An intelligent vehicle corner module system and control method for cross-domain cooperative control

Abstract

The application relates to the technical field of automobile chassis control, and provides an intelligent vehicle angle module system and a control method for cross-domain cooperative control, the system is realized based on a hierarchical computing architecture and comprises the following: a parameter estimator used for fusing sensor data and outputting vehicle motion state and road adhesion parameters; a motion prediction controller used for running a nonlinear model prediction control algorithm and calculating generalized force demand of the whole vehicle based on a vehicle dynamics model; a diagnosis fault-tolerant module used for diagnosing the state of an actuator and generating a health state flag bit; an inter-wheel optimization distributor used for constructing and solving a quadratic programming problem with multiple constraints, decomposing the generalized force demand into optimal control instructions of each actuator; and a local controller used for closed-loop execution of the instructions. Through cooperative calculation of the above modules, the method realizes closed-loop control from perception, decision, optimization to execution, and realizes active fault tolerance through online re-optimization of the problem when the actuator fails, effectively solving the problems of multi-system coupling control, excessive model simplification and insufficient functional safety redundancy in the integrated chassis system.

Description

Technical Field

[0001] This application relates to the field of automotive chassis control technology, and in particular to an intelligent vehicle corner module system and control method with cross-domain collaborative control. Background Technology

[0002] Drive-by-wire chassis and corner module technology are key directions for the development of intelligent electric vehicles. They achieve precise wheel-end force vector control by highly integrating drive, braking, steering, and suspension actuators near the wheel. However, this integration also brings a series of technical problems, such as strong coupling control of multiple systems, increased unsprung mass, difficulties in thermal management, loose control architecture, oversimplification of models, single optimization objective, and insufficient functional safety redundancy.

[0003] In some technologies, independent system controllers interact with commands through upper-level coordinators, making it difficult to handle millisecond-level dynamic coupling. Control strategies often rely on simplified linear models, failing to fully consider tire nonlinearity, actuator dynamics, and inter-system coupling effects. Force distribution strategies often focus only on a single performance indicator, failing to comprehensively balance tracking accuracy, stability, ride comfort, economy, and actuator load within a unified optimization framework. Furthermore, actuator failure responses are mostly limited to passive switching, lacking proactive fault-tolerant control based on real-time optimization and redistribution.

[0004] Therefore, there is an urgent need for an intelligent corner module system and its collaborative control method that is reasonably feasible in configuration, has strict theoretical support in control, and has multiple redundancies in safety, so as to fully realize the technical potential of integrated drive-by-wire chassis. Summary of the Invention

[0005] This application provides a cross-domain collaborative control intelligent vehicle corner module system and control method to solve the shortcomings of existing corner module technology in terms of unsprung mass, system coupling, collaborative control and functional safety.

[0006] In a first aspect, this application provides an intelligent vehicle corner module system for cross-domain cooperative control, comprising: A parameter estimator is used to fuse sensor data and output vehicle motion state and road adhesion parameters; A motion prediction controller is used to receive motion commands and calculate generalized force requirements based on a prediction model and the vehicle's motion state. The prediction model is composed of a vehicle nonlinear dynamics model and a tire nonlinear model. The generalized force requirements include the required longitudinal resultant force and the required yaw moment. The diagnostic fault-tolerant module is used to detect the actuator status and generate a health status flag. The wheel-to-wheel optimization allocator is used to receive the generalized force demand and road surface adhesion parameters, and perform optimization under preset constraints to generate control commands for the actuator. The preset constraints include tire adhesion boundaries, actuator physical limits, and health status flags. A local controller is used to drive the corresponding actuator to perform actions according to the control instructions; The diagnostic fault-tolerant module is also used to receive feedback parameters generated by the actuator and regenerate the health status flag to trigger the re-optimization of the inter-wheel optimization allocator.

[0007] Secondly, this application provides a cross-domain cooperative control method for an intelligent vehicle corner module system, comprising: By fusing sensor data through a parameter estimator, the vehicle motion state and road surface adhesion parameters are output. The motion prediction controller receives motion commands and calculates generalized force requirements based on the prediction model and the vehicle's motion state. The prediction model is composed of a vehicle nonlinear dynamics model and a tire nonlinear model. The generalized force requirements include the required longitudinal resultant force and the required yaw moment. The actuator status is detected by the diagnostic fault-tolerant module, and a health status flag is generated. The generalized force demand and road surface adhesion parameters are received by the inter-wheel optimization distributor, and optimization is performed under preset constraints to generate control commands for the actuator. The preset constraints include tire adhesion boundaries, actuator physical limits, and health status flags. The local controller drives the corresponding actuator to perform actions according to the control instructions. The diagnostic fault-tolerant module receives feedback parameters generated by the actuator and regenerates the health status flag to trigger the re-optimization of the inter-wheel optimization allocator.

[0008] As can be seen from the above technical solutions, a cross-domain collaborative control intelligent vehicle corner module system and control method are provided. The system is implemented based on a hierarchical computing architecture and includes: a parameter estimator for fusing sensor data and outputting vehicle motion state and road surface adhesion parameters; a motion prediction controller for running a nonlinear model predictive control algorithm to calculate the generalized force requirements of the entire vehicle based on the vehicle dynamics model; a diagnostic fault-tolerant module for diagnosing actuator states and generating health status flags; an inter-wheel optimization allocator for constructing and solving a quadratic programming problem with multiple constraints, decomposing the generalized force requirements into optimal control commands for each actuator; and a local controller for closed-loop execution of the commands. The method achieves closed-loop control from perception, decision-making, optimization to execution through the collaborative computing of the above modules, and achieves active fault tolerance by online resolving the optimization problem when actuators fail. This effectively solves the problems of multi-system coupled control, excessive model simplification, and insufficient functional safety redundancy in integrated chassis systems. Attached Figure Description

[0009] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 A schematic diagram of the structure of the intelligent vehicle corner module system for cross-domain collaborative control provided in this application embodiment; Figure 2 A schematic diagram of the actuator provided in an embodiment of this application; Figure 3 A schematic diagram of the actuator structure provided in the embodiments of this application; Figure 4 This is a schematic diagram of a braking system provided in an embodiment of this application. Detailed Implementation

[0011] The embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following examples do not represent all embodiments consistent with this application.

