Hydrogen energy heavy truck omni-directional freedom optimization method

Abstract

The application relates to the technical field of automobile chassis dynamics control, and discloses a hydrogen energy heavy truck omnidirectional freedom optimization method, which comprises the following steps: acquiring multi-source sensor data and hydrogen fuel cell system state data of the hydrogen energy heavy truck, and performing time synchronization and noise preprocessing; using a nineteen-freedom vehicle dynamics model and an adaptive filtering algorithm, real-time identification is performed on key state parameters of the vehicle; a hierarchical omnidirectional coordinated control architecture is constructed, which comprises a master control layer, a scheduling layer and an execution layer, the master control layer generates a global motion control target; multi-objective optimization is solved in the scheduling layer, and an expected force distribution instruction is calculated; the execution layer converts the instruction into a bottom-layer execution control signal, and drives the coordinated action of each subsystem to realize omnidirectional freedom optimization control. The application can break through the limitations of poor coordination and multi-objective optimization difficulty caused by independent control of each subsystem of a traditional chassis, and realizes accurate control of the hydrogen energy heavy truck under complex working conditions.

Description

Technical Field

[0001] This application relates to the field of automotive chassis dynamics control technology, and in particular to an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks. Background Technology

[0002] Hydrogen-powered heavy-duty trucks, with their advantages of long range, zero emissions, and high load capacity, have become an important development direction in the fields of green logistics and heavy-duty transportation. As a complex electromechanical-hydraulic coupled system, hydrogen-powered heavy-duty trucks typically integrate advanced chassis systems such as distributed electric drive axles, electromechanical braking, steer-by-wire, and active suspension. The coordinated operation of these subsystems is crucial to ensuring the safety and stability of the vehicle under high-speed driving, heavy-load turning, and complex road conditions. However, hydrogen-powered heavy-duty trucks are characterized by a high center of gravity, large moment of inertia, and significant nonlinear dynamic characteristics. Furthermore, during long-distance transportation, the vehicle's mass and center of gravity distribution undergo dynamic time-varying changes as hydrogen fuel is continuously consumed, posing a significant challenge to the precise control of the vehicle.

[0003] Most existing heavy-duty truck chassis control technologies employ independent, decentralized control architectures. For example, anti-lock braking systems, electronic stability programs, and active suspension systems often operate independently, lacking top-level coordination and deep inter-system linkage. Existing control models struggle to achieve optimal tire force distribution in longitudinal, lateral, and vertical spaces when dealing with extreme conditions or multi-degree-of-freedom coupled motions, easily leading to conflicting control objectives between subsystems and decreased vehicle handling stability. Furthermore, traditional control algorithms are typically based on simplified two- or seven-degree-of-freedom linear models, neglecting the impact of frame torsion, cab suspension motion, and suspension geometric nonlinearity on vehicle attitude, and rarely considering mass parameter drift caused by hydrogen fuel consumption. This results in significant deviations between the control model and actual vehicle conditions. In practical applications, these deviations often cause control system response lag or overshoot, leading to severe vehicle roll, large path tracking deviations, and even instability and rollover risks, severely limiting the performance of hydrogen-powered heavy-duty truck intelligent chassis. Therefore, there is an urgent need to develop an omnidirectional degree-of-freedom optimization method capable of comprehensively sensing changes in vehicle state and achieving deep integration and coordinated control of multi-dimensional actuators. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks, which effectively overcomes the limitations of poor coordination and difficulty in multi-objective optimization caused by the independent control of each subsystem of the traditional chassis, and achieves precise control of hydrogen-powered heavy trucks under complex working conditions.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] Firstly, this application provides a method for optimizing the omnidirectional degrees of freedom of hydrogen-powered heavy-duty trucks, including:

[0008] Acquire multi-source sensor data and hydrogen fuel cell system status data of hydrogen-powered heavy trucks. The multi-source sensor data includes inertial measurement unit data, global positioning system data and wheel speed sensor data. Perform time synchronization and noise preprocessing on the multi-source sensor data.

[0009] Based on the preprocessed data, the key state parameters of the vehicle are identified in real time using a 19-DOF vehicle dynamics model and an adaptive filtering algorithm. The key state parameters include the center of gravity sideslip angle, the road adhesion coefficient, and the vehicle mass that changes dynamically with hydrogen fuel consumption.

[0010] A hierarchical omnidirectional coordinated control architecture is constructed, which includes a main control layer, a scheduling layer, and an execution layer. The main control layer analyzes the vehicle's desired motion state based on the driver's input and generates a global motion control target based on key state parameters.

[0011] At the scheduling layer, a model predictive control algorithm based on converter neural network optimization is used to perform multi-objective optimization of the global motion control objective and calculate the expected force distribution commands for each tire in the longitudinal, lateral and vertical directions.

[0012] The execution layer receives the desired force allocation command and transforms it into underlying execution control signals for the distributed electric drive axle, electromechanical braking system, steer-by-wire system, and semi-active air suspension system, driving each subsystem to coordinate its actions to achieve omnidirectional degree of freedom optimized control.

[0013] As a preferred embodiment of the omnidirectional degree of freedom optimization method for hydrogen-powered heavy trucks of the present invention, the longitudinal acceleration, lateral acceleration and yaw rate of the vehicle are collected by an inertial measurement unit.

[0014] Vehicle location coordinates and elevation change information are collected using the Global Positioning System;

[0015] The rotational speed of the four wheels is collected using wheel speed sensors;

[0016] An unscented Kalman filter algorithm is used to fuse the acquired longitudinal acceleration data to eliminate accumulated sensor errors. A measurement equation is then established, and the calculation formula is as follows:

[0017]

[0018] in, for The measurement vector at any given time includes the velocity output by the Global Positioning System (GPS) and the reference velocity output by the wheel speed sensor. for The state vector at any given time includes the actual longitudinal vehicle speed and the acceleration deviation. It is a nonlinear measurement function. The measurement noise sequence follows a covariance of... Gaussian distribution;

[0019] The noise covariance matrix is ​​dynamically corrected using adaptive weighting coefficients. The correction formula is as follows:

[0020]

[0021] in, for The time-corrected measurement noise covariance matrix; This is the forgetting factor, with a value ranging from 0 to 1; To predict measurement values, This indicates the matrix transpose.

[0022] As a preferred embodiment of the omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks of the present invention, a nineteen-degree-of-freedom vehicle dynamics model is constructed, comprising six degrees of freedom for the vehicle body, three degrees of freedom for the cab, eight degrees of freedom for the rotation and jogging of the four wheels, one degree of freedom for the steering system, and one degree of freedom for the powertrain.

