle-mounted ECU controls the steering action of the motor vehicle according to the active steering angle ⁇ C and yaw moment Mc , the actual center of mass sideslip angle and yaw rate of the motor vehicle feedback approach the formula 5
  • the calculated ⁇ r value and ⁇ r value can avoid accidents such as instability caused by improper driving.
  • Fig. 1 is a schematic diagram of the principle of the present invention
  • Fig. 2 is the actual steering angle-time graph when the vehicle ECU controls the vehicle to change lanes at a vehicle speed of 28 m/s according to the active steering angle ⁇ C and the yaw moment Mc ;
  • Fig. 3 is a comparison diagram of the reference value of the center of mass slip angle (ie, the ⁇ r value calculated by formula 5) and the real side value of the control result when the vehicle performs the lane change action in Fig. 2;
  • Fig. 4 is a comparison chart of the reference value of the yaw rate (that is, the value of ⁇ r calculated by formula 5) and the actual value of the control result when the vehicle performs the lane change in Fig. 2 .
  • a kind of active steering and yaw moment self-learning cooperative control method of the present invention is used in motor vehicle, and motor vehicle has vehicle-mounted ECU, and vehicle-mounted ECU is connected with the vehicle speed sensor that is used to obtain vehicle speed v x information, is used to obtain the steering angle sensor of driver's steering angle information ⁇ f , is used to obtain the yaw rate sensor of yaw rate ⁇ information and is used to obtain the centroid side slip angle sensor of center of mass side slip angle ⁇ information;
  • each sensor is conventional technologies, such as the angle sensor being arranged at the steering shaft of the steering wheel of the motor vehicle, and the yaw rate sensor and the center of mass side slip angle sensor being arranged at the center of mass of the motor vehicle;
  • the vehicle-mounted ECU stores a constant K 1 representing the identification gain, a constant ⁇ representing the online learning gain, a time constant ⁇ r , a time constant ⁇ ⁇ , a vehicle mass m, a distance l f from the center of mass to the front axle, a distance l r from the center of mass to the rear axle, and a constant K 2 representing the control gain;
  • Front wheel cornering stiffness C f , rear wheel cornering stiffness C r , vehicle mass m, distance l f from the center of mass to the front axle, and distance l r from the center of mass to the rear axle are all determined by the vehicle manufacturer and stored in the on-board ECU.
  • I Z is the moment of inertia around the vertical (vertical) direction of the motor vehicle, which is determined by the motor vehicle manufacturer and stored in the on-board ECU;
  • the first step is to construct the basic equations stored in the on-board ECU
  • Equation 1 The basic equations include Equation 1 to Equation 15;
  • Equation 1 is the parameterized neural network:
  • ⁇ is the sideslip angle of the center of mass
  • ⁇ is the yaw rate
  • a ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 matrix in the real field R
  • ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 matrix in the real field R
  • ⁇ (x) is the sigmoidal activation function in the neural network
  • ⁇ (x) [ ⁇ (x 1 ), ⁇ (x 2 )] T
  • ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 matrix in the real field R
  • u [ ⁇ c , M c ] T
  • ⁇ R 2 is a 2-dimensional column vector in the real field R
  • ⁇ C is the active steering angle
  • M c is the yaw moment
  • ⁇ C and M c is the calculation result that is finally used to control the vehicle
  • the second sub-step is to design the recognizer and online self-learning rate, and construct formula 3 and formula 4;
  • Equation 3 is the identifier used to achieve system identification:
  • Formula 4 is the online self-learning rate:
  • ⁇ is a constant obtained by trial and error and stored in the on-board ECU, ⁇ >0 and represents the online learning gain;
  • the third sub-step is to construct the control target reference model expressed by formula 5, and generate the target tracking signal;
  • x r is the reference state
  • x r [ ⁇ r ⁇ r ] T
  • ⁇ r is the sideslip angle of the target center of mass
  • ⁇ r is the target yaw rate
  • ⁇ r and ⁇ r are the calculation results of formula 5 and serve as target tracking signals
  • I Z is the moment of inertia of the motor vehicle around the vertical (vertical) direction, the unit is kg m2 (ie kilogram ⁇ m2 ), determined by the motor vehicle manufacturer and stored in the on-board ECU;
  • the fourth sub-step is to construct the controller expressed by formula 6 based on the approximate dynamic programming theory, so as to realize the cooperative control of active steering and yaw moment self-learning;
  • Equation 6 The u 1 in Equation 6 is used to ensure that the steady-state error of the control converges to zero, and its expression is Equation 7:
  • K 2 is a constant indicating the control gain obtained by trial and error and stored in the on-board ECU, K 2 >0;
  • the u 2 in formula 6 is used to ensure the optimal control performance, and it is designed according to the approximate dynamic programming theory, specifically obtained through formula 8 to formula 15;
  • Equation 8 is the evaluation function V as follows:
  • Q ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 diagonal matrix in the real number field R, which represents the weight of the tracking error e 2 in the optimal control evaluation function, which is obtained by trial and error and stored in the vehicle ECU;
  • P ⁇ R 2 ⁇ 2 is a 2 ⁇ 2 diagonal matrix in the real number field R, representing the weight of u 2 in the optimal control evaluation function, which is obtained by trial and error and stored in the vehicle ECU;
  • Equation 9 is the Hamiltonian function:
  • W V is the ideal weight vector
  • ⁇ (e 2 ) is the sigmoidal activation function in the neural network
  • Equation 11
  • Equation 12 Equation 12
  • Formula 12 is:
  • the second step is to calculate the values of the active steering angle ⁇ C and the yaw moment Mc online according to the following sub-steps during the running of the vehicle, and control the running state of the motor vehicle according to the calculation results of ⁇ C and Mc ;
  • the first sub-step is that the ECU collects original real-time parameter values, including the steering angle ⁇ f value measured by the steering angle sensor when the driver manipulates the steering wheel, the vehicle speed v x value from the vehicle speed sensor, the mass center side slip angle ⁇ value from the mass center slip angle sensor, and the yaw rate ⁇ value from the yaw rate sensor;
  • the second sub-step is the calculation step of the identifier and the reference model of the control target
  • the on-board ECU provides the center of mass sideslip angle ⁇ value and the yaw rate ⁇ value to the identifier described in formula 3, and calculates as the basis for calculating the value of u1 in Equation 6;
  • the vehicle-mounted ECU provides the steering angle ⁇ f value and the vehicle speed v x value to the control target reference model described in formula 5, and calculates the target mass center side slip angle ⁇ r value and the target yaw rate ⁇ r value, which are used as the calculation basis for the parameters e 2 required by u 1 and u 2 in formula 6;
  • the third sub-step is that the on-board ECU converts ⁇ r , ⁇ r and Provided to the controller described in Formula 6, the active steering angle ⁇ C and the yaw moment M c are calculated through Formula 6 to Formula 15;
  • the vehicle-mounted ECU controls the steering action of the steering wheel of the motor vehicle according to the active steering angle ⁇ C calculated in real time, and controls the braking action of the brake of the motor vehicle according to the yaw moment M c calculated in real time;
  • the vehicle-mounted ECU controls the vehicle to change lanes at a vehicle speed of 28 m/s according to the active steering angle ⁇ C and yaw moment Mc .
  • the control results show that the method proposed in the present invention can realize self-learning cooperative control of active steering and yaw moment without model information.