[0012] like Figure 1As shown, some embodiments of this application provide a cross-domain collaborative control intelligent vehicle corner module system, including: a full-dimensional state and parameter estimator (FPE), a vehicle motion prediction controller (VMPC), a wheel-to-wheel optimal distributor (WOD), a diagnostic fault-tolerant module, and a local controller and actuators. The actuator structure includes a drive system, a braking system, a steering system, and a suspension system, wherein the actuators are a motor in the drive system, a brake in the braking system, a steering motor in the steering system, and an actuator in the suspension system, respectively.

[0013] The parameter estimator is used to fuse sensor data and output vehicle motion state and road adhesion parameters. Specifically, it receives data from multiple sensors on the vehicle, such as inertial measurement units, wheel speed sensors, and steering angle sensors.

[0014] For example, by integrating sensor data from the vehicle's sensor network inertial measurement unit, wheel speed, motor current / rotation angle, suspension displacement, etc., the vehicle's center of gravity sideslip angle can be estimated in real time. yaw rate longitudinal force of tires Tire lateral force and road surface adhesion coefficient Among them, the vehicle's center of gravity sideslip angle yaw rate longitudinal force of tires Lateral force This is vehicle motion status data.

[0015] Among them, the road surface adhesion parameter is the friction coefficient between the tire and the road surface, which is a physical quantity reflecting the road surface conditions.

[0016] The parameter estimator employs a two-layer cascaded extended Kalman filter. For the vehicle state, the extended Kalman filter (EKF) uses acceleration, angular velocity, wheel speed, and steering angle sensors provided by the inertial measurement unit (IMU) as input to estimate... And wheel angular velocity deviation, based on vehicle dynamics equations, the observation model is directly linked to sensor readings.

[0017] For tire-road parameters, the longitudinal stiffness of each tire is estimated using the vehicle state output and motor torque / electro-mechanical brake (EMB) pressure (as an approximation of tire force observation) as inputs. Lateral stiffness and road surface friction coefficient A loosely coupled design is adopted to reduce computational complexity.

[0018] The motion predictive controller receives motion commands from the vehicle's advanced driver assistance system or the driver. Based on the preset path, vehicle speed planning, and vehicle motion state provided by the parameter estimator, it uses a nonlinear model predictive control to calculate the desired generalized force / torque of the whole vehicle with the optimization goal of vehicle motion tracking accuracy and stability.

[0019] Specifically, the motion prediction controller receives motion commands and calculates generalized force requirements based on the prediction model and the vehicle's motion state. The prediction model is composed of a vehicle nonlinear dynamics model and a tire nonlinear model. The generalized force requirements include the required longitudinal resultant force and the required yaw moment.

[0020] In some embodiments, the vehicle nonlinear dynamics model is defined by state equations, the state variables of which are given by the following equations: ; in, The longitudinal velocity of the vehicle's center of gravity. The lateral velocity of the vehicle's center of gravity. Let yaw rate be the vehicle's angular velocity. For the first The angular velocity of each wheel; The state equations are defined by the following dynamic relations: The following formula represents the longitudinal motion of the vehicle's center of gravity: ; in, For the overall vehicle quality, For tire force, For the first Each wheel turns, Calculated from the tire model, To create vertical synergy in demand; The following formula is for the lateral motion of the vehicle's center of gravity: ; in, The estimated lateral resultant force of the entire vehicle in the current state; The following formula is for the movement of the vehicle's swing arm: ; in, For the moment of inertia of yaw rotation, The distance from the center of mass to the front and rear axles. The wheelbase is the distance between the wheels. To meet the yaw moment requirement; The following formula is for the yaw motion of a vehicle: ; in, Let be the moment of inertia of the wheel. For the first Motor torque of each wheel The effective rolling radius of the wheel, For the first The frictional braking torque of each wheel; The tire force nonlinear model is a nonlinear brush model used to calculate the tire longitudinal force and tire lateral force. The nonlinear brush model calculates the total tangential force of the tire based on the comprehensive slip ratio. The calculation of the total tangential force satisfies the following formula: exist In this case, ; exist In this case, ; in, For the overall slip ratio, , For longitudinal slip ratio, The lateral slip ratio, For the total tangential force, The coefficient of friction of the road surface. For vertical loads, , This refers to the longitudinal and lateral stiffness of the tire.

[0021] The tire longitudinal force and tire lateral force are proportionally distributed based on the total tangential force and the overall slip ratio, and the tire longitudinal force is calculated according to the following formula: ; The lateral force of the tire is calculated according to the following formula: ; Longitudinal slip ratio Calculate according to the following formula: ; Lateral slip ratio Calculate according to the following formula: ; in, This represents the component of the wheel center velocity in the tire coordinate system.

[0022] In some embodiments, the motion predictive controller is configured to run a nonlinear model predictive control algorithm, the control algorithm comprising: taking the vehicle motion state as the current state, the process being executed in each control cycle, the core of which is to solve a rolling optimization problem in a finite time domain to calculate the optimal generalized force requirement.

[0023] Specifically, in each control cycle The motion predictive controller solves the following finite-time optimization problem. First, the latest vehicle motion state obtained from the parameter estimator is used as the initial state of the optimization problem. Then, an optimization problem defined in the prediction time domain is constructed. The mathematical description of this optimization problem is as follows: The optimization objective (cost function) is minimized as shown in the following formula: ; in, To predict the length of the time domain, For the desired yaw rate, The sideslip angle is the angle of the centroid. To control the weight of the input rate of change, This is the weight matrix. For the desired longitudinal force, and As the safety boundary for the centroid sideslip angle, Let be the integral variable. The optimization process must satisfy the following constraints: 1. Vehicle dynamics model constraints (state equations), see the following formula: ; in, Determined jointly by the vehicle nonlinear dynamics model and the tire nonlinear model; 2. State safety constraint: This constraint limits the centroid sideslip angle within the stable safety boundary, as shown in the following formula: ; 3. Control input constraints: These constraints limit the range of values ​​for the generalized force requirement to conform to the actual physical conditions of the vehicle. See the following formula: The optimal control sequence is obtained by solving the above problem using numerical optimization methods. and take it in The value at time is sent as the current instruction to the inter-round optimizer.