[0023] The longitudinal and lateral forces of the tire are calculated based on the improved Duguff tire model, and the road adhesion coefficient is estimated by combining the unscented Kalman filter algorithm. The tire force calculation formula is as follows:

[0024]

[0025] in, For the first The combined force of the tires of each wheel The road surface adhesion coefficient, For the first Vertical load on each wheel These are tire stiffness parameters. For longitudinal slip ratio, This is the correlation coefficient for tire slip angle. This is the tire's nonlinear characteristic function;

[0026] To address the mass change caused by fuel consumption in hydrogen-powered heavy-duty trucks, a recursive least squares mass identification model based on the longitudinal kinetic equation is established. The calculation formula is as follows:

[0027]

[0028] in, for The overall vehicle quality at all times For the initial vehicle weight, for The hydrogen consumption rate of the hydrogen fuel cell system at any given time. This is a mass correction term based on dynamic residual correction.

[0029] As a preferred embodiment of the omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks of the present invention, the main control layer receives signals of accelerator pedal opening, brake pedal displacement and steering wheel angle.

[0030] The ideal yaw rate and ideal sideslip angle of the vehicle are calculated based on a two-degree-of-freedom reference model. The calculation formula is as follows:

[0031]

[0032] in, For the ideal yaw rate, For longitudinal vehicle speed, Wheelbase For insufficient turning coefficient, The steering angle of the front wheels;

[0033] Using the ideal yaw rate and ideal center of gravity sideslip angle as tracking targets, and combining the vehicle mass and road adhesion coefficient, a global motion control target is generated. The global motion control target includes the vehicle's required longitudinal force, required lateral force, and required yaw moment.

[0034] As a preferred embodiment of the omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy-duty trucks of the present invention, an objective function for omnidirectional degree-of-freedom optimization is defined. The objective function includes a tracking error term, a control input increment term, and an actuator constraint penalty term, and the calculation formula is as follows:

[0035]

[0036] in, To optimize performance metrics, To predict the time domain, To control the time domain, for The first time predicted The system output vector for each step includes longitudinal velocity, lateral velocity, and yaw rate. As the reference trajectory vector, To control the input increment vector, including the four-wheel drive torque increment and steering angle increment, The state weight matrix is... To control the weight matrix;

[0037] A converter neural network module is introduced to accelerate the prediction of initial values ​​for solving the model predictive control algorithm. The converter neural network module takes the current vehicle state and environmental parameters as input and outputs optimized control sequence initial values.

[0038] Based on the objective function, a quadratic programming solution is performed to calculate the optimal tire force distribution vector for each wheel, under the premise of satisfying the tire friction circle constraint and the actuator physical limit constraint.

[0039] As a preferred embodiment of the omnidirectional degree of freedom optimization method for hydrogen-powered heavy trucks of the present invention, for longitudinal force distribution, the torque command and braking pressure command of each wheel drive motor are calculated based on the motor efficiency distribution diagram of the distributed electric drive axle and the response characteristics of the electromechanical braking system, so as to realize drive anti-slip and regenerative braking energy recovery.

[0040] For lateral force distribution, the lateral component in the desired force distribution command is converted into the wheel angle command of the steer-by-wire system;

[0041] For vertical force distribution, the air spring stiffness adjustment command and damper damping force adjustment command of the semi-active air suspension system are calculated based on the vehicle pitch angle and roll angle.

[0042] The drive motor torque command, braking pressure command, wheel angle command, air spring stiffness adjustment command, and shock absorber damping force adjustment command are synchronously sent to the corresponding underlying controller via the vehicle Ethernet bus.

[0043] As a preferred embodiment of the omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy-duty trucks of the present invention, an energy coupling management mechanism is established, and a hydrogen fuel cell power prediction module is integrated into the omnidirectional degree-of-freedom optimization process.

[0044] The hydrogen fuel cell power prediction module predicts the power demand within a future time window based on the global motion control target, and dynamically adjusts the power distribution ratio of the distributed electric drive bridge in combination with the state of charge of the power battery.

[0045] When the vehicle is detected to be braking on a long downhill slope, regenerative braking is prioritized through the distributed electric drive axle, and the recovered energy is stored in the power battery. At the same time, the output power of the hydrogen fuel cell is reduced to maintain energy balance.

[0046] As a preferred embodiment of the omnidirectional degree of freedom optimization method for hydrogen-powered heavy trucks of the present invention, the working status of the distributed electric drive axle, electromechanical braking system, steer-by-wire system and semi-active air suspension system is monitored in real time.

[0047] When a failure is detected in an actuator, the control is reconfigured using the redundant degrees of freedom of the remaining healthy actuators.

[0048] If the front wheel steering system fails, an additional yaw moment is generated by the difference in longitudinal driving force between the left and right wheels to assist the vehicle in completing the steering action.

[0049] If braking fails on a single wheel, the braking force is redistributed to the remaining three wheels, and the suspension damping is adjusted to suppress sudden changes in vehicle posture caused by asymmetric braking force.

[0050] Secondly, this invention provides an omnidirectional degree-of-freedom optimization system for hydrogen-powered heavy-duty trucks, comprising: a multi-source data sensing and preprocessing module for acquiring multi-source sensor data and hydrogen fuel cell system state data of the hydrogen-powered heavy-duty truck, the multi-source sensor data including inertial measurement unit data, global positioning system data, and wheel speed sensor data, and performing time synchronization and noise preprocessing on the multi-source sensor data; a vehicle state identification and parameter estimation module for identifying key vehicle state parameters in real time based on the preprocessed data, using a 19-degree-of-freedom vehicle dynamics model and an adaptive filtering algorithm, the key state parameters including the center of gravity sideslip angle, road adhesion coefficient, and vehicle mass dynamically changing with hydrogen fuel consumption; and a hierarchical omnidirectional coordinated control architecture module for constructing a hierarchical omnidirectional coordinated control framework. The architecture comprises a main control layer, a scheduling layer, and an execution layer. The main control layer analyzes the vehicle's desired motion state based on driver input and generates a global motion control objective based on key state parameters. The multi-objective optimization and force distribution module, at the scheduling layer, uses a model predictive control algorithm based on converter neural network optimization to perform multi-objective optimization on the global motion control objective, calculating the desired force distribution commands for each tire in the longitudinal, lateral, and vertical directions. The distributed underlying collaborative execution module receives the desired force distribution commands from the execution layer and transforms them into underlying execution control signals for the distributed electric drive axle, electromechanical braking system, steer-by-wire system, and semi-active air suspension system, driving the coordinated actions of each subsystem to achieve omnidirectional degree-of-freedom optimized control.

