Dynamic braking element arrangement and control device for preventing side tipping of cruise vehicle

Abstract

The invention provides a dynamic braking element arrangement and control device for preventing a cruise vehicle from rolling, and relates to the technical field of vehicle braking control systems. The problems that in the prior art, when the complex overturning risk of a high-gravity-center vehicle is dealt with, risk sources cannot be distinguished, and control strategy switching is rigid are comprehensively solved. According to the method, the identification precision and response speed of the control system to overturning risks of different sources are improved, and the smoothness and precision of the control process are improved on the premise of ensuring the limit safety, so that the overall robustness and adaptability of the system in a complex and dynamic change scene are enhanced.

Description

Technical Field

[0001] This invention relates to the field of vehicle braking control system technology, specifically to a dynamic braking element arrangement and control device for preventing the side tilt of a parade float. Background Technology

[0002] With the continuous improvement of automation and intelligence in special vehicles, vehicle dynamic stability control technology is undergoing a profound evolution from passive safety to active safety, and from single operating conditions to full-scenario adaptive control. Especially in the application field of vehicles with high center of gravity and non-standard structures, how to achieve accurate prediction and robust intervention of complex and nonlinear rollover risks through deep fusion and collaborative control of multi-source heterogeneous information has become a key technical direction and cutting-edge research hotspot for promoting the development of vehicle active safety technology.

[0003] However, when applying existing vehicle stability control technology to specific scenarios such as parade floats, which have characteristics such as extremely low speeds, high centers of gravity, and dynamic changes, the inventors found in their research that the existing technology still faces the following areas that require further improvement:

[0004] The risk perception dimension is singular, lacking the ability to decouple the sources of rollover risk. Existing technologies, such as the method disclosed in Chinese patent CN114368369B, primarily construct their control logic around the vehicle's high-speed dynamic response and road surface friction coefficient. Such solutions typically rely on sensors such as inertial measurement units (IMUs) to holistically perceive changes in vehicle attitude, but they struggle to distinguish whether these changes are caused by quasi-static tilting resulting from predictable internal load transfers (including performer movement) or by instantaneous impacts caused by unpredictable external environmental abrupt changes (including a single wheel hitting an obstacle). This "hybrid perception" mode of risk sources prevents the control system from addressing the root cause, making it difficult to achieve an optimal balance between ensuring safety redundancy and maintaining ride comfort.

[0005] Inflexible control strategy switching makes it difficult to adapt to the continuous evolution of risks. Traditional stability control systems often employ control mode switching logic based on discrete thresholds when dealing with different risk levels. That is, when a certain state parameter exceeds a preset threshold, the system forcibly switches from a fixed control mode (including the normal stability mode) to another (including the emergency intervention mode). This "step-like" control strategy switching may cause jitter or discontinuity in the system's control output near the critical point, affecting the smoothness and accuracy of control. Especially in the process where the risk of overturning is a continuous and gradual accumulation, this rigid switching method makes it difficult to achieve refined and on-demand distribution of control torque output. Summary of the Invention

[0006] The purpose of this invention is to provide a dynamic braking element arrangement and control device for preventing parade floats from tilting, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A dynamic braking element arrangement and control device for preventing the tilt of parade floats, specifically comprising:

[0009] The first center of mass state perception module is used to acquire dynamic axle load data at multiple wheel positions of the float in real time, and determine the real-time center of mass position of the float based on the dynamic axle load data.

[0010] The second grounding margin detection module is used to determine the real-time grounding margin of the wheel by applying a preset probing braking torque pulse to the brake of at least one wheel and monitoring the angular velocity response of the wheel.

[0011] The risk source decoupling module is used to calculate an expected grounding margin based on the real-time centroid position determined by the first centroid state perception module through a preset benchmark model, and compare the expected grounding margin with the real-time grounding margin determined by the second grounding margin detection module to generate a prediction residual signal characterizing the difference between internal and external risk sources.

[0012] The adaptive selection module for control strategy is used to adaptively select between a preset feedforward stabilization control strategy based on the real-time centroid position and a feedback stabilization control strategy based on the real-time grounding margin, according to the characteristics of the predicted residual signal.