[0024] The motion predictive controller uses a model to perform rolling optimization calculations within a limited time frame to find a series of control actions that enable the vehicle's future state to optimally track the target command. This series of control actions is summarized into two key physical quantities at the vehicle level: the required longitudinal resultant force and the required yaw moment.

[0025] Among them, the longitudinal force of demand determines whether the vehicle should accelerate or decelerate, and the yaw moment of demand determines how the vehicle should steer.

[0026] The wheel-to-wheel optimization allocator receives the generalized force demand and road surface adhesion parameters, performs optimization under the condition of considering preset constraints, and generates control commands for the actuator. The preset constraints include tire adhesion boundaries, actuator physical limits, and health status flags.

[0027] The motion prediction controller outputs the generalized force demand at the vehicle level. The wheel-to-wheel optimization distributor decomposes the vehicle demand into 16 independent control quantities for the four corner modules, which are the specific instructions for each tire actuator. These include: the drive of the four motors, regenerative braking torque, etc. Four friction braking forces (When EMB is activated), four active steering angles And four suspensions as power .

[0028] However, actuators have capability limits, and there are physical boundaries to the adhesion between the tires and the ground. Furthermore, actuators may malfunction or experience performance degradation. The inter-wheel optimization allocator is used to perform optimal resource allocation under these conditions.

[0029] The wheel-to-wheel optimization allocator receives the generalized force demand from the motion prediction controller as the total task that must be completed. At the same time, it receives road adhesion parameters from the parameter estimator to calculate the maximum force that each tire can provide on the current road surface without slipping, i.e., the tire adhesion boundary.

[0030] The system also includes a diagnostic fault-tolerant module for detecting actuator status to generate a health status flag. The inter-wheel optimization allocator also receives the health status flag, which dynamically indicates whether each actuator is currently available and whether its performance is good.

[0031] The diagnostic fault-tolerant module incorporates model- and signal analysis-based fault diagnosis. For motor / EMB faults, it compares the commanded torque / force with the actual output force estimated based on current / pressure, and combines this with temperature signals to determine demagnetization, overheating, or mechanical jamming. For steering faults, it monitors the steering motor current and steering angle tracking error. For suspension faults, it compares the commanded power and actuator feedback.

[0032] Once a fault is diagnosed, the diagnostic fault-tolerant module immediately updates the actuator health status flag. This triggers the reconfiguration of WOD constraints. If necessary, the control objectives of VMPC are adjusted, such as reducing the expected performance of yaw moment tracking in the event of steering redundancy failure, and prioritizing path tracking safety.

[0033] In some embodiments, the wheel-to-wheel optimization allocator performs optimization by constructing and solving a quadratic programming problem; the quadratic programming problem includes constraints and an objective function; the optimization variables of the quadratic programming problem are the control commands of each actuator; the constraints of the quadratic programming problem include: equality constraints for achieving the generalized force requirement, tire adhesion boundary inequality constraints determined based on the road adhesion parameters and tire load, and inequality constraints on the capabilities of each actuator determined based on the health status flag; the objective function is configured to simultaneously minimize tracking error and control cost.

[0034] Solving quadratic programming problems can make control actions as smooth and efficient as possible while satisfying the generalized force requirements. The process of solving quadratic programming problems is subject to multiple constraints, including: the force of each tire cannot exceed the adhesion boundary, the output of each actuator cannot exceed its physical limit, and the output capacity limit of actuators marked as faulty or degraded will be reduced or zeroed accordingly.

[0035] By solving the quadratic programming problem online, the wheel-to-wheel optimization allocator generates a set of specific control instructions assigned to each actuator to ensure that the overall vehicle objectives are achieved in the most reasonable way under various real-world constraints. Furthermore, when the capabilities of some actuators decline, the task is automatically reassigned to other healthy actuators.

[0036] To meet real-time solution requirements, the wheel-to-wheel optimization allocator applies a reasonable linear approximation to the system model. Specifically, within a single control cycle, assuming smooth changes in vehicle state, a linear tire model is used to approximate the relationship between tire force and vehicle motion. That is, the tire lateral force and the sideslip angle approximately satisfy a linear relationship: ; ; in, The real-time equivalent lateral stiffness of the i-th tire is provided by the parameter estimator. This refers to the tire slip angle.

[0037] Based on the above linearization assumptions, the contribution of each tire force to the generalized force of the vehicle can be linearly mapped through a time-varying control efficiency matrix. The construction of this matrix depends on the current wheel steering angle and information such as tire side stiffness provided by the parameter estimator. Therefore, the nonlinear coupling relationship between the generalized force demand and tire forces is transformed into a linear equality constraint: ; in, This refers to the longitudinal force component in the time-varying control efficiency matrix B. For the lateral force component of the time-varying control efficiency matrix B, The tire slip angle is estimated by FPE. For the first The current turning angle of the wheel.

[0038] This equation constraint, as the core constraint of the quadratic programming problem, ensures that the allocation result must follow the instructions from the higher level.

[0039] By solving this quadratic programming problem online, the inter-wheel optimization allocator generates a set of specific control commands assigned to each actuator (drive motor, brake, steering motor, suspension actuator). When an actuator's health status flag degrades due to a fault, its capability constraints in the optimization problem are automatically tightened. The system then automatically reassigns the control task to other healthy actuators through optimization calculations, thereby achieving active fault-tolerant control.

[0040] To facilitate online optimization, the lateral force contribution term is treated as a known value from the previous time step, thus linearizing the control efficiency matrix. Simultaneously, considering that the steering angle itself is also a control variable, the steering angle increment is... It is also used as an optimization variable to optimize the steering response.

[0041] The output of the motion predictive controller is a vector containing all the instructions from the underlying actuators. To meet the requirements of real-time solving, the system model needs to be reasonably simplified. A complete control input vector v is defined, containing all independent control instructions from the four corner modules, totaling sixteen variables: ; in, For motor drive or regenerative braking torque, For friction braking force, This is the steering angle increment. It provides power to the suspension.

[0042] The core of the optimization problem is to establish the mathematical relationship between the generalized force demand and the control input vector v. Assuming small steering angles and minimal state changes within a single control cycle, the forces of each tire are approximated using a linear model. The longitudinal tire force is determined by the balance between the motor torque and the braking torque, while the lateral tire force is approximated using a linear model. Based on this, a time-varying control efficiency matrix B and a constant disturbance term o are constructed, forming the equality constraints for the quadratic programming problem, as shown in the following equation: ; in, This is the time-varying control efficiency matrix. This is a constant term for the constant perturbation determined by the current state.