[0051] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks.

[0052] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention overcomes the limitations of poor coordination and difficulty in multi-objective optimization caused by the independent control of each subsystem of a traditional heavy-duty truck chassis. By synchronously collecting data from multiple sources of sensors and the state of the hydrogen fuel cell system, and introducing a 19-DOF vehicle dynamics model tailored to the complex characteristics of heavy-duty trucks, it achieves deep perception of the vehicle's dynamic state. In particular, by combining an adaptive filtering algorithm to identify the dynamically changing vehicle mass and road adhesion coefficient with hydrogen fuel consumption in real time, it eliminates the adverse effects of time-varying mass parameters on control accuracy, providing a precise "model benchmark" for omnidirectional coordinated control. The hierarchical omnidirectional coordinated control architecture constructed by this invention can effectively decouple the complex vehicle control task into top-level motion planning, mid-level force distribution, and bottom-level execution control, ensuring both... The clarity of the control logic further enhances the system's response speed. At the core scheduling layer, an innovative model predictive control algorithm based on converter neural network optimization is adopted. This algorithm leverages the powerful sequence prediction capability of neural networks to accelerate the optimization process, solving the problem of long computation time in traditional control prediction algorithms when dealing with high-dimensional multi-objective optimization. This ensures the real-time and optimal generation of control commands. Finally, by transforming the optimized tire force distribution commands into coordinated actions of the distributed electric drive axle, electromechanical braking, steer-by-wire, and semi-active air suspension, the system achieves comprehensive linkage and precise control of the vehicle in longitudinal drive braking, lateral steering, and vertical suspension movement. This significantly improves the driving stability, ride comfort, and energy utilization efficiency of hydrogen-powered heavy trucks under complex working conditions, effectively ensuring the vehicle's all-domain dynamic safety. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a schematic diagram illustrating an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy-duty trucks proposed in this invention;

[0055] Figure 2 This is a schematic diagram of an omnidirectional degree-of-freedom optimization system for hydrogen-powered heavy trucks proposed in this invention;

[0056] Figure 3 This is a schematic diagram of an electronic device proposed by the present invention. Detailed Implementation

[0057] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0058] Example 1, referring to Figure 1 As the first embodiment of the present invention, a method for optimizing the omnidirectional degrees of freedom of a hydrogen-powered heavy-duty truck is provided, comprising:

[0059] S1. Acquire multi-source sensor data and hydrogen fuel cell system status data of the hydrogen-powered heavy truck. The multi-source sensor data includes inertial measurement unit data, global positioning system data and wheel speed sensor data. Perform time synchronization and noise preprocessing on the multi-source sensor data.

[0060] S2. Based on the preprocessed data, the key state parameters of the vehicle are identified in real time using the 19-DOF vehicle dynamics model and adaptive filtering algorithm. The key state parameters include the center of gravity sideslip angle, road adhesion coefficient and the vehicle mass that changes dynamically with hydrogen fuel consumption.

[0061] S3. Construct a hierarchical omnidirectional coordinated control architecture, which includes a main control layer, a scheduling layer and an execution layer. The main control layer analyzes the vehicle's desired motion state based on the driver's input and generates a global motion control target based on key state parameters.

[0062] S4. At the scheduling layer, a model predictive control algorithm based on converter neural network optimization is adopted to perform multi-objective optimization solution for the global motion control objective and calculate the expected force distribution commands of each tire in the longitudinal, lateral and vertical directions.

[0063] S5. The execution layer receives the desired force allocation command and converts it into a low-level execution control signal for the distributed electric drive axle, electromechanical braking system, steer-by-wire system and semi-active air suspension system, driving each subsystem to work together to achieve omnidirectional degree of freedom optimization control.

[0064] Specifically, firstly, this invention constructs a closed-loop "perception-decision-control" system suitable for the complex electromechanical-hydraulic coupling system of hydrogen-powered heavy-duty trucks. At the hardware level, the system relies on a high-speed communication network built upon an in-vehicle Ethernet and a flexible data rate controller local area network bus, connecting various sensors (including a six-axis inertial measurement unit, a dual-antenna differential positioning module, and a high-precision wheel speed encoder) and actuators distributed throughout the vehicle body. The core omnidirectional coordination controller uses a high-performance automotive-grade chip, possessing high computing power to support complex dynamic calculations and neural network inference, and integrating a redundancy design at automotive safety integrity level D. In terms of software logic, this invention breaks through the limitations of traditional single subsystems (such as ABS and ESP) operating independently. By introducing a high-fidelity nineteen-degree-of-freedom model, it deeply restores the dynamic characteristics of the vehicle under heavy loads, high speeds, and complex road conditions, especially providing targeted compensation for the time-varying mass characteristics of hydrogen-powered heavy-duty trucks caused by hydrogen fuel consumption. By employing a layered architecture, vehicle control tasks are decoupled into three levels: top-level motion planning, mid-level torque distribution, and bottom-level execution. Furthermore, a converter neural network is innovatively introduced at the scheduling layer to accelerate the solution process of model predictive control, solving the engineering challenge of nonlinear multi-objective optimization algorithms failing to converge within millisecond-level control cycles. Ultimately, this method achieves omnidirectional coordination of the drive, braking, steering, and suspension systems, effectively improving vehicle handling stability, ride comfort, and energy efficiency, ensuring the operational safety of hydrogen-powered heavy trucks throughout their entire lifecycle and under all operating conditions.

[0065] Specifically, S1 includes the following sub-steps:

[0066] The vehicle's longitudinal acceleration, lateral acceleration, and yaw rate are collected using an inertial measurement unit.