[0013] The braking torque distribution execution module is used to calculate and issue braking torque commands to the braking elements of the float according to the selected control strategy, so as to generate a stabilizing torque to suppress tilting.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: It constructs two parallel sensing channels with clear physical meaning: firstly, through a first centroid state sensing module, it accurately tracks predictable macroscopic centroid drift caused by internal load changes using dynamic axle load data; secondly, through a second ground margin detection module, it directly and rapidly quantifies unpredictable microscopic tire ground contact state changes caused by external sudden shocks using active braking detection technology. A risk source decoupling module is set up. This module generates a novel quantitative index—risk source characteristic spectrum separation degree—by comparing the outputs of the two sensing channels, i.e., the "expected state" and the "actual state." This index can accurately quantify the contribution of unpredictable external risks, thereby achieving clear decoupling of internal and external risk sources. Based on this, this invention further proposes a smooth weighted fusion control framework based on control mode authority factors, upgrading the discrete strategy "selection" to a continuous strategy "ratio," achieving a shock-free smooth transition from feedforward dominance to feedback dominance. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the modules of the present invention;

[0016] Figure 2 This is a flowchart illustrating the execution logic of the first centroid state perception module of the present invention.

[0017] Figure 3 This is a flowchart illustrating the execution logic of the second ground margin detection module of the present invention.

[0018] Figure 4 This is a flowchart illustrating the execution logic of the risk source decoupling module of the present invention.

[0019] Figure 5 This is a flowchart illustrating the execution logic of the control strategy adaptive selection module and the braking torque allocation execution module of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Example 1:

[0023] Please see Figures 1 to 5This invention provides a technical solution: a dynamic braking element arrangement and control device for preventing the tilt of parade floats, specifically comprising:

[0024] The first centroid state perception module is used to acquire dynamic axle load data at multiple wheel positions of the float in real time, and determine the real-time centroid position of the float based on the dynamic axle load data.

[0025] The second grounding margin detection module is used to determine the real-time grounding margin of the wheel by applying a preset probing braking torque pulse to the brake of at least one wheel and monitoring the angular velocity response of the wheel.

[0026] The risk source decoupling module is used to calculate an expected grounding margin based on the real-time centroid position determined by the first centroid state sensing module through a preset benchmark model, and compare the expected grounding margin with the real-time grounding margin determined by the second grounding margin detection module to generate a prediction residual signal characterizing the difference between internal and external risk sources.

[0027] The adaptive selection module for control strategy is used to adaptively select between a preset feedforward stabilization control strategy based on the real-time centroid position and a feedback stabilization control strategy based on the real-time grounding margin, according to the characteristics of the predicted residual signal.

[0028] The braking torque distribution execution module is used to calculate and send braking torque commands to the braking elements of the float according to the selected control strategy, so as to generate a stabilizing torque to suppress roll.

[0029] To further explain, the first centroid state perception module includes: dynamic axle load data collected by sensors installed at the four corner suspensions; specifically, based on the four corner dynamic axle loads and preset geometric parameters, the lateral and longitudinal centroid coordinates of the float are calculated in real time by solving the static torque balance equation; the lateral and longitudinal centroid coordinates are combined and analyzed to obtain the normalized centroid offset.

[0030] To further explain, the preset geometric parameters include the vehicle wheelbase and the vehicle track width; before solving the static torque balance equation, the first center of mass state perception module also includes calculating the total mass of the vehicle;

[0031] The specific steps for calculating the lateral center of mass coordinates include: calculating the difference between the total dynamic axle loads on the left and right sides, multiplying the difference by half the vehicle track width, and then dividing the result by the product of the vehicle's total mass and gravitational acceleration.

[0032] The steps for calculating the longitudinal centroid coordinates specifically include: calculating the difference in total dynamic axle load between the front and rear axles, multiplying the difference by half the vehicle wheelbase, and then dividing the result by the product of the vehicle's total mass and gravitational acceleration; the dynamic axle loads collected by the sensors at the four corners are low-pass filtered to remove high-frequency noise before being used for calculation.