[0043] This equation constraint mandates that the allocation result must accurately meet the vehicle motion requirements of the motion prediction controller.

[0044] The optimization solution must be performed under multiple physical and logical constraints, which are expressed as linear inequalities.

[0045] First, the resultant force of each tire is constrained by the road adhesion limit. Linearizing the nonlinear adhesion ellipse using an inscribed regular octagon yields a set of linear inequalities: ; in, , For the first Estimated vertical load of each tire.

[0046] For the tire attachment elliptical constraint, the resultant force of each tire must not exceed the boundary determined by the current vertical load and the friction coefficient. An inscribed octagonal linearization approximation is used to transform the nonlinear elliptical constraint into a set of linear inequalities. Furthermore, each actuator has its own physical capability limits and operating range. ; ; ; ; ; in, and The motor torque output limit is the motor temperature. The function, This is the maximum clamping force for the EMB. This is the braking system health status flag, a scalar value between 0 and 1. This indicates that the EMB is completely healthy; This indicates a partial degradation in performance (e.g., due to overheating, the maximum force is reduced). This indicates that the EMB has completely failed. This flag is updated in real time by the diagnostic fault-tolerant module. , To limit the rate of increase in steering angle, , Mechanical limit for the total steering angle of the wheel. , This refers to the range of action of the suspension actuator.

[0047] Then, a braking energy recovery priority strategy is set. Let the total required braking force be... (When negative), regenerative braking from the motor is given priority, defining the regenerative braking capacity coefficient. At this point, the clamping force ,in, This indicates that a positive value is being used. This constraint ensures that friction braking is only activated when regenerative braking is insufficient.

[0048] Reintegrate fault tolerance mechanisms. This is achieved through health status flags of the drive, braking, and steering subsystems. Modify the upper and lower limits of the corresponding actuator. For example... (EMB failure), then... and adjust To compensate as much as possible by other healthy braking units.

[0049] Regarding the constraints on health status flag bits, in addition to ,definition This is a health status flag for the drive system. It represents the... The health status of each motor (1 for healthy, 0 for unhealthy). If Then let in the constraints .

[0050] This is a health status flag for the steering system, representing the first... The health of each steering actuator, if Then let in the constraints This triggers a higher-level vehicle steering fault-tolerance strategy (using differential braking to compensate for steering failure). The standard form of the quadratic programming problem solved by WOD is as follows: ; ; ; in, It involves finding the minimum value of the objective function and the corresponding value v. As an equality constraint, corresponding to the aforementioned , The inequality constraints combine all the linear inequalities mentioned above. v represents the control matrix and the weight matrix. sum vector The design of these is crucial, as they encode multiple competing objectives: The main objective (force tracking) is achieved through equality constraints. Achieve. Secondary objective (minimize control cost) in Set diagonal weights in the middle and minimize them. (Reduce friction loss) (To ensure smooth steering) and the deviation of motor torque from the high-efficiency operating range.

[0051] Tire load equalization Negative values ​​are assigned to off-diagonal elements (use with caution), or the vertical load estimation weights are adjusted iteratively to encourage force distribution to tires with large adhesion margins.

[0052] Thermal management and energy consumption based on real-time temperature The weight of the corresponding actuator command is dynamically increased, so that it tends to reduce output while meeting the main objectives, thereby achieving thermal load balance.

[0053] For an adaptive weight adjustment strategy, the weight matrix and V MPC In It's not fixed; it's estimated based on driving mode and road surface adhesion. and urgency (Relative time to obstacles) is adjusted online, see the following formula: ; ; ; in, It directly affects the weight of the yaw rate tracking error in VMPC and the penalty term for tire utilization in WOD; Affecting the suspension's dynamics The optimization objectives are the weights and the vertical acceleration of the vehicle body; The weights affecting the priority of regenerative braking. to These are calibration parameters.

[0054] By solving the above optimization problem online, the wheel-to-wheel optimization allocator generates a set of specific control instructions assigned to each actuator. The local controller is used to drive the corresponding actuator to perform actions according to the control instructions. For example, the drive motor local controller controls the motor to output a specific torque, and the steering motor local controller controls the wheel to turn to a specific angle.

[0055] The diagnostic fault-tolerant module is also used to receive feedback parameters from the actuator and regenerate the health status flag to trigger the re-optimization of the wheel-to-wheel optimizer. For example, the actual output force of the motor is read by a current sensor, and the actual clamping force of the brake is read by a pressure sensor. The diagnostic fault-tolerant module updates the corresponding health status flag, and the updated health status flag is sent to the wheel-to-wheel optimizer in real time.

[0056] When the inter-wheel optimization allocator performs optimization calculations in the next control cycle, it uses new constraints that reflect the actuator failure state. This causes the solution to the optimization problem, i.e., the generated control commands, to change automatically. Tasks are no longer assigned to the failed actuator, or tasks are assigned to other actuators for compensation. This process is proactive and based on online optimization and reallocation. It does not require switching to a separate backup controller; instead, within the existing optimization framework, the control strategy is naturally reconstructed and tasks are redistributed by updating the constraints.

[0057] like Figure 2 , 3 As shown, in some embodiments, the system further includes a drive system, a braking system, a steering system, and a suspension system; the actuators include a motor in the drive system, a brake in the braking system, a steering motor in the steering system, and an actuator in the suspension system.

[0058] The drive system is used to provide drive torque and includes a motor as the core power output actuator. The motor is an axial-radial hybrid flux permanent magnet synchronous hub motor, which is directly integrated inside or near the wheel.

[0059] In some embodiments, the motor of the drive system is a hub motor, the stator of the hub motor includes a first part coupled to the axial magnetic circuit and a second part coupled to the radial magnetic circuit, and the rotor of the hub motor is a double-sided symmetrical permanent magnet structure; The hub motor has a composite magnetic circuit structure, which includes a radial flux switching motor module and an axial flux switching motor module. The rotor of the radial flux switching motor module and the rotor of the axial flux switching motor module are connected by a central clutch. The central clutch is configured to engage under a first preset operating condition to jointly output power through the radial flux switching motor module and the axial flux switching motor module, and to disengage under a second preset operating condition to output power through the axial flux switching motor module to provide driving torque.