[0067] Vehicle location coordinates and elevation change information are collected using the Global Positioning System;

[0068] The rotational speed of the four wheels is collected using wheel speed sensors;

[0069] An unscented Kalman filter algorithm is used to fuse the acquired longitudinal acceleration data to eliminate accumulated sensor errors. A measurement equation is then established, and the calculation formula is as follows:

[0070]

[0071] in, for The measurement vector at any given time includes the velocity output by the Global Positioning System (GPS) and the reference velocity output by the wheel speed sensor. for The state vector at any given time includes the actual longitudinal vehicle speed and the acceleration deviation. It is a nonlinear measurement function. The measurement noise sequence follows a covariance of... Gaussian distribution;

[0072] For example, in specific calculations, it is assumed that the vehicle is in a state of uniform acceleration and linear motion, and the sampling time interval is... .exist At that time, the longitudinal velocity observation value obtained by the system from the Global Positioning System was... The average wheel speed (reference speed) calculated using four wheel speed sensors and after removing outliers is: At this point, the measurement vector is constructed. State vector Includes the predicted true longitudinal vehicle speed from the previous moment. Zero bias error of the accelerometer in the inertial measurement unit Nonlinear measurement function The state space is mapped to the measurement space, and the predicted measurement values ​​are calculated. At this point, the measurement residuals are calculated. Assume the initial measurement noise covariance. Diagonal elements are By performing nonlinear propagation and weighted calculation through unscented transformation, and updating the state vector using residuals, the final corrected optimal estimate of the true longitudinal vehicle speed may be: It also updates the zero bias error estimate of the accelerometer in real time, thereby eliminating measurement noise and drift of a single sensor.

[0073] The noise covariance matrix is ​​dynamically corrected using adaptive weighting coefficients. The correction formula is as follows:

[0074]

[0075] in, for The time-corrected measurement noise covariance matrix; This is the forgetting factor, with a value ranging from 0 to 1; To predict measurement values, Indicates matrix transpose;

[0076] For example, in specific calculations, a forgetting factor is set. This value determines the length of the algorithm's "memory" of historical noise statistics. The noise covariance matrix from the previous time step. identity matrix The measurement residual calculated at the current moment. Then the outer product of the residuals Substituting into the correction formula, the updated... Through this dynamic correction, when a vehicle enters a tunnel causing a sudden drop in GPS signal quality (the residuals increase, for example, the first residual from...), the GPS signal quality is adjusted accordingly. jump to When this happens, the algorithm will automatically and significantly increase... The noise variance term of the corresponding positioning system is used to reduce the weight of the signal source in the subsequent filter gain calculation, and more reliance is placed on the integral data of the inertial measurement unit for state recursion, thus ensuring the continuity and robustness of vehicle speed estimation in the signal loss area.

[0077] Specifically, step S1 effectively solves the problem of single sensor failure or accuracy degradation under specific operating conditions through the physical integration and data fusion of multi-source heterogeneous sensors. In the actual operation of hydrogen-powered heavy-duty trucks, the data sampling frequencies and transmission delays of different sensors vary, easily leading to data misalignment in time. This step first aligns the timestamps of all sensor data at the microsecond level based on a precision clock protocol, ensuring that subsequent algorithms process the vehicle state at the same moment. Subsequently, to address the issue of the inertial measurement unit being susceptible to high-frequency vibration interference from the engine, a fourth-order Butterworth low-pass filter is used to filter out the interference. The above-mentioned high-frequency noise. To address the multipath effect and signal loss problems of the Global Positioning System (GPS) in urban canyons or tunnels, an adaptive unscented Kalman filter algorithm with online noise statistical characteristic estimation function was designed. This algorithm can not only fuse multi-source information to estimate the vehicle's motion state, but also sense the health of sensors in real time (reflected by the noise covariance matrix), achieving "immunity" to abnormal observations. This process provides high-confidence, low-latency, and high-precision vehicle motion state reference data for subsequent dynamic control, which is a prerequisite for achieving omnidirectional precise control;

[0078] Specifically, S2 includes the following sub-steps:

[0079] A nineteen-degree-of-freedom vehicle dynamics model was constructed, including six degrees of freedom for the vehicle body, three degrees of freedom for the cab, eight degrees of freedom for the rotation and hop of the four wheels, one degree of freedom for the steering system, and one degree of freedom for the powertrain.

[0080] The longitudinal and lateral forces of the tire are calculated based on the improved Duguff tire model, and the road adhesion coefficient is estimated by combining the unscented Kalman filter algorithm. The tire force calculation formula is as follows:

[0081]

[0082] in, For the first The combined force of the tires of each wheel The road surface adhesion coefficient, For the first Vertical load on each wheel These are tire stiffness parameters. For longitudinal slip ratio, This is the correlation coefficient for tire slip angle. This is the tire's nonlinear characteristic function;

[0083] For example, in specific calculations, for the vehicle's left front wheel ( The vertical load was calculated using a 19-DOF vehicle dynamics model. (Heavy-duty truck fully loaded with braking load transfer). The current road surface is identified as wet asphalt pavement; the estimated peak adhesion coefficient is... Tire longitudinal stiffness parameters Calibrated as The longitudinal slip ratio is estimated using wheel speed sensors and vehicle speed. Side slip angle correlation coefficient Based on slip ratio and sideslip angle (Assuming) The composite function is calculated to obtain Tire nonlinear characteristic function This describes the decay characteristics of tire force after it reaches saturation; the value here is [value missing]. Substitute into the formula to calculate the resultant force of the tires. (This is a simplified calculation illustration; in actual calculations, the longitudinal and lateral components need to be calculated separately, and friction circle constraints need to be considered.) This calculation process runs in real time in the edge controller, with an update frequency of up to [missing information]. This ensures that the control system can accurately grasp the limit of friction between the tires and the road surface, preventing uncontrollable slippage or sideslip when the vehicle applies driving or braking torque.

[0084] To address the mass change caused by fuel consumption in hydrogen-powered heavy-duty trucks, a recursive least squares mass identification model based on the longitudinal kinetic equation is established. The calculation formula is as follows:

[0085]

[0086] in, for The overall vehicle quality at all times For the initial vehicle weight, for The hydrogen consumption rate of the hydrogen fuel cell system at any given time. This is a mass correction term based on dynamic residual correction.

[0087] For example, in specific calculations, the initial mass of a hydrogen-powered heavy-duty truck when it leaves the warehouse fully loaded. Continuous high-speed operation time of vehicles (2 hours). Instantaneous hydrogen consumption rate transmitted by the hydrogen fuel cell system via the vehicle bus. The total consumption calculated by integration is (Relying solely on flow meter integration, the mass change may seem small, but considering the potential for cargo loading and unloading, water consumption by sprinklers, etc., during long-distance transportation, the actual mass change could be much larger.) At this point, the recursive least squares algorithm is initiated, based on Newton's second law. Utilizing feedback from the drive system Driving force (such as) ), estimated (Air resistance and rolling resistance, such as) ) and high-precision accelerometers Acceleration (e.g.) ) Calculate the dynamic mass estimate The algorithm fuses the integral calculations and dynamic estimates using Kalman filtering to calculate the correction term. Assuming the final revised Updated to This dual verification mechanism, which combines physical flow statistics and dynamic inversion, ensures that the vehicle mass parameters remain accurate even when there are cumulative errors in the hydrogen flow meter or changes in external loads (such as increased weight due to mud and water adhering to the vehicle body in rainy weather), providing accurate input for subsequent inertial torque calculations.