Landscapes

  • Engineering & Computer Science (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Theoretical Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Steering Control In Accordance With Driving Conditions (AREA)

Abstract

一种主动转向和横摆力矩自学习协同控制方法，第一步骤是构建储存于车载ECU中的基础方程，第二步骤是在车辆的行驶过程中，车载ECU按如下子步骤在线计算主动转向角δ C和横摆力矩M c的值，并根据δ C和M c的计算结果控制机动车的运行状态；第一子步骤是ECU采集原始实时参数值，第二子步骤是辨识器和控制目标参考模型计算步骤；第三子步骤是主动转向角δ C和横摆力矩M c计算步骤；重复第二步骤，该控制方法无须系统控制模型即可实现主动转向和横摆力矩自学习协同控制，修正驾驶员的转向操作，克服不当驾驶，使车拐弯时趋向于中性转向中性转向，机动车反馈的实际质心侧偏角和横摆角速度趋近于公式5计算得到的β r值和γr值，避免不当驾驶引起的失稳等事故。

Description

一种主动转向和横摆力矩自学习协同控制方法 技术领域
本发明涉及属于自动控制、信息技术和先进制造领域，具体涉及针对具有未知模型的主动转向和横摆力矩协同控制问题。
背景技术
先进的主动底盘控制系统在实现追求更好的操纵性和稳定性、更安全汽车的目标方面起着关键作用。对于实际底盘系统来说，应具有更高的智能性，以适应各种路况。
主动转向控制系统可向驾驶员的转向指令施加额外的转向角，并通过调节轮胎横向力直接影响车辆横向动态行为。
然而，当轮胎进入非线性区域时，主动转向控制系统无法产生足够的轮胎侧向力，因此其在线性操纵区域的性能受到限制。而横摆力矩控制系统对于线性和非线性操纵区域的车辆稳定性非常有效，其通过车辆左右两侧的差速制动，产生适量的校正横摆力矩，以保持车辆稳定性。
但由于轮胎磨损和制动引起的显著减速，横摆力矩控制系统仅适用于有限的操纵。单独的主动转向控制系统和横摆力矩控制系统无法实现全面最优的安全性能。因此，需要集成主动转向控制系统和横摆力矩控制系统，通过两者之间的协同控制方法实现跟踪期望的横摆角速度和侧滑角，得到在不同驾驶动作下获得令人满意的车辆行驶稳定性性能。
轮胎转弯刚度参数、纵向车速、轮胎的非线性特性等参数所带来的模型不确定性给控制器的设计带来巨大困难。现有常用的控制方法是基于线性二次型调节方法(LQR)，其主要缺点在于必须事先精确地知道悬架系统模型才能找到最优控制律。此外，反馈控制增益是由离线求解黎卡提(Riccati)方程获得的，一旦获得控制器的反馈增益，它们就不能随着不同的路面外界输入及相关车辆不确定参数而改变。因此，需要一种更有效的控制策略来实时自适应地处理不同行驶工况下参数随时间变化的智能底盘控制问题。
发明内容
本发明的目的在于提供一种主动转向和横摆力矩自学习协同控制方法，事先不需要获知悬架系统模型，应对模型不确定性给控制器的设计带来巨大困难的问题，只需要传感器数据即可进行控制，用于实时自适应地处理不同行驶工况下参数随时间变化的智能底盘控制问题。
为实现上述目的，本发明的一种主动转向和横摆力矩自学习协同控制方法用于机动车，机动车具有车载ECU，车载ECU连接有用于获取车速v x信息的车速传感器、用于获取驾驶 员转向角信息δ f的转角传感器，用于获取横摆角速度γ信息的横摆角速度传感器和用于获取质心侧偏角β信息的质心侧偏角传感器；
各传感器的设置位置为常规技术，如转角传感器设置于机动车的方向盘的转向轴处，横摆角速度传感器和质心侧偏角传感器设于机动车的质心处；
车载ECU内存储有表示辨识增益的常数K 1、表示在线学习增益的常数Γ、时间常数τ r、时间常数τ β、车辆质量m、质心到前轴的距离l f、质心到后轴的距离l r和表示控制增益的常数K 2；
第一步骤是构建储存于车载ECU中的基础方程，包括辨识器、控制目标参考模型和控制器；
基础方程包括公式1至公式15；
第一步骤第一子步骤是构建公式1和公式2；公式1是参数化神经网络：
Figure PCTCN2022138265-appb-000001
公式1中，x＝[β,γ] T∈R 2为实数域R中的2维列向量，β为质心侧偏角，其单位为弧度rad，γ为横摆角速度，其单位为弧度每秒rad/s，a∈R 2×2为实数域R中的2×2矩阵，ω∈R 2×2为实数域R中的2×2矩阵，σ(x)为神经网络中的sigmoidal激励函数，σ(x)＝[σ(x 1),σ(x 2)] T，λ∈R 2×2为实数域R中的2×2矩阵，u＝[δ c,M c] T∈R 2为实数域R中的2维列向量；δ C为主动转向角，其单位为弧度rad，M c为横摆力矩，其单位为N·m；δ C和M c是最终用于控制车辆的计算结果；
公式2为公式1的改写形式：
Figure PCTCN2022138265-appb-000002
其中θ＝[a,w,λ] T，ψ＝[x,σ(x),u] T；
第一步骤第二子步骤是设计辨识器及在线自学习率，构建公式3和公式4；
公式3是用于实现系统辨识的辨识器：
Figure PCTCN2022138265-appb-000003
公式3中，
Figure PCTCN2022138265-appb-000004
表示辨识误差，
Figure PCTCN2022138265-appb-000005
为辨识状态，
Figure PCTCN2022138265-appb-000006
表示辨识参数，K 1＞0为试凑得到并存储于车载ECU内的常数，表示辨识增益；公式3的辨识结果是
Figure PCTCN2022138265-appb-000007