[0060] Specifically, the stator is divided into upper and lower parts, which are coupled to the axial magnetic circuit and the radial magnetic circuit respectively; the rotor is a double-sided symmetrical permanent magnet structure, forming a composite magnetic circuit.

[0061] The radial flux switching motor module is located on the outside. Utilizing its flux switching principle and radial magnetic field structure, it provides extremely high torque density and strong overload capacity in the low-speed range, meeting the vehicle's starting, acceleration, and hill-climbing requirements. The axial flux switching motor module is located on the inside. Taking advantage of the flat and high power density of the axial flux structure, it operates efficiently in the high-speed range, achieving constant power expansion over a wide speed range.

[0062] The central clutch connects the rotors of the two modules. The central clutch is a device that can be controlled to engage or disengage. It is used to connect or disconnect the power transmission path between two rotating parts. In the low-speed, high-torque condition, i.e. the first preset condition, the clutch engages and both modules work together. In the high-speed cruising condition, i.e. the second preset condition, the radial flux switching motor module can be disengaged, and only the axial flux switching motor module is driven, reducing iron loss and wind friction loss and improving efficiency. This configuration solves the contradiction between torque and power range from a physical perspective.

[0063] The braking system is a system that provides deceleration force to a vehicle. This system employs an architecture that incorporates multiple braking methods working in tandem.

[0064] In some embodiments, the braking system includes a drive motor, a first brake, and a second brake; The drive motor is used for regenerative braking, which is achieved through the motor in the drive system. When the motor operates as a generator, it can convert kinetic energy into electrical energy.

[0065] The first brake uses a ball screw transmission mechanism to generate frictional braking force. The first brake is a device that generates frictional braking force through an electrically driven mechanical mechanism. It uses a ball screw transmission mechanism to convert rotational motion into linear clamping force.

[0066] The second brake is an electromechanical hydraulic brake, which is arranged in parallel with the ball screw transmission mechanism. It includes a hydraulic piston chamber and a hydraulic source. The hydraulic piston chamber is a closed cavity that can generate linear motion by pushing the piston with hydraulic pressure. The independent hydraulic source is a hydraulic pressure supply device that is independent of the traditional brake master cylinder.

[0067] The second brake is configured to, when the first brake fails, connect the hydraulic source and hydraulically push the piston chamber to generate braking force.

[0068] like Figure 4 As shown, specifically, the braking system is a three-stage hybrid architecture consisting of regenerative braking, electromechanical braking (EMB), i.e., the first brake, and hydraulic backup (hEMB), i.e., the second brake.

[0069] The first-stage regenerative braking, directly implemented by the drive motor, is the primary means of daily braking and energy recovery, and has the highest priority.

[0070] The second-stage pure EMB uses a ball screw-type electromechanical brake with high dynamic response, which quickly intervenes when the regenerative braking force is insufficient or fails, providing precise friction braking force.

[0071] The third-stage hybrid EMB (hEMB) incorporates a hydraulic piston chamber connected in parallel with the motor-screw drive path of the second-stage EMB. Under normal circumstances, the hydraulic chamber is closed; however, in the event of a motor or circuit failure in the EMB, the system can connect an independent hydraulic source to drive the piston and generate clamping force, serving as a mechanical backup.

[0072] The steering system controls the direction of a vehicle's movement. This system employs a steer-by-wire kingpin steering mechanism. Steer-by-wire kingpin steering means that steering commands are transmitted via electrical signals, eliminating the direct mechanical connection between the steering wheel and the wheels; the steering actuator directly drives the steering knuckle to rotate around the kingpin axis.

[0073] In some embodiments, the steering system employs a steer-by-wire kingpin steering structure and integrates a multi-level redundant safety mechanism; The multi-level redundancy safety mechanism includes an electrical redundancy mechanism and a mechanical backup; the electrical redundancy mechanism includes at least two steering motors connected in parallel via differential gears; the mechanical backup includes a backup driveshaft engaged by an electromagnetic clutch, the backup driveshaft being configured to transmit manual steering torque to the wheels in the event of electrical system failure.

[0074] Specifically, the steering system adopts a steer-by-wire kingpin steering structure, eliminating traditional steering tie rods, steering knuckles, and other components. The core of the steering system is the steering module and steering arm. The steering module integrates a steering motor and a reduction mechanism. The steering module directly drives the steering angle module to rotate relative to the vehicle frame through the steering arm, thereby steering the wheels.

[0075] The kingpin axis is designed to be adjustable. Through the variable kingpin inclination system, the kingpin inclination angle can be adjusted within a certain range, and the wheel camber angle can be adjusted simultaneously to achieve decoupled adjustment of kingpin inclination and wheel camber.

[0076] The steering system supports a 360-degree steering angle, enabling the vehicle to achieve a variety of special driving modes. The steering control system automatically calculates the optimal steering angle for each wheel based on the driving mode.

[0077] The steering system adopts a dual-motor redundancy design. The two motors are connected in parallel through differential gears and work together under normal circumstances. Even if one motor fails, it can still provide steering capability. It also integrates a mechanical backup mechanism. When the electrical system fails completely, it switches to manual steering mode through an electromagnetic clutch. At this time, the steering torque is transmitted to the steering wheel through the backup drive shaft.

[0078] The steering wheel module uses active force feedback technology to generate appropriate steering torque based on vehicle status, road conditions and driving mode. The road feel simulation algorithm is based on the vehicle dynamics model and takes into account factors such as lateral acceleration, tire slip angle and vehicle speed to provide the driver with a realistic and safe steering feel. In autonomous driving mode, the steering wheel can automatically retract to free up interior space.

[0079] The suspension system is a system that connects the vehicle body to the wheels and manages the vehicle's attitude and vibration. This system uses a design that combines active suspension with variable geometry.

[0080] like Figure 4 As shown, in some embodiments, the suspension system includes an air spring unit, an electric damper, and a multi-link geometry adjustment mechanism; The air spring unit is configured to adjust the spring stiffness and chassis height by adjusting the internal air pressure; the electric damper is configured to provide damping force; the multi-link geometry adjustment mechanism includes multiple motor-driven links that adjust the camber and toe angles of the wheels by changing the link length or connection point position. The suspension system also includes a control system configured to calculate actuation commands for the air spring unit, electric damper, and multi-link geometry adjustment mechanism based on predicted road inputs and vehicle conditions, in order to coordinately adjust vehicle attitude and vibration level.