[0088] Specifically, step S2 overcomes the limitations of traditional simplified models (such as two-DOF bicycle models) in describing the roll, pitch, and torsional motion of heavy trucks by constructing a high-fidelity 19-DOF vehicle model. In particular, the introduction of a three-DOF (vertical, roll, and pitch) cab suspension allows the control algorithm to fully consider the impact of cab swaying on driver operation and comfort, a factor neglected by traditional models. Regarding parameter identification, this step innovatively solves the unique time-varying mass problem of hydrogen-powered heavy trucks. Due to the release of high-pressure hydrogen from the hydrogen cylinder, the emission of water generated by the fuel cell reaction, and changes in the long-distance transportation environment, the vehicle's inertial parameters are in a state of dynamic micro-variation. By combining hydrogen system state data with a dynamic inversion algorithm, online real-time correction of the vehicle's mass and center of gravity position is achieved, eliminating control errors caused by model parameter mismatch. Simultaneously, real-time estimation of the road adhesion coefficient defines a safe "dynamic stability envelope" for the vehicle, ensuring that subsequent control commands are always limited within physical limits, avoiding vehicle instability caused by blindly increasing control torque.

[0089] Specifically, S3 includes the following sub-steps:

[0090] The main control layer receives signals of accelerator pedal opening, brake pedal displacement, and steering wheel angle.

[0091] The ideal yaw rate and ideal sideslip angle of the vehicle are calculated based on a two-degree-of-freedom reference model. The calculation formula is as follows:

[0092]

[0093] in, For the ideal yaw rate, For longitudinal vehicle speed, Wheelbase For insufficient turning coefficient, The steering angle of the front wheels;

[0094] For example, in specific calculations, the vehicle's longitudinal speed (about ), vehicle wheelbase calibrated understeering coefficient Set as The driver inputs the front wheel steering angle at the entrance to the curve. (about (degree). Substitute into the formula, the denominator part Molecular part Then the ideal yaw rate (about (degrees per second). This value represents the optimal steering response rate that the driver expects the vehicle to achieve under current high-speed conditions, based on the vehicle's steady-state response characteristics. If the actual measured yaw rate is much greater than this value (e.g., reaching...), the driver will be disappointed. This indicates that the vehicle is exhibiting oversteer (fishtailing); conversely, understeer (pushing) indicates understeer. The control system will then... To track the target, subsequent closed-loop control law calculations are performed;

[0095] Using the ideal yaw rate and ideal center of gravity sideslip angle as tracking targets, and combining the vehicle mass and road adhesion coefficient, a global motion control target is generated. The global motion control target includes the vehicle's required longitudinal force, required lateral force, and required yaw moment.

[0096] Specifically, step S3 designs a hierarchical control architecture to solve the complex control problems of multi-input multi-output systems. The main control layer, acting as the "brain" of the entire control system, is responsible for interpreting the driver's driving intentions. Its core lies in using a reference model to calculate the ideal posture the vehicle "should" exhibit under the current operating conditions. This is achieved by introducing an understeer coefficient proportional to the square of the vehicle speed. The reference model can automatically adjust the ideal steering sensitivity according to vehicle speed, ensuring that the vehicle is agile at low speeds and stable at high speeds, meeting the driver's psychological expectations. When generating the global motion control target, the algorithm uses the road adhesion coefficient identified by S2. Dynamically limit the reference value. For example, on icy or snowy roads. Even if the driver makes a sharp turn of the steering wheel, the ideal yaw rate generated by the main control layer will be forcibly limited within physical limits to prevent the vehicle from completely losing control due to excessively high control targets. The generated global control target (such as total required longitudinal force) Total demand yaw moment This provides explicit and secure mathematical constraints for the optimization calculations of the next scheduling layer;

[0097] The specific S4 includes the following sub-steps:

[0098] Define the objective function for omnidirectional degree-of-freedom optimization. The objective function includes a tracking error term, a control input increment term, and an actuator constraint penalty term. The calculation formula is as follows:

[0099]

[0100] in, To optimize performance metrics, To predict the time domain, To control the time domain, for The first time predicted The system output vector for each step includes longitudinal velocity, lateral velocity, and yaw rate. As the reference trajectory vector, To control the input increment vector, including the four-wheel drive torque increment and steering angle increment, The state weight matrix is... To control the weight matrix;

[0101] For example, in specific calculations, the prediction time domain is set. Control time domain State weight matrix It is a diagonal matrix, and its diagonal elements are set to... These correspond to the tracking weights of longitudinal velocity, lateral velocity, and yaw rate, respectively. The magnitude of these values ​​indicates the system's priority in ensuring yaw stability under the current operating conditions (weights). (Maximum). Control weight matrix Diagonal elements are This is used to punish drastic changes in control quantities to protect the actuator. At what moment, based on the model, predict the future number of... The longitudinal velocity deviation of the step is Lateral velocity deviation is The yaw rate deviation is The contribution of the tracking error term in this step is approximately... If the optimized control quantity (such as the left front wheel drive torque) changes... (After normalization) Then control the contribution of the incremental term. The optimizer adjusts the control sequence within the constraints. This will make it possible in the future Overall performance index within the step Minimize, thereby finding the optimal balance between rapidly tracking the target and maintaining smooth control;

[0102] A converter neural network module is introduced to accelerate the prediction of initial values ​​for solving the model predictive control algorithm. The converter neural network module takes the current vehicle state and environmental parameters as input and outputs optimized control sequence initial values.

[0103] Based on the objective function, a quadratic programming solution is performed to calculate the optimal tire force distribution vector for each wheel, under the premise of satisfying the tire friction circle constraint and the actuator physical limit constraint.

[0104] Specifically, step S4 is the core computational layer of the omnidirectional degree-of-freedom optimization method. While traditional model predictive control can handle multi-constraint optimization problems, its online solution of quadratic programming is time-consuming (typically requiring tens of milliseconds), making it difficult to meet the millisecond-level control requirements of heavy-duty trucks traveling at high speeds. This invention innovatively introduces a converter neural network as a "hot-start" mechanism for model predictive control. The converter neural network utilizes its powerful sequence modeling and attention mechanisms to learn a large number of optimal control trajectories under standard operating conditions offline. In real-time operation, based on the current vehicle state (such as speed and sideslip angle) and environmental parameters (such as adhesion coefficient), it directly predicts a set of control sequences extremely close to the optimal solution as initial values ​​for the quadratic programming solver. Experiments show that this "conjecture-correction" mechanism reduces the average number of iterations for quadratic programming solutions from [missing information]. The next reduction to This reduces computation time. That's all. In this way, the scheduling layer can achieve this in a very short time (…). Taking into account tire friction circle constraints (to prevent tire lock-up or slippage), motor peak torque constraints, and suspension travel constraints, the optimal force distribution of the four wheels in the longitudinal, lateral, and vertical directions is calculated, truly realizing the real-time implementation of complex multi-objective optimization.