并提供给公式7；
公式4是在线自学习率：
Figure PCTCN2022138265-appb-000008
公式4中，Γ为试凑得到并存储于车载ECU内的常数，Γ＞0并表示在线学习增益；
第一步骤第三子步骤是构建公式5表达的控制目标参考模型，产生目标跟踪信号；
公式5是：
Figure PCTCN2022138265-appb-000009
公式5中，x r为参考状态，x r＝[β r γ r] T,β r为目标质心侧偏角，γ r为目标横摆角速 度，β r和γ r均为公式5的计算结果并作为目标跟踪信号；
公式5中，
Figure PCTCN2022138265-appb-000010
其中，
Figure PCTCN2022138265-appb-000011
Figure PCTCN2022138265-appb-000012
Figure PCTCN2022138265-appb-000013
I Z是机动车绕垂向的转动惯量，单位为kg·m 2，由机动车厂家确定并存储在车载ECU中；
Figure PCTCN2022138265-appb-000014
m为车辆质量，其单位为kg；l f为质心到前轴的距离，其单位为米，l r为质心到后轴的距离，其单位为米，v x为由车速传感器获取的车速，其单位为米每秒，C f为为前轮侧偏刚度，其单位为N/rad，C r为后轮侧偏刚度，其单位为N/rad，；
第一步骤第四子步骤是基于近似动态规划理论构建公式6表达的控制器，以实现主动转向和横摆力矩自学习协同控制；
公式6是控制器：u＝u 1+u 2；
公式6中的u 1用来确保控制的稳态误差收敛到零，其表达式为公式7：
Figure PCTCN2022138265-appb-000015
公式7中，
Figure PCTCN2022138265-appb-000016
表示λ的广义逆，其中
Figure PCTCN2022138265-appb-000017
表示跟踪误差，x r来自公式5；
K 2为试凑得到并存储于车载ECU内的表示控制增益的常数，K 2＞0；
公式6中的u 2用来确保控制的性能最优，根据近似动态规划理论设计而来，具体通过公式8至公式15得到；
公式8是评价函数V如下：
Figure PCTCN2022138265-appb-000018
公式8中，Q∈R 2×2为实数域R中的2×2对角矩阵，表示优化控制评价函数中跟踪误差e 2的权重，通过试凑得到并存储于车载ECU内；
P∈R 2×2为实数域R中的2×2对角矩阵，表示优化控制评价函数中u 2的权重，通过试凑得到并存储于车载ECU内；
公式9是哈密顿函数：
Figure PCTCN2022138265-appb-000019
公式9中，
Figure PCTCN2022138265-appb-000020
表示V关于e 2的偏导数，由于最优评价函数V是未知的，需通过公式10中的神经网络来逼近V，
公式10是：
Figure PCTCN2022138265-appb-000021
公式10中，W V为理想权值向量，σ(e 2)为神经网络中的sigmoidal激励函数；
由公式10推导得到
Figure PCTCN2022138265-appb-000022
的表达式为公式11，
公式11是：
Figure PCTCN2022138265-appb-000023
将公式11代入公式9中得哈密顿函数即公式12，
公式12是：
Figure PCTCN2022138265-appb-000024
令公式(12)的等式左边等于零可得公式13，
公式13是：
Figure PCTCN2022138265-appb-000025
公式13中，
Figure PCTCN2022138265-appb-000026
由最小二乘法的原理可得公式14，
公式14是：W V＝(N TN) -1N TM；
由公式(12)通过求解
Figure PCTCN2022138265-appb-000027
可得公式15，
公式15是：
Figure PCTCN2022138265-appb-000028
第二步骤是在车辆的行驶过程中，车载ECU按如下子步骤在线计算主动转向角δ C和横摆力矩M c的值，并根据δ C和M c的计算结果控制机动车的运行状态。
第二步骤第一子步骤是ECU采集原始实时参数值，包括来自驾驶员操纵方向盘时转角传感器测得的转向角δ f数值、来自车速传感器的车速v x数值、来自质心侧偏角传感器的质心侧偏角β数值以及来自横摆角速度传感器的横摆角速度γ数值。
第二步骤第二子步骤是辨识器和控制目标参考模型计算步骤；
车载ECU将质心侧偏角β数值和横摆角速度γ数值提供给公式3所述的辨识器，计算得到
Figure PCTCN2022138265-appb-000029
作为计算公式6中u 1数值的基础；
并提供给公式7从而计算得到公式6所需要的u 1数值；
同时车载ECU将转向角δ f数值和车速v x数值提供给公式5所述的控制目标参考模型，计算得到目标质心侧偏角β r值和目标横摆角速度γ r值，作为公式6中u 1和u 2所需要的参数e 2的计算基础。
第二步骤第三子步骤是车载ECU将β r、γ r和
Figure PCTCN2022138265-appb-000030
提供给公式6所述的控制器，通过公式6至公式15计算得到主动转向角δ C和横摆力矩M c；
车载ECU按照实时计算得到的主动转向角δ C控制机动车方向盘的转向动作，并按照实时计算得到的横摆力矩M c控制机动车制动器的制动动作；
重复执行第二步骤，实现在线无模型主动转向和横摆力矩自学习协同控制。
本发明具有如下的优点：
本发明摆脱了以往必须事先精确地知道悬架系统模型才能找到最优控制律的约束，能够实时自适应地处理不同行驶工况下参数随时间变化的智能底盘控制问题。