[0033] The core function of the first centroid state perception module in this embodiment is to continuously and accurately calculate the real-time two-dimensional centroid coordinates, representing the overall mass distribution of the vehicle, by collecting and processing multiple dynamic axle load data at specific locations on the vehicle's suspension system in real time, and combining this with the vehicle's inherent preset geometric parameters and a determined physical and mechanical model. This provides a dynamic and reliable macroscopic state benchmark for subsequent risk assessment and control decisions. In this embodiment, the key parameters used to determine the real-time centroid position of the float are defined as follows:

[0034] Four-corner dynamic axle loads, their parameter symbols are as follows: , , , The physical meaning of is the instantaneous vertical load force acting on the four suspension support points of the left front, right front, left rear, and right rear wheels respectively during vehicle operation, measured in Newtons. This parameter directly reflects the total mass of the vehicle and its real-time distribution at the four support points.

[0035] The total mass of the vehicle, its parameter symbol is The physical meaning of _____ is the total mass of the float and all its loads, measured in kilograms. The wheelbase, denoted by W, is the lateral distance between the centerlines of the left and right wheels of a vehicle, measured in meters; this parameter is a preset, unchanging geometric constant. The wheelbase, denoted by L, is the longitudinal distance between the centerlines of the front and rear axles of a vehicle, measured in meters.

[0036] Lateral centroid coordinates, with parameter notation as follows: Its physical meaning is the offset distance of the vehicle's total center of mass along the lateral Y-axis in the vehicle coordinate system with the vehicle's geometric center as the origin, measured in meters. The positive or negative value of this parameter indicates whether the center of mass is biased towards the right or left side of the vehicle.

[0037] The longitudinal centroid coordinates, with parameter notation as follows: Its physical meaning is the offset distance of the vehicle's total center of mass along the longitudinal X-axis in the same vehicle coordinate system, with the unit being meters. The positive or negative value of this parameter indicates whether the center of mass is offset towards the front or rear of the vehicle. Gravitational acceleration, with the symbol g, is the intensity of the gravitational field at the Earth's surface, with the unit being meters per second squared. Four-corner dynamic axle loads { , , , The method for obtaining the load cell is to install a high-precision pad-type or pin-type load cell between the suspension system and the axle of each wheel of the float. Specifically, the installation location is the lower support plate of the air spring or the coil spring seat in the suspension system. In this embodiment, the load cell is the HBM U9C series force sensor. The raw voltage signal output by the sensor is amplified and filtered by a signal conditioner. In this embodiment, a second-order Butterworth low-pass filter with a cutoff frequency of 50Hz is used to filter out high-frequency road vibration noise. The signal is then converted into a digital quantity by an analog-to-digital converter (ADC), and finally converted into force value data in Newton units by a calibration curve pre-stored inside the vehicle control unit (ECU). Further explanation of the above parameters follows:

[0038] Total vehicle mass Derived from the definition of mass in fundamental physics; its calculation method is to take the dynamic axis loads at the four corners { , , , Summing the values, then dividing the sum by the gravitational acceleration g. Lateral center of mass coordinates. This is derived from the principle of static moment equilibrium of rigid bodies in classical mechanics. The calculation method involves calculating the total axle load of the right wheel of the vehicle (…). and The sum of ( ) and the total axle load of the left wheel ( ) and The difference between the sum of the two values ​​is multiplied by half the vehicle track width W to obtain a lateral moment caused by axle load imbalance; then this lateral moment is divided by the total vehicle mass. The product of the product with the gravitational acceleration g is used to obtain the transverse centroid coordinates.

[0039] Longitudinal centroid coordinates It originates from the principle of static moment equilibrium of rigid bodies. Its calculation method involves calculating the total axle load of the vehicle's front axle (…). and The sum of ( ) and the total axle load of the rear axle ( ) and The difference between the sum of the axle loads and the total mass of the vehicle is multiplied by half the vehicle wheelbase L to obtain the longitudinal moment caused by the axle load imbalance; then this longitudinal moment is divided by the total mass of the vehicle. The product of this and the gravitational acceleration g yields the longitudinal centroid coordinates. At the preset measurement time, the dynamic shaft loads at the four corners, after being collected and processed by the sensor, are as follows: , , , The preset vehicle geometry parameters are: wheelbase W = 2.5 meters, wheelbase L = 6.0 meters. The acceleration due to gravity g is taken as 9.8 m / s². The total vehicle mass... The calculated result is: (15000+25000+14000+26000) / 9.8≈8163.3 kg. Lateral centroid coordinates The calculation result is: [(25000+26000)-(15000+14000)]×(2.5 / 2) / (8163.3×9.8)≈0.344 meters. This result indicates that the center of gravity is offset to the right of the vehicle by 0.344 meters. (Longitudinal center of gravity coordinates) The calculation result is: [(14000+26000)-(15000+25000)]×(6.0 / 2) / (8163.3×9.8)=0 meters. This result indicates that the center of gravity is located longitudinally at the center of the front and rear axles of the vehicle. The complete calculation process for realizing the function of the first center of gravity state perception module in this embodiment is executed by the vehicle controller (ECU) every 10 milliseconds according to a preset cycle. The specific steps are as follows:

[0040] 1.1) The initial input is the dynamic axle load at the four corners, which is obtained in real time from the signal conditioning circuits of four independent load cells and has undergone digital processing. , , , The current value of}. Simultaneously, the preset vehicle track width W, vehicle wheelbase L, and constant values ​​of gravitational acceleration g are read from the controller's non-volatile memory;

[0041] 1.2) Total Mass Calculation Steps: The four input corner dynamic axle load values ​​are arithmetically summed to obtain the total vertical load force. Then, this total vertical load force is divided by the gravitational acceleration g to obtain the vehicle's total mass for the current cycle. .

[0042] 1.3) Steps for calculating the transverse centroid: Calculate the total axle load on the right side, i.e. and The sum. Calculate the total axle load on the left side, i.e. and Sum. Calculate the difference in total axle load between the left and right sides. Multiply this difference by half the vehicle's track width W. Calculate the total vehicle mass. Multiply by the acceleration due to gravity g. Divide the result of "multiplying this difference by half the vehicle's wheelbase W" by the "total vehicle mass". Multiplying the result by the gravitational acceleration g yields the final transverse centroid coordinates. .

[0043] 1.4) Steps for calculating the longitudinal centroid: a. Calculate the total axle load of the front axle, i.e. and The sum. b. Calculate the total axle load of the rear axle, i.e. and c. Calculate the difference in total axle load between the front and rear axles. d. Multiply this difference by half the vehicle wheelbase L. e. Divide the result of step d by the calculated total vehicle mass. The product of this and the gravitational acceleration g yields the final longitudinal centroid coordinates. .

[0044] 1.5) The transverse centroid coordinates obtained within this calculation period and longitudinal centroid coordinates As a two-dimensional coordinate pair { , The data is output to a shared memory area or broadcast via the vehicle's CAN bus for use by the risk source decoupling module in the device of this invention. In the calculation process of this embodiment, the core fusion mechanism is reflected in the structured fusion of multiple independent sensor data through a rigid body static model to derive the system-level lateral centroid coordinates. and longitudinal centroid coordinates This embodiment employs a fusion method driven by a physics model. Specifically, it uses the principle of rigid body static torque balance as the fusion framework. This method was chosen because, for the low-speed operation of the parade floats, the dynamic acceleration effect of the vehicles is negligible, and their force state highly conforms to static or quasi-static equilibrium conditions. Therefore, using the static torque balance equation as the fusion model has significant advantages such as clear physical meaning, simple calculation, strong robustness, and no need for extensive data training, making it suitable for real-time applications in vehicle-mounted embedded systems.

[0045] Lateral centroid coordinates The relationship with the dynamic axle load at the four corners is as follows: With the total axle load on the right side Positively correlated with the total axis load on the left. They are negatively correlated. That is, when the load on the right side increases or the load on the left side decreases, the center of mass will shift to the right.

[0046] Longitudinal centroid coordinates The relationship with the dynamic axle load at the four corners is as follows: Total axle load of the rear axle Positively correlated with the total axle load of the front axle There is a negative correlation. That is, when the rear axle load increases or the front axle load decreases, the center of mass will shift rearward. The normalized center of mass shift is denoted as... Normalized centroid offset The value range is limited to the interval [0, 1], normalized centroid offset. The method for obtaining this information is as follows: Based on vehicle design and safety specifications, a lateral safety boundary for the permissible offset of the vehicle's center of gravity is pre-defined. and longitudinal security boundary These two boundary values ​​represent the maximum permissible distance that the center of mass can deviate in any direction. Exceeding this distance indicates that the vehicle is in