[0105] The specific S5 includes the following sub-steps:

[0106] For longitudinal force distribution, based on the motor efficiency distribution diagram of the distributed electric drive bridge and the response characteristics of the electromechanical braking system, the torque command and braking pressure command of each wheel drive motor are calculated to achieve drive anti-slip and regenerative braking energy recovery.

[0107] For lateral force distribution, the lateral component in the desired force distribution command is converted into the wheel angle command of the steer-by-wire system;

[0108] For vertical force distribution, the air spring stiffness adjustment command and damper damping force adjustment command of the semi-active air suspension system are calculated based on the vehicle pitch angle and roll angle.

[0109] The drive motor torque command, braking pressure command, wheel angle command, air spring stiffness adjustment command, and shock absorber damping force adjustment command are synchronously sent to the corresponding underlying controller via the vehicle Ethernet bus.

[0110] Specifically, step S5 is responsible for converting the abstract physical quantities (force / torque) calculated by the scheduling layer into specific low-level actuator electrical signals and coordinating the actions of each subsystem. For the distributed electric drive axle, the system consults the pre-stored motor efficiency distribution map and, under the premise of meeting the total driving force requirements, prioritizes the allocation of torque to the motors in the high-efficiency zone. Alternatively, when yaw torque assistance is required, it commands one side of the motor to drive while the other side recovers energy (braking), achieving differential steering while simultaneously recovering energy. For lateral force distribution, the system converts the target lateral force into precise steering wheel angle commands and sends them to the steer-by-wire system for execution. For vertical force distribution, when S1 detects excessive lateral acceleration causing body roll, S5 immediately sends a command to increase the damping value of the outer wheel shock absorber (e.g., from...). Adjust to The system incorporates increased air spring stiffness to provide strong lateral support and suppress roll; however, stiffness and damping are reduced when traversing bumpy roads to enhance comfort. All commands are transmitted via the vehicle's Ethernet network. The periodic synchronous broadcast is sent to the controllers of each subsystem, ensuring a high degree of coordination in the execution of the vehicle's actions and realizing true omnidirectional (longitudinal, lateral, and vertical) degree of freedom control;

[0111] In some embodiments, a method for optimizing the omnidirectional degrees of freedom of a hydrogen-powered heavy truck may further include the following steps:

[0112] Establish an energy coupling management mechanism and integrate a hydrogen fuel cell power prediction module into the omnidirectional degree of freedom optimization process;

[0113] The hydrogen fuel cell power prediction module predicts the power demand within a future time window based on the global motion control target, and dynamically adjusts the power distribution ratio of the distributed electric drive bridge in combination with the state of charge of the power battery.

[0114] When the vehicle is detected to be braking on a long downhill slope, regenerative braking is prioritized through the distributed electric drive axle, and the recovered energy is stored in the power battery. At the same time, the output power of the hydrogen fuel cell is reduced to maintain energy balance.

[0115] Specifically, this implementation method is deeply optimized for the energy characteristics of hydrogen-powered heavy-duty trucks. Hydrogen fuel cells have relatively slow dynamic response characteristics, making them difficult to adapt to sudden power surges during rapid acceleration or heavy-load uphill driving. This embodiment utilizes the predictive characteristics of model predictive control algorithms to anticipate power demands several seconds in advance. For example, before the vehicle enters an uphill section (via map or road condition perception), the system proactively increases the fuel cell output power, using the power battery for peak shaving and valley filling. During long downhill driving, the system dynamically adjusts the regenerative braking ratio of the distributed electric drive axle, efficiently converting the vehicle's significant gravitational potential energy into electrical energy and recovering it into the power battery. If the power battery's state of charge approaches its upper limit (e.g., ...), the system can adjust the regenerative braking ratio of the distributed electric drive axle to efficiently recover the vehicle's enormous gravitational potential energy into electrical energy and return it to the power battery. The system automatically coordinates the intervention of the electromechanical braking system (friction braking) to prevent battery overcharging. This deep coupling of energy management and dynamic control not only ensures the vehicle's power performance and braking safety, but also significantly reduces hydrogen consumption and extends the vehicle's driving range by maximizing energy recovery and optimizing the fuel cell operating point.

[0116] In some embodiments, a method for optimizing the omnidirectional degrees of freedom of a hydrogen-powered heavy truck may further include the following steps:

[0117] Real-time monitoring of the operating status of the distributed electric drive axle, electromechanical braking system, steer-by-wire system, and semi-active air suspension system;

[0118] When a failure is detected in an actuator, the control is reconfigured using the redundant degrees of freedom of the remaining healthy actuators.

[0119] If the front wheel steering system fails, an additional yaw moment is generated by the difference in longitudinal driving force between the left and right wheels to assist the vehicle in completing the steering action.

[0120] If braking fails on a single wheel, the braking force is redistributed to the remaining three wheels, and the suspension damping is adjusted to suppress sudden changes in vehicle posture caused by asymmetric braking force.

[0121] Specifically, this implementation significantly improves the safety of hydrogen-powered heavy trucks under extreme fault conditions. While the steer-by-wire chassis system eliminates mechanical connections, it also introduces the potential risk of electronic system failure. This invention fully utilizes the multi-degree-of-freedom redundancy characteristics of distributed drive systems to construct a robust fault-tolerant control mechanism. For example, when the steer-by-wire motor is detected to be unable to control the vehicle's steering angle due to overheating or circuit failure, the system immediately switches to "differential steering mode." Through precise calculation, it commands the right wheel to output driving force (such as...). ), the left wheel outputs braking force (such as The system utilizes the significant longitudinal force difference between the two sides to generate an additional yaw moment, forcibly changing the vehicle's course and enabling it to safely pull over. If the left front wheel brake fails, the system not only redistributes the braking force of the remaining three wheels to maintain a constant total braking force, but also instantly adjusts the stiffness and damping of the diagonal (right rear wheel) and same-side (left rear wheel) suspensions to suppress violent yaw and nose-diving caused by asymmetrical braking, preventing rollover. This fault-tolerant control capability based on omnidirectional degree-of-freedom coupling is unattainable by traditional fuel-powered heavy trucks, fully demonstrating the structural advantages and intelligent level of distributed drive hydrogen-powered heavy trucks.