采用本发明，无须系统控制模型，车载ECU就能够实现对主动转向和横摆力矩的自学习协同控制，计算出理想的主动转向角δ C和横摆力矩M c，对制动和转向进行调整和控制，修正驾驶员的转向操作，克服不当驾驶，使车拐弯时趋向于中性转向；车载ECU按照主动转向角δ C和横摆力矩M c控制机动车转向动作后，机动车反馈的实际质心侧偏角和横摆角速度趋近于公式5计算得到的β r值和γ r值，避免不当驾驶引起的失稳等事故。
附图说明
图1是本发明的原理示意图；
图2是车载ECU按主动转向角δ C和横摆力矩M c控制机动车在车速为28米/秒的情况下 变道时的实测转向角－时间曲线图；
图3是车辆进行图2中的变道动作时，质心侧偏角的参考值(即公式5计算得到的β r值)与控制结果实侧值的对比图；
图4是车辆进行图2中的变道动作时，横摆角速度的参考值(即公式5计算得到的γ r值)与控制结果实侧值的对比图。
具体实施方式
如图1所示，本发明的一种主动转向和横摆力矩自学习协同控制方法用于机动车，机动车具有车载ECU，车载ECU连接有用于获取车速v x信息的车速传感器、用于获取驾驶员转向角信息δ f的转角传感器，用于获取横摆角速度γ信息的横摆角速度传感器和用于获取质心侧偏角β信息的质心侧偏角传感器；
各传感器的设置位置为常规技术，如转角传感器设置于机动车的方向盘的转向轴处，横摆角速度传感器和质心侧偏角传感器设于机动车的质心处；
车载ECU内存储有表示辨识增益的常数K 1、表示在线学习增益的常数Γ、时间常数τ r、时间常数τ β、车辆质量m、质心到前轴的距离l f、质心到后轴的距离l r和表示控制增益的常数K 2；
前轮侧偏刚度C f、后轮侧偏刚度C r、车辆质量m、质心到前轴的距离l f、质心到后轴的距离l r均由机动车厂家确定并存储在车载ECU中。I Z是机动车绕垂向(竖向)的转动惯量，由机动车厂家确定并存储在车载ECU中；
第一步骤是构建储存于车载ECU中的基础方程；
基础方程包括公式1至公式15；
各公式中，x＝[β,γ] T∈R 2为实数域R中的2维列向量；β为质心侧偏角，其单位为弧度rad；γ为横摆角速度，其单位为弧度每秒rad/s；a∈R 2×2为实数域R中的2×2矩阵；ω∈R 2×2为实数域R中的2×2矩阵；σ(x)为神经网络中的sigmoidal激励函数；σ(x)＝[σ(x 1),σ(x 2)] T；λ∈R 2×2为实数域R中的2×2矩阵；u＝[δ c,M c] T∈R 2为实数域R中的2维列向量；δ C为主动转向角，其单位为弧度rad，M c为横摆力矩，其单位为N·m(即牛顿×米)；θ＝[a,w,λ] T，ψ＝[x,σ(x),u] T；
Figure PCTCN2022138265-appb-000031
表示辨识误差；
Figure PCTCN2022138265-appb-000032
为辨识状态；
Figure PCTCN2022138265-appb-000033
表示辨识参数；K 1为试凑得到并存储于车载ECU内的表示辨识增益的常数，K 1＞0；Γ为试凑得到并存储于车载ECU内的表示在线学习增益的常数，Γ＞0；β r为目标质心侧偏角，γ r为目标横摆角速度，m为车辆质量，l f为质心到前轴的距离；l r为质心到后轴的距离，v x为由车速传感器获取的 车速，C f为前轮侧偏刚度，C r为后轮侧偏刚度；x r为参考状态，K 2为试凑得到并存储于车载ECU内的表示控制增益的常数；Q∈R 2×2为实数域R中的2×2对角矩阵并表示公式8中跟踪误差e 2的权重；P∈R 2×2为实数域R中的2×2对角矩阵并表示公式8中u 2的权重；P和Q分别通过试凑得到并存储于车载ECU内；V是评价函数；W V为理想权值向量，σ(e 2)为神经网络中的sigmoidal激励函数；I Z是机动车绕垂向(竖向)的转动惯量，单位为千克×米 2；
第一子步骤是构建公式1和公式2；公式1是参数化神经网络：
Figure PCTCN2022138265-appb-000034
公式1中，x＝[β,γ] T∈R 2为实数域R中的2维列向量，β为质心侧偏角，γ为横摆角速度，a∈R 2×2为实数域R中的2×2矩阵，ω∈R 2×2为实数域R中的2×2矩阵，σ(x)为神经网络中的sigmoidal激励函数，σ(x)＝[σ(x 1),σ(x 2)] T，λ∈R 2×2为实数域R中的2×2矩阵，u＝[δ c,M c] T∈R 2为实数域R中的2维列向量；δ C为主动转向角，M c为横摆力矩；δ C和M c是最终用于控制车辆的计算结果；
公式2为公式1的改写形式：
Figure PCTCN2022138265-appb-000035
其中θ＝[a,w,λ] T，ψ＝[x,σ(x),u] T；
第二子步骤是设计辨识器及在线自学习率，构建公式3和公式4；
公式3是用于实现系统辨识的辨识器：
Figure PCTCN2022138265-appb-000036
公式3中，
Figure PCTCN2022138265-appb-000037
表示辨识误差，
Figure PCTCN2022138265-appb-000038
为辨识状态，
Figure PCTCN2022138265-appb-000039
表示辨识参数，K 1＞0为试凑得到并存储于车载ECU内的常数，表示辨识增益；公式3的辨识结果是
Figure PCTCN2022138265-appb-000040
并提供给公式7；
公式4是在线自学习率：
Figure PCTCN2022138265-appb-000041
公式4中，Γ为试凑得到并存储于车载ECU内的常数，Γ＞0并表示在线学习增益；
第三子步骤是构建公式5表达的控制目标参考模型，产生目标跟踪信号；
公式5是：
Figure PCTCN2022138265-appb-000042