[0081] Specifically, the electric damper can efficiently reduce vibration and collect vibration energy. The geometric adjustment mechanism consists of adjustable-length links driven by multiple motors, such as upper and lower forks. By changing the link length and connection point position, the wheel alignment parameters (camber, toe, etc.) can be adjusted in real time.

[0082] The suspension control system integrates a road preview function, using a forward-looking camera and LiDAR to identify road surface unevenness in advance and pre-adjust suspension parameters. Based on the predicted road input and vehicle state, the control system uses an optimal control algorithm to calculate the optimal output of each suspension actuator, significantly reducing vehicle body vibration acceleration. Furthermore, the system can adjust damping and actuation force in real time based on suspension travel signals, achieving optimized control of vehicle attitude and vibration.

[0083] The suspension system offers several preset modes, such as Comfort, Sport, Off-road, and Load. In Comfort mode, the suspension focuses on isolating road vibrations for a smooth ride; in Sport mode, suspension stiffness increases and damping is enhanced for precise handling; in Off-road mode, the chassis height is raised and suspension travel is increased; in Load mode, suspension parameters are automatically adjusted according to the load to maintain vehicle level. The system can also learn driver preferences and automatically adjust mode parameters.

[0084] To enhance safety, the hEHB hydraulic backup automatically takes over in the event of EMB failure; in the event of a total electrical failure, the mechanical redundancy mechanism can directly drive the hydraulic piston via pedal force. If one winding of the dual-winding motor fails, the other winding can maintain 70% steering capability; in the event of a total failure, the system switches to the backup driveshaft via a clutch. Communication is based on Ethernet and supports a hybrid scheduling of time-triggered and event-triggered commands to ensure the real-time nature of critical instructions.

[0085] The implementation methods of this application will be described in detail below in conjunction with the operation flow of the control system.

[0086] The coordinated control under normal high-speed cruise conditions follows the following procedure: (1) System initialization: The vehicle is powered on, all sensors and actuators pass self-tests, the FPE starts working, and the VMPC and WOD weights are set to comfort mode, i.e. High, medium, medium.

[0087] (2) State estimation: Based on signals such as IMU and wheel speed, FPE estimates whether the vehicle is currently in a straight-line cruising state. Good road surface adhesion ( Health status .

[0088] (3) Vehicle motion control: VMPC receives cruise control commands ( Since the state is stable, the NMPC optimization outputs almost zero generalized force instructions to maintain the current state.

[0089] (4) Optimal allocation between rounds: WOD receives a generalized force command that is close to zero. In its cost function, the energy recovery weight is... It takes effect, but regenerative braking is not activated because the required braking force is zero. Steering angle increment weighting prevents unnecessary minor steering adjustments. Distribution result: Each motor outputs a balanced, small drive torque to maintain vehicle speed; each EMB braking force is zero; the steering angle remains neutral; the suspension performs fine-tuning with the goal of optimizing ride comfort.

[0090] (5) Execution and closed loop: Each actuator executes the ALC command, and the vehicle cruises smoothly. FPE continuously estimates the state, and the system periodically (e.g., every 10ms) repeats steps (3)-(5).

[0091] The coordinated control in emergency double-lane-shift obstacle avoidance conditions follows the procedure below: (1) Triggering and Perception: The forward sensor detects a sudden obstacle, the planning layer generates an emergency double-track path, provides the desired lateral acceleration and yaw rate sequence, and then... The parameter is marked as high urgency.

[0092] (2) Adaptive adjustment: Automatic switching of system weights: Rapidly increase ( (Project Leading) reduce, The FPE estimate decreases as the vehicle enters a high-dynamic state.

[0093] (3) Vehicle motion control: VMPC’s NMPC tracks the rapidly changing desired yaw rate with high weight. While maintaining stable longitudinal vehicle speed, after real-time optimization, a set of time-varying, relatively large values ​​is output. and appropriate (Possibly negative, to moderately reduce speed).

[0094] (4) Optimal allocation between rounds: WOD receives a large value Its solver operates under the following key constraints: Attachment constraints: based on real-time estimation. and Calculate the maximum available resultant force for each tire. Actuator constraints: The motor is allowed to generate maximum torque, and the EMB is ready to intervene. Objective function: Satisfy the generalized force equation constraints with extremely high priority (stability first).

[0095] The possible allocation result is: a positive drive torque is commanded to the left front wheel. and slight positive steering angle A strong regenerative braking torque is commanded to the right rear wheel. And in conjunction with EMB braking force The right front wheel and left rear wheel are coordinated and adjusted. This distribution strategy utilizes both drive / brake torque vector control and active steering to collaboratively generate the required yaw moment.

[0096] (5) Example of fault-tolerant operation: Suppose that during obstacle avoidance, the right rear wheel EMB reports a performance degradation due to overheating ( corresponding (Downgraded). The diagnostic fault-tolerant module response indicates that the braking constraints of that wheel have been updated when the WOD is solved in the next control cycle. The optimization result automatically redistributes some braking force demand to the left rear wheel or other wheels' regenerative braking, while potentially fine-tuning the steering angle command to compensate for slight changes in yaw moment generation, thus maintaining obstacle avoidance stability with almost imperceptible accuracy.

[0097] The following procedure applies to low-adhesion cornering braking energy recovery: (1) Condition recognition: FPE identifies the entry into a low-adhesion road surface by using the relationship between tire stiffness and slip ratio. The vehicle is in a stable cornering state.

[0098] (2) Weight adjustment: because Decrease and automatically increase ( item), Reduce appropriately.

[0099] (3) Vehicle motion control: The driver requests braking. VMPC calculates the required... However, in its nonlinear model predictive control algorithm, the centroid sideslip angle is enhanced. Constraints are imposed to prevent it from exceeding the stability threshold.

[0100] (4) Optimal Wheel-to-Wheel Braking: Solving the WOD problem faces stringent adhesion constraints. The strategy is as follows: Braking Distribution: Prioritize regenerative braking on non-drive wheels (or wheels with less load) (if their motors are available), as excessive regenerative braking force on drive wheels can easily cause drive slip. Friction braking is used as a supplement, but the braking force of each wheel is strictly limited within the adhesion ellipse. Steering Compensation: Distribute small steering angle compensations to counteract yaw interference that may be caused by uneven braking force distribution. Suspension Assist: Adjust the stiffness of the inner and outer suspensions to optimize load transfer and maximize the available adhesion of the inner tires in the curve.