[0122] Example 2, refer to Figure 2 This is the second embodiment of the present invention, which provides an omnidirectional degree-of-freedom optimization system for a hydrogen-powered heavy-duty truck, comprising: a multi-source data sensing and preprocessing module, used to acquire multi-source sensor data of the hydrogen-powered heavy-duty truck and state data of the hydrogen fuel cell system, the multi-source sensor data including inertial measurement unit data, global positioning system data, and wheel speed sensor data, and performing time synchronization and noise preprocessing on the multi-source sensor data; a vehicle state identification and parameter estimation module, used to identify key vehicle state parameters in real time based on the preprocessed data, using a 19-degree-of-freedom vehicle dynamics model and an adaptive filtering algorithm, the key state parameters including the center of gravity sideslip angle, road adhesion coefficient, and the vehicle mass dynamically changing with hydrogen fuel consumption; and a hierarchical omnidirectional coordinated control architecture module, used to construct a hierarchical omnidirectional... The coordinated control architecture comprises a main control layer, a scheduling layer, and an execution layer. The main control layer analyzes the vehicle's desired motion state based on driver input and generates a global motion control objective based on key state parameters. A multi-objective optimization and force distribution module, at the scheduling layer, employs a model predictive control algorithm based on a converter neural network to perform multi-objective optimization on the global motion control objective, calculating the desired force distribution commands for each tire in the longitudinal, lateral, and vertical directions. A distributed underlying collaborative execution module receives the desired force distribution commands from the execution layer and transforms them into underlying execution control signals for the distributed electric drive axle, electromechanical braking system, steer-by-wire system, and semi-active air suspension system, driving the coordinated actions of each subsystem to achieve omnidirectional degree-of-freedom optimized control.

[0123] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks.

[0124] like Figure 3As shown, the electronic device may include a processor 610, a communications interface 620, a memory 630, and a communication bus 640. The processor 610, communications interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions from the memory 630 to execute an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy-duty trucks.

[0125] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute a method for optimizing the omnidirectional degrees of freedom of a hydrogen-powered heavy truck.

[0127] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks.

[0128] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0129] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0130] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0131] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for optimizing the omnidirectional degrees of freedom of a hydrogen-powered heavy-duty truck, characterized in that, include: The system acquires multi-source sensor data and hydrogen fuel cell system status data from hydrogen-powered heavy-duty trucks. The multi-source sensor data includes inertial measurement unit data, global positioning system data, and wheel speed sensor data. Time synchronization and noise preprocessing are then performed on the multi-source sensor data. Based on the preprocessed data, the key state parameters of the vehicle are identified in real time using a 19-DOF vehicle dynamics model and an adaptive filtering algorithm. The key state parameters include the center of gravity sideslip angle, the road adhesion coefficient, and the vehicle mass that changes dynamically with hydrogen fuel consumption. A hierarchical omnidirectional coordinated control architecture is constructed, which includes a main control layer, a scheduling layer and an execution layer. The main control layer analyzes the vehicle's desired motion state based on the driver's operation input and generates a global motion control target based on the key state parameters. In the scheduling layer, a model predictive control algorithm based on converter neural network optimization is used to perform multi-objective optimization solution for the global motion control objective, and calculate the expected force distribution commands of each tire in the longitudinal, lateral and vertical directions. The execution layer receives the desired force allocation command and converts it into a low-level execution control signal for the distributed electric drive axle, electromechanical braking system, steer-by-wire system and semi-active air suspension system, driving each subsystem to work together to achieve omnidirectional degree of freedom optimized control.

2. The omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks according to claim 1, characterized in that, The process involves acquiring multi-source sensor data and hydrogen fuel cell system status data from a hydrogen-powered heavy-duty truck. The multi-source sensor data includes inertial measurement unit data, global positioning system data, and wheel speed sensor data. The process also includes performing time synchronization and noise preprocessing on the multi-source sensor data. Specifically: The inertial measurement unit collects the vehicle's longitudinal acceleration, lateral acceleration, and yaw rate. The vehicle's location coordinates and elevation change information are collected using the global positioning system. The rotational speed of the four wheels is collected by the wheel speed sensor; The acquired longitudinal acceleration is fused using an unscented Kalman filter algorithm to eliminate sensor cumulative errors. A measurement equation is then established, and the calculation formula is as follows: ； in, for The measurement vector at any given time includes the velocity output by the GPS and the reference velocity output by the wheel speed sensor. for The state vector at any given time includes the actual longitudinal vehicle speed and the acceleration deviation. It is a nonlinear measurement function. The measurement noise sequence follows a covariance of... Gaussian distribution; The noise covariance matrix is ​​dynamically corrected using adaptive weighting coefficients. The correction formula is as follows: ； in, for The time-corrected measurement noise covariance matrix; This is the forgetting factor, with a value ranging from 0 to 1; To predict measurement values, This indicates the matrix transpose.

3. The omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks according to claim 2, characterized in that, Based on the preprocessed data, a 19-DOF vehicle dynamics model and an adaptive filtering algorithm are used to identify key vehicle state parameters in real time. These key state parameters include the center of gravity sideslip angle, road adhesion coefficient, and the vehicle mass that dynamically changes with hydrogen fuel consumption. Specifically, the steps are as follows: A nineteen-degree-of-freedom vehicle dynamics model is constructed, comprising six degrees of freedom for the vehicle body, three degrees of freedom for the cab, eight degrees of freedom for the rotation and hop of the four wheels, one degree of freedom for the steering system, and one degree of freedom for the powertrain. Based on the improved Duguff tire model, the longitudinal and lateral forces of the tire are calculated. Combined with the unscented Kalman filter algorithm, the road adhesion coefficient is estimated. The tire force calculation formula is as follows: ； in, For the first The combined force of the tires of each wheel The road surface adhesion coefficient, For the first Vertical load on each wheel These are tire stiffness parameters. For longitudinal slip ratio, This is the correlation coefficient for tire slip angle. This is the tire's nonlinear characteristic function; To address the mass change caused by fuel consumption in hydrogen-powered heavy-duty trucks, a recursive least squares mass identification model based on the longitudinal kinetic equation is established. The calculation formula is as follows: ； in, for The overall vehicle quality at all times For the initial vehicle weight, for The hydrogen consumption rate of the hydrogen fuel cell system at any given time. This is a mass correction term based on dynamic residual correction.