公式5中，x r为参考状态，x r＝[β r γ r] T,β r为目标质心侧偏角，γ r为目标横摆角速度，β r和γ r均为公式5的计算结果并作为目标跟踪信号；
公式5中，
Figure PCTCN2022138265-appb-000043
其中，
Figure PCTCN2022138265-appb-000044
Figure PCTCN2022138265-appb-000045
Figure PCTCN2022138265-appb-000046
I Z是机动车绕垂向(竖向)的转动惯量，单位为kg·m2(即千克×米 2)，由机动车厂家确定并存储在车载ECU中；
Figure PCTCN2022138265-appb-000047
m为车辆质量，其单位为kg(千克)；l f为质心到前轴的距离，其单位为米；l r为质心到后轴的距离，其单位为米；v x为由车速传感器获取的车速，其单位为m/s(即米每秒)；C f为为前轮侧偏刚度，其单位为N/rad(即牛顿/弧度)；C r为后轮侧偏刚度，其单位为N/rad；
第四子步骤是基于近似动态规划理论构建公式6表达的控制器，以实现主动转向和横摆力矩自学习协同控制；
公式6是控制器：u＝u 1+u 2；
公式6中的u 1用来确保控制的稳态误差收敛到零，其表达式为公式7：
Figure PCTCN2022138265-appb-000048
公式7中，
Figure PCTCN2022138265-appb-000049
表示λ的广义逆，其中
Figure PCTCN2022138265-appb-000050
表示跟踪误差，x r来自公式5；
K 2为试凑得到并存储于车载ECU内的表示控制增益的常数，K 2＞0；
公式6中的u 2用来确保控制的性能最优，根据近似动态规划理论设计而来，具体通过公式8至公式15得到；
公式8是评价函数V如下：
Figure PCTCN2022138265-appb-000051
公式8中，Q∈R 2×2为实数域R中的2×2对角矩阵，表示优化控制评价函数中跟踪误差e 2的权重，通过试凑得到并存储于车载ECU内；
P∈R 2×2为实数域R中的2×2对角矩阵，表示优化控制评价函数中u 2的权重，通过试凑得到并存储于车载ECU内；
公式9是哈密顿函数：
Figure PCTCN2022138265-appb-000052
公式9中，
Figure PCTCN2022138265-appb-000053
表示V关于e 2的偏导数，由于最优评价函数V是未知的，需通过公式10中的神经网络来逼近V，
公式10是：
Figure PCTCN2022138265-appb-000054
公式10中，W V为理想权值向量，σ(e 2)为神经网络中的sigmoidal激励函数；
由公式10推导得到
Figure PCTCN2022138265-appb-000055
的表达式为公式11，
公式11是：
Figure PCTCN2022138265-appb-000056
将公式11代入公式9中得哈密顿函数即公式12，
公式12是：
Figure PCTCN2022138265-appb-000057
令公式(12)的等式左边等于零可得公式13，
公式13是：
Figure PCTCN2022138265-appb-000058
公式13中，
Figure PCTCN2022138265-appb-000059
由最小二乘法的原理可得公式14，
公式14是：W V＝(N TN) -1N TM；
由公式(12)通过求解
Figure PCTCN2022138265-appb-000060
可得公式15，
公式15是：
Figure PCTCN2022138265-appb-000061
第二步骤是在车辆的行驶过程中，车载ECU按如下子步骤在线计算主动转向角δ C和横摆力矩M c的值，并根据δ C和M c的计算结果控制机动车的运行状态；
第一子步骤是ECU采集原始实时参数值，包括来自驾驶员操纵方向盘时转角传感器测得的转向角δ f数值、来自车速传感器的车速v x数值、来自质心侧偏角传感器的质心侧偏角β数值以及来自横摆角速度传感器的横摆角速度γ数值；
第二子步骤是辨识器和控制目标参考模型计算步骤；
车载ECU将质心侧偏角β数值和横摆角速度γ数值提供给公式3所述的辨识器，计算得到
Figure PCTCN2022138265-appb-000062
作为计算公式6中u 1数值的基础；
并提供给公式7从而计算得到公式6所需要的u 1数值；
同时车载ECU将转向角δ f数值和车速v x数值提供给公式5所述的控制目标参考模型，计算得到目标质心侧偏角β r值和目标横摆角速度γ r值，作为公式6中u 1和u 2所需要的参数e 2的计算基础；
第三子步骤是车载ECU将β r、γ r和
Figure PCTCN2022138265-appb-000063
提供给公式6所述的控制器，通过公式6至公式15计算得到主动转向角δ C和横摆力矩M c；
车载ECU按照实时计算得到的主动转向角δ C控制机动车方向盘的转向动作，并按照实时计算得到的横摆力矩M c控制机动车制动器的制动动作；
重复执行第二步骤，实现在线无模型主动转向和横摆力矩自学习协同控制。
如图2至图4所示，车载ECU按主动转向角δ C和横摆力矩M c控制机动车在车速为28米/秒的情况下变道，控制结果表明本发明所提出的方法可以在不需要模型信息的情况下实现对主动转向和横摆力矩的自学习协同控制，质心侧偏角和横摆角速度均趋近于参考值，机动车变道动作平滑稳定，远离侧翻风险。
以上实施例仅用以说明而非限制本发明的技术方案，尽管参照上述实施例对本发明进行了详细说明，本领域的普通技术人员应当理解：依然可以对本发明进行修改或者等同替换，而不脱离本发明的精神和范围的任何修改或局部替换，其均应涵盖在本发明的权利要求范围当中。