[0101] While prioritizing low-friction cornering braking stability, the system intelligently selects available energy recovery methods, achieving the best trade-off between safety and energy efficiency.

[0102] In summary, the corner module system provided in this application can improve ride comfort and handling. The suspension system effectively suppresses unsprung resonance through inertial mass adjustment, significantly improving tire contact with the ground without increasing actual weight. The decoupled thrust rod design reduces longitudinal impact transmission, further improving ride comfort and enhancing overall coordination and extreme stability. Based on the predictive-distribution integrated architecture of model predictive control, it achieves unified optimization of the four major systems of drive, braking, steering, and suspension, fundamentally avoiding control target conflicts and ensuring the stability and controllability of the vehicle under extreme conditions.

[0103] Furthermore, the control theory is advanced and engineering-feasible, possessing high precision and adaptability to all operating conditions. A predictive controller is constructed based on a high-fidelity nonlinear vehicle model and a tire model, and a quadratic programming approach considering tire adhesion boundaries and actuator dynamics is used for real-time optimal allocation. This theoretically ensures the accuracy and robustness of the control strategy under all operating conditions, enabling real-time optimization and engineering implementation. Through reasonable model simplification (such as tire constraint linearization) and the embedding of efficient solution algorithms (such as the interior point method), the entire advanced control algorithm can be stably operated on the vehicle's high-performance domain controller, combining theoretical advancement with engineering practicality.

[0104] The three-layer coaxial compact layout greatly optimizes the internal space utilization of the corner modules. At the same time, the use of lightweight materials such as carbon fiber connecting rods effectively reduces unsprung mass, resulting in a lower overall system weight compared to traditional distributed solutions.

[0105] With comprehensive safety redundancy and fault tolerance capabilities, a system-level safety architecture with triple redundancy across electrical, hydraulic, and mechanical systems is constructed, meeting the highest level of functional safety requirements. It integrates proactive fault diagnosis and fault-tolerant control logic based on health state estimation. When some actuators fail, the system can achieve graceful functional degradation through rapid reconfiguration and optimized reallocation of control laws, ensuring core safety.

[0106] By prioritizing braking energy recovery and combining it with suspension vibration energy recovery technology, the efficient conversion and reuse of mechanical energy is achieved, thereby improving the overall energy efficiency of the system.

[0107] Based on the aforementioned cross-domain cooperative control intelligent vehicle corner module system, some embodiments of this application also provide a control method for the cross-domain cooperative control intelligent vehicle corner module system, including: By fusing sensor data through a parameter estimator, the vehicle motion state and road surface adhesion parameters are output. The motion prediction controller receives motion commands and calculates generalized force requirements based on the prediction model and the vehicle's motion state. The prediction model is composed of a vehicle nonlinear dynamics model and a tire nonlinear model. The generalized force requirements include the required longitudinal resultant force and the required yaw moment. The actuator status is detected by the diagnostic fault-tolerant module, and a health status flag is generated. The generalized force demand and road adhesion parameters are received by the inter-wheel optimization distributor. Optimization is performed under the condition of considering preset constraints to generate control commands for the actuator. The preset constraints include tire adhesion boundaries, actuator physical limits, and health status flags. The local controller drives the corresponding actuator to perform actions according to the control instructions. The diagnostic fault-tolerant module receives feedback parameters from the actuator and regenerates the health status flag to trigger the re-optimization of the inter-wheel optimization allocator.

[0108] Similar parts between the embodiments provided in this application can be referred to mutually. The specific implementation methods provided above are only a few examples under the overall concept of this application and do not constitute a limitation on the scope of protection of this application. For those skilled in the art, any other implementation methods extended from the solution of this application without creative effort shall fall within the scope of protection of this application.

Claims

1. A cross-domain collaborative control intelligent vehicle corner module system, characterized in that, include: A parameter estimator is used to fuse sensor data and output vehicle motion state and road adhesion parameters; A motion prediction controller is used to receive motion commands and calculate generalized force requirements based on a prediction model and the vehicle's motion state. The prediction model is composed of a vehicle nonlinear dynamics model and a tire nonlinear model. The generalized force requirements include the required longitudinal resultant force and the required yaw moment. The diagnostic fault-tolerant module is used to detect the actuator status and generate a health status flag. The wheel-to-wheel optimization allocator is used to receive the generalized force demand and road surface adhesion parameters, and perform optimization under preset constraints to generate control commands for the actuator. The preset constraints include tire adhesion boundaries, actuator physical limits, and health status flags. A local controller is used to drive the corresponding actuator to perform actions according to the control instructions; The diagnostic fault-tolerant module is also used to receive feedback parameters generated by the actuator and regenerate the health status flag bit to trigger the re-optimization of the inter-wheel optimization allocator; The motion predictive controller is configured to run a nonlinear model predictive control algorithm, the control algorithm comprising: taking the vehicle motion state as the current state, and performing rolling optimization in a finite time domain based on the predictive model and the vehicle motion state to calculate the generalized force demand; The inter-round optimization allocator performs optimization by constructing and solving a quadratic programming problem; the quadratic programming problem includes constraints and an objective function. The optimization variables of the quadratic programming problem are the control commands of each actuator; The constraints of the quadratic programming problem include: equality constraints for achieving the generalized force requirement, tire adhesion boundary inequality constraints determined based on the road surface adhesion parameters and tire load, and inequality constraints on the capabilities of each actuator determined based on the health status flag. The objective function is configured to simultaneously minimize tracking error and control cost.