4. The omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks according to claim 3, characterized in that... A hierarchical omnidirectional coordinated control architecture is constructed, comprising a main control layer, a scheduling layer, and an execution layer. The main control layer parses the vehicle's desired motion state based on driver input and generates a global motion control target based on the key state parameters. Specifically, the steps are as follows: The main control layer receives signals of accelerator pedal opening, brake pedal displacement, and steering wheel angle. The ideal yaw rate and ideal sideslip angle of the vehicle are calculated based on a two-degree-of-freedom reference model. The calculation formula is as follows: ； in, For the ideal yaw rate, For longitudinal vehicle speed, Wheelbase For insufficient turning coefficient, The steering angle of the front wheels; Using the ideal yaw rate and the ideal center of gravity sideslip angle as tracking targets, and combining the vehicle mass and the road surface adhesion coefficient, the global motion control target is generated. The global motion control target includes the vehicle's required longitudinal force, required lateral force, and required yaw moment.

5. The omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks according to claim 4, characterized in that, In the scheduling layer, a model predictive control algorithm based on converter neural network optimization is used to perform multi-objective optimization to solve the global motion control objective, and to calculate the desired force distribution commands for each tire in the longitudinal, lateral, and vertical directions. Specifically, the steps are as follows: Define an objective function for omnidirectional degree-of-freedom optimization, which includes a tracking error term, a control input increment term, and an actuator constraint penalty term, and calculate the formula as follows: ； in, To optimize performance metrics, To predict the time domain, To control the time domain, for The first time predicted The system output vector for each step includes longitudinal velocity, lateral velocity, and yaw rate. As the reference trajectory vector, To control the input increment vector, including the four-wheel drive torque increment and steering angle increment, The state weight matrix is... To control the weight matrix; A converter neural network module is introduced to accelerate the prediction of the initial values ​​for solving the model predictive control algorithm. The converter neural network module takes the current vehicle state and environmental parameters as input and outputs the optimized control sequence initial values. Based on the objective function, a quadratic programming solution is performed to calculate the optimal tire force distribution vector for each wheel, under the premise of satisfying the tire friction circle constraint and the actuator physical limit constraint.

6. The omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy-duty trucks according to claim 5, characterized in that, The execution layer receives the desired force allocation command and converts it into low-level execution control signals for the distributed electric drive axle, electromechanical braking system, steer-by-wire system, and semi-active air suspension system, driving the coordinated action of each subsystem to achieve omnidirectional degree-of-freedom optimized control. Specifically, this involves: For longitudinal force distribution, based on the motor efficiency distribution diagram of the distributed electric drive bridge and the response characteristics of the electromechanical braking system, the torque command and braking pressure command of each wheel drive motor are calculated to achieve drive anti-slip and regenerative braking energy recovery. For lateral force distribution, the lateral component in the desired force distribution command is converted into the wheel angle command of the steer-by-wire system; For vertical force distribution, the air spring stiffness adjustment command and damper damping force adjustment command of the semi-active air suspension system are calculated based on the vehicle pitch angle and roll angle. The drive motor torque command, the braking pressure command, the wheel angle command, the air spring stiffness adjustment command, and the shock absorber damping force adjustment command are synchronously sent to the corresponding underlying controller via the vehicle Ethernet bus.

7. The omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy-duty trucks according to claim 1, characterized in that, The method further includes: An energy coupling management mechanism is established, and a hydrogen fuel cell power prediction module is integrated into the omnidirectional degree of freedom optimization process; The hydrogen fuel cell power prediction module predicts the power demand within a future time window based on the global motion control target, and dynamically adjusts the power allocation ratio of the distributed electric drive bridge in conjunction with the state of charge of the power battery. When the vehicle is detected to be braking on a long downhill slope, regenerative braking is preferentially performed through the distributed electric drive axle, and the recovered energy is stored in the power battery. At the same time, the output power of the hydrogen fuel cell is reduced to maintain energy balance.

8. The omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy-duty trucks according to claim 1, characterized in that, The method also includes establishing an actuator fault-tolerant control mechanism, specifically: Real-time monitoring of the operating status of the distributed electric drive axle, the electromechanical braking system, the steer-by-wire system, and the semi-active air suspension system; When a failure is detected in an actuator, the control is reconfigured using the redundant degrees of freedom of the remaining healthy actuators. If the front wheel steering system fails, an additional yaw moment is generated by the difference in longitudinal driving force between the left and right wheels to assist the vehicle in completing the steering action. If braking fails on a single wheel, the braking force is redistributed to the remaining three wheels, and the suspension damping is adjusted to suppress sudden changes in vehicle posture caused by asymmetric braking force.

9. A hydrogen-powered heavy-duty truck omnidirectional degree-of-freedom optimization system, characterized in that, include: The multi-source data sensing and preprocessing module is used to acquire multi-source sensor data and hydrogen fuel cell system status data of hydrogen-powered heavy trucks. The multi-source sensor data includes inertial measurement unit data, global positioning system data and wheel speed sensor data, and performs time synchronization and noise preprocessing on the multi-source sensor data. The vehicle state identification and parameter estimation module is used to identify key vehicle state parameters in real time based on the preprocessed data, using a 19-DOF vehicle dynamics model and an adaptive filtering algorithm. The key state parameters include the center of gravity sideslip angle, road surface adhesion coefficient, and vehicle mass that changes dynamically with hydrogen fuel consumption. A hierarchical omnidirectional coordinated control architecture module is used to construct a hierarchical omnidirectional coordinated control architecture. The architecture includes a main control layer, a scheduling layer, and an execution layer. The main control layer analyzes the vehicle's desired motion state based on the driver's operation input and generates a global motion control target based on the key state parameters. The multi-objective optimization and force distribution module is used in the scheduling layer to perform multi-objective optimization on the global motion control objective using a model predictive control algorithm based on converter neural network optimization, and to calculate the expected force distribution commands for each tire in the longitudinal, lateral and vertical directions. The distributed underlying collaborative execution module is used to receive the desired force allocation instruction from the execution layer and convert the desired force allocation instruction into underlying execution control signals for the distributed electric drive axle, electromechanical braking system, steer-by-wire system and semi-active air suspension system, so as to drive the coordinated action of each subsystem to achieve omnidirectional degree of freedom optimization control.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements an omnidirectional degree-of-freedom optimization method for hydrogen-powered heavy trucks as described in any one of claims 1-8.

Images