Claims (4)

  1. 一种主动转向和横摆力矩自学习协同控制方法，用于机动车，机动车具有车载ECU，车载ECU连接有用于获取车速v x信息的车速传感器、用于获取驾驶员转向角信息δ f的转角传感器，用于获取横摆角速度γ信息的横摆角速度传感器和用于获取质心侧偏角β信息的质心侧偏角传感器；
    各传感器的设置位置为常规技术，如转角传感器设置于机动车的方向盘的转向轴处，横摆角速度传感器和质心侧偏角传感器设于机动车的质心处；
    车载ECU内存储有表示辨识增益的常数K 1、表示在线学习增益的常数Γ、时间常数τ r、时间常数τ β、车辆质量m、质心到前轴的距离l f、质心到后轴的距离l r和表示控制增益的常数K 2；
    其特征在于：
    第一步骤是构建储存于车载ECU中的基础方程，包括辨识器、控制目标参考模型和控制器；
    基础方程包括公式1至公式15；
    第一步骤第一子步骤是构建公式1和公式2；公式1是参数化神经网络：
    Figure PCTCN2022138265-appb-100001
    公式1中，x＝[β,γ] T∈R 2为实数域R中的2维列向量，β为质心侧偏角，其单位为弧度rad，γ为横摆角速度，其单位为弧度每秒rad/s，a∈R 2×2为实数域R中的2×2矩阵，ω∈R 2×2为实数域R中的2×2矩阵，σ(x)为神经网络中的sigmoidal激励函数，σ(x)＝[σ(x 1),σ(x 2)] T，λ∈R 2×2为实数域R中的2×2矩阵，u＝[δ c,M c] T∈R 2为实数域R中的2维列向量；δ C为主动转向角，其单位为弧度rad，M c为横摆力矩，其单位为N·m；δ C和M c是最终用于控制车辆的计算结果；
    公式2为公式1的改写形式：
    Figure PCTCN2022138265-appb-100002
    其中θ＝[a,w,λ] T，ψ＝[x,σ(x),u] T；
    第一步骤第二子步骤是设计辨识器及在线自学习率，构建公式3和公式4；
    公式3是用于实现系统辨识的辨识器：
    Figure PCTCN2022138265-appb-100003
    公式3中，
    Figure PCTCN2022138265-appb-100004
    表示辨识误差，
    Figure PCTCN2022138265-appb-100005
    为辨识状态，
    Figure PCTCN2022138265-appb-100006
    表示辨识参数，K 1＞0为试凑得到并存储于车载ECU内的常数，表示辨识增益；公式3的辨识结果是
    Figure PCTCN2022138265-appb-100007
    并提供给公式7；
    公式4是在线自学习率：
    Figure PCTCN2022138265-appb-100008
    公式4中，Γ为试凑得到并存储于车载ECU内的常数，Γ＞0并表示在线学习增益；
    第一步骤第三子步骤是构建公式5表达的控制目标参考模型，产生目标跟踪信号；
    公式5是：
    Figure PCTCN2022138265-appb-100009
    公式5中，x r为参考状态，x r＝[β r γ r] T,β r为目标质心侧偏角，γ r为目标横摆角速度，β r和γ r均为公式5的计算结果并作为目标跟踪信号；
    公式5中，
    Figure PCTCN2022138265-appb-100010
    其中，
    Figure PCTCN2022138265-appb-100011
    Figure PCTCN2022138265-appb-100012
    Figure PCTCN2022138265-appb-100013
    I Z是机动车绕垂向的转动惯量，单位为kg·m 2，由机动车厂家确定并存储在车载ECU中；
    Figure PCTCN2022138265-appb-100014
    m为车辆质量，其单位为kg；l f为质心到前轴的距离，其单位为米，l r为质心到后轴的距离，其单位为米，v x为由车速传感器获取的车速，其单位为米每秒，C f为为前轮侧偏刚度，其单位为N/rad，C r为后轮侧偏刚度，其单位为N/rad，；
    第一步骤第四子步骤是基于近似动态规划理论构建公式6表达的控制器，以实现主动转向和横摆力矩自学习协同控制；
    公式6是控制器：u＝u 1+u 2；
    公式6中的u 1用来确保控制的稳态误差收敛到零，其表达式为公式7：
    Figure PCTCN2022138265-appb-100015
    公式7中，
    Figure PCTCN2022138265-appb-100016
    表示λ的广义逆，其中
    Figure PCTCN2022138265-appb-100017