2. The intelligent vehicle corner module system for cross-domain collaborative control according to claim 1, characterized in that, The nonlinear dynamics model of the vehicle is defined by a state equation, and the state variables of the state equation are shown in the following equation: ； in, The longitudinal velocity of the vehicle's center of gravity. The lateral velocity of the vehicle's center of gravity. Let yaw rate be the vehicle's angular velocity. For the first The angular velocity of each wheel; The state equations are defined by the following dynamic relations: The following formula represents the longitudinal motion of the vehicle's center of gravity: ； in, For the overall vehicle quality, For tire force, For the first Each wheel turns, Calculated from the tire model, To create vertical synergy in demand; The following formula is for the lateral motion of the vehicle's center of gravity: ； in, The estimated lateral resultant force of the entire vehicle in the current state; The following formula is the equation for the movement of the vehicle's swing arm: ； in, For the moment of inertia of yaw rotation, The distance from the center of mass to the front and rear axles. The wheelbase is the distance between the wheels. To meet the yaw moment requirement; The following formula is for the yaw motion of a vehicle: ； in, Let be the moment of inertia of the wheel. For the first Motor torque of each wheel The effective rolling radius of the wheel, For the first The frictional braking torque of each wheel; The tire force nonlinear model is a nonlinear brush model used to calculate the tire longitudinal force and tire lateral force. The nonlinear brush model calculates the total tangential force of the tire based on the comprehensive slip ratio. The tire longitudinal force and tire lateral force are proportionally distributed according to the total tangential force and the comprehensive slip ratio. The calculation of the total tangential force satisfies the following formula: exist In this case, ; exist In this case, ; in, For the overall slip ratio, , For longitudinal slip ratio, The lateral slip ratio, For the total tangential force, The coefficient of friction of the road surface. For vertical loads, , This refers to the longitudinal and lateral stiffness of the tire.

3. The intelligent vehicle corner module system for cross-domain collaborative control according to claim 1, characterized in that, The system also includes a drive system, a braking system, a steering system, and a suspension system; The drive system is used to provide drive torque, the braking system is used to generate braking force and regenerative braking, the steering system is used to provide steering torque to change wheel angle, and the suspension system is used to provide vertical force to adjust vehicle attitude and wheel alignment parameters. The actuators include a motor in the drive system, a brake in the braking system, a steering motor in the steering system, and an actuator in the suspension system.

4. The intelligent vehicle corner module system for cross-domain collaborative control according to claim 3, characterized in that, The motor of the drive system is a hub motor. The stator of the hub motor includes a first part coupled to the axial magnetic circuit and a second part coupled to the radial magnetic circuit. The rotor of the hub motor is a double-sided symmetrical permanent magnet structure. The hub motor has a composite magnetic circuit structure, which includes a radial flux switching motor module and an axial flux switching motor module. The rotor of the radial flux switching motor module and the rotor of the axial flux switching motor module are connected by a central clutch. The central clutch is configured to engage under a first preset operating condition to jointly output power through the radial flux switching motor module and the axial flux switching motor module, and to disengage under a second preset operating condition to output power through the axial flux switching motor module to provide driving torque.

5. The intelligent vehicle corner module system for cross-domain collaborative control according to claim 3, characterized in that, The braking system includes a first brake and a second brake; the braking system also includes a drive motor. The drive motor is used for regenerative braking; The first brake uses a ball screw transmission mechanism to generate frictional braking force; The second brake is a hybrid electromechanical hydraulic brake, which is connected in parallel with the ball screw transmission mechanism and includes a hydraulic piston chamber and a hydraulic source; The second brake is configured to, when the first brake fails, connect the hydraulic source and hydraulically push the hydraulic piston chamber to generate braking force.

6. The intelligent vehicle corner module system for cross-domain collaborative control according to claim 3, characterized in that, The steering system adopts a steer-by-wire kingpin steering structure and integrates a multi-level redundant safety mechanism; The multi-level redundancy safety mechanism includes an electrical redundancy mechanism and a mechanical backup; the electrical redundancy mechanism includes at least two steering motors connected in parallel via differential gears to form electrical redundancy; the mechanical backup includes an electromagnetic clutch and a backup drive shaft, the electromagnetic clutch being configured to connect the backup drive shaft to the steering drive chain when the electrical redundancy mechanism fails, so that the manual steering torque is transmitted to the wheels through the backup drive shaft, thereby changing the wheel angle.

7. The intelligent vehicle corner module system for cross-domain collaborative control according to claim 3, characterized in that, The suspension system includes an air spring unit, an electric damper, and a multi-link geometry adjustment mechanism; The air spring unit is configured to adjust the spring stiffness and chassis height by adjusting the internal air pressure; the electric damper is configured to provide damping force; the multi-link geometry adjustment mechanism includes multiple motor-driven links that adjust the camber and toe angles of the wheels by changing the link length or connection point position. The suspension system is configured to calculate the actuation commands of the air spring unit, electric damper, and multi-link geometry adjustment mechanism based on predicted road input and vehicle state, in order to adjust the vehicle body posture and vibration level.

8. A control method for a cross-domain collaborative control intelligent vehicle corner module system, characterized in that, include: By fusing sensor data through a parameter estimator, the vehicle motion state and road surface adhesion parameters are output. The motion prediction controller receives motion commands and calculates generalized force requirements based on the prediction model and the vehicle's motion state. The prediction model is composed of a vehicle nonlinear dynamics model and a tire nonlinear model. The generalized force requirements include the required longitudinal resultant force and the required yaw moment. The actuator status is detected by the diagnostic fault-tolerant module, and a health status flag is generated. The generalized force demand and road surface adhesion parameters are received by the inter-wheel optimization distributor, and optimization is performed under preset constraints to generate control commands for the actuator. The preset constraints include tire adhesion boundaries, actuator physical limits, and health status flags. The local controller drives the corresponding actuator to perform actions according to the control instructions. The diagnostic fault-tolerant module receives feedback parameters generated by the actuator and regenerates the health status flag to trigger the re-optimization of the inter-wheel optimizer. The motion predictive controller runs a nonlinear model predictive control algorithm, the control algorithm including: taking the vehicle motion state as the current state, and performing rolling optimization in a finite time domain based on the predictive model and the vehicle motion state to calculate the generalized force demand; The optimization is performed by constructing and solving a quadratic programming problem through the inter-round optimization allocator; the quadratic programming problem includes constraints and an objective function. The optimization variables of the quadratic programming problem are the control commands of each actuator; The constraints of the quadratic programming problem include: equality constraints for achieving the generalized force requirement, tire adhesion boundary inequality constraints determined based on the road surface adhesion parameters and tire load, and inequality constraints on the capabilities of each actuator determined based on the health status flag. The objective function is configured to simultaneously minimize tracking error and control cost.

Images