    表示跟踪误差，x r来自公式5；
    K 2为试凑得到并存储于车载ECU内的表示控制增益的常数，K 2＞0；
    公式6中的u 2用来确保控制的性能最优，根据近似动态规划理论设计而来，具体通过公式8至公式15得到；
    公式8是评价函数V如下：
    Figure PCTCN2022138265-appb-100018
    公式8中，Q∈R 2×2为实数域R中的2×2对角矩阵，表示优化控制评价函数中跟踪误差e 2的权重，通过试凑得到并存储于车载ECU内；
    P∈R 2×2为实数域R中的2×2对角矩阵，表示优化控制评价函数中u 2的权重，通过试凑得到并存储于车载ECU内；
    公式9是哈密顿函数：
    Figure PCTCN2022138265-appb-100019
    公式9中，
    Figure PCTCN2022138265-appb-100020
    表示V关于e 2的偏导数，由于最优评价函数V是未知的，需通过公式10中的神经网络来逼近V，
    公式10是：
    Figure PCTCN2022138265-appb-100021
    公式10中，W V为理想权值向量，σ(e 2)为神经网络中的sigmoidal激励函数；
    由公式10推导得到
    Figure PCTCN2022138265-appb-100022
    的表达式为公式11，
    公式11是：
    Figure PCTCN2022138265-appb-100023
    将公式11代入公式9中得哈密顿函数即公式12，
    公式12是：
    Figure PCTCN2022138265-appb-100024
    令公式(12)的等式左边等于零可得公式13，
    公式13是：
    Figure PCTCN2022138265-appb-100025
    公式13中，
    Figure PCTCN2022138265-appb-100026
    由最小二乘法的原理可得公式14，
    公式14是：W V＝(N TN) -1N TM；
    由公式(12)通过求解
    Figure PCTCN2022138265-appb-100027
    可得公式15，
    公式15是：
    Figure PCTCN2022138265-appb-100028
    第二步骤是在车辆的行驶过程中，车载ECU按如下子步骤在线计算主动转向角δ C和横摆力矩M c的值，并根据δ C和M c的计算结果控制机动车的运行状态。
  2. 根据权利要求1所述的一种主动转向和横摆力矩自学习协同控制方法，其特征在于：
    第二步骤第一子步骤是ECU采集原始实时参数值，包括来自驾驶员操纵方向盘时转角传感器测得的转向角δ f数值、来自车速传感器的车速v x数值、来自质心侧偏角传感器的质心侧偏角β数值以及来自横摆角速度传感器的横摆角速度γ数值。
  3. 根据权利要求2所述的一种主动转向和横摆力矩自学习协同控制方法，其特征在于：
    第二步骤第二子步骤是辨识器和控制目标参考模型计算步骤；
    车载ECU将质心侧偏角β数值和横摆角速度γ数值提供给公式3所述的辨识器，计算得到
    Figure PCTCN2022138265-appb-100029
    作为计算公式6中u 1数值的基础；
    并提供给公式7从而计算得到公式6所需要的u 1数值；
    同时车载ECU将转向角δ f数值和车速v x数值提供给公式5所述的控制目标参考模型，计算得到目标质心侧偏角β r值和目标横摆角速度γ r值，作为公式6中u 1和u 2所需要的参数e 2的计算基础。
  4. 根据权利要求3所述的一种主动转向和横摆力矩自学习协同控制方法，其特征在于：
    第二步骤第三子步骤是车载ECU将β r、γ r和
    Figure PCTCN2022138265-appb-100030
    提供给公式6所述的控制器，通过公式6至公式15计算得到主动转向角δ C和横摆力矩M c；
    车载ECU按照实时计算得到的主动转向角δ C控制机动车方向盘的转向动作，并按照实时计算得到的横摆力矩M c控制机动车制动器的制动动作；
    重复执行第二步骤，实现在线无模型主动转向和横摆力矩自学习协同控制。
PCT/CN2022/138265 2022-01-18 2022-12-11 一种主动转向和横摆力矩自学习协同控制方法 WO2023138258A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210052344.1A CN114312750B (zh) 2022-01-18 2022-01-18 一种主动转向和横摆力矩自学习协同控制方法
CN202210052344.1 2022-01-18

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/395,716 Continuation US20240132152A1 (en) 2022-01-18 2023-12-25 Self-learning collaborative control method for active steering and yaw moment

Publications (1)

Publication Number Publication Date
WO2023138258A1 true WO2023138258A1 (zh) 2023-07-27

Family

ID=81029642

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/138265 WO2023138258A1 (zh) 2022-01-18 2022-12-11 一种主动转向和横摆力矩自学习协同控制方法

Country Status (2)

Country Link
CN (1) CN114312750B (zh)
WO (1) WO2023138258A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114312750B (zh) * 2022-01-18 2022-08-09 郑州轻工业大学 一种主动转向和横摆力矩自学习协同控制方法
CN115195757B (zh) * 2022-09-07 2023-08-04 郑州轻工业大学 电动公交起步驾驶行为建模及识别训练方法
CN117311346A (zh) * 2023-09-15 2023-12-29 广东省机场管理集团有限公司工程建设指挥部 机器人横向稳定协同转向控制方法、装置、终端及介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20130052790A (ko) * 2011-11-14 2013-05-23 현대자동차주식회사 차량 동적 성능 향상 장치 및 방법
US20150291147A1 (en) * 2014-04-11 2015-10-15 Mando Corporation Apparatus and method for electronic stability control of vehicle
CN111746633A (zh) * 2020-07-02 2020-10-09 南京航空航天大学 一种基于强化学习的车辆分布式转向驱动系统控制方法
CN113022551A (zh) * 2021-03-26 2021-06-25 华南理工大学 一种操纵杆线控转向控制方法
CN114312750A (zh) * 2022-01-18 2022-04-12 郑州轻工业大学 一种主动转向和横摆力矩自学习协同控制方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103019098B (zh) * 2012-12-15 2016-09-14 江苏大学 汽车底盘集成系统鲁棒控制器及构造方法
CN109050661B (zh) * 2018-09-20 2020-04-17 合肥工业大学 电子差速和主动差动转向的协调控制方法及协调控制装置
CN111487870B (zh) * 2020-04-26 2022-09-16 贵州理工学院 一种柔性主动悬架系统中自适应反演控制器的设计方法
CN112987687B (zh) * 2021-02-08 2022-05-24 南京航空航天大学 一种云-端融合的智能线控底盘健康监测系统及方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20130052790A (ko) * 2011-11-14 2013-05-23 현대자동차주식회사 차량 동적 성능 향상 장치 및 방법
US20150291147A1 (en) * 2014-04-11 2015-10-15 Mando Corporation Apparatus and method for electronic stability control of vehicle
CN111746633A (zh) * 2020-07-02 2020-10-09 南京航空航天大学 一种基于强化学习的车辆分布式转向驱动系统控制方法
CN113022551A (zh) * 2021-03-26 2021-06-25 华南理工大学 一种操纵杆线控转向控制方法
CN114312750A (zh) * 2022-01-18 2022-04-12 郑州轻工业大学 一种主动转向和横摆力矩自学习协同控制方法

Also Published As

Publication number Publication date
CN114312750A (zh) 2022-04-12
CN114312750B (zh) 2022-08-09

Similar Documents

Publication Publication Date Title
WO2023138258A1 (zh) 一种主动转向和横摆力矩自学习协同控制方法
US11364895B2 (en) Yaw motion control method for four-wheel distributed vehicle
CN109849899B (zh) 一种电动轮汽车电液复合车身稳定控制系统及方法
CN109144076B (zh) 一种多车辆横纵向耦合协同控制系统及控制方法
US6909957B2 (en) Method for controlling yaw and transversal dynamics in a road vehicle
CN111645755B (zh) 一种控制方法和装置
JPH05502422A (ja) 自動車の操縦性の改良方法
CN113753080B (zh) 一种自动驾驶汽车横向运动自适应参数控制方法
JP2005014901A (ja) アクティブ法線力調整システムを備えたビークルダイナミクスコントロールシステムおよびその調整方法
CN111002976B (zh) 一种基于模糊自适应pid控制的智能车辆抗侧风控制方法
US20190118858A1 (en) Adaptive wheel base rear steering control
WO2022266824A1 (zh) 一种转向控制方法及装置
CN111142534B (zh) 一种智能车横纵向综合轨迹跟踪方法及控制系统
CN113183950A (zh) 一种电动汽车主动前轮转向的自适应控制方法
JP2008197848A (ja) ファジィ制御装置、レーン走行支援装置、および操舵補助装置
CN113031443A (zh) 一种带有主动安全和自适应预瞄的车辆横向运动控制方法
CN109849898B (zh) 基于遗传算法混合优化gpc的车辆横摆稳定性控制方法
CN114148403B (zh) 一种线控转向系统多工况稳定性控制方法
JP4613668B2 (ja) 車両挙動制御装置および車両挙動制御方法
JP2003231429A (ja) 車両の中心対称面の両側での荷重移動を考慮した横方向力の測定値による車両の軌道へのアクション
Shim et al. Using µ feedforward for vehicle stability enhancement
CN114407880B (zh) 一种无人驾驶紧急避障路径跟踪方法
CN114044003B (zh) 前后双轴转向车辆的循迹控制方法
CN114312847B (zh) 一种自动驾驶车辆的横向控制方法及装置
US20240132152A1 (en) Self-learning collaborative control method for active steering and yaw moment

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22921682

Country of ref document: EP

Kind code of ref document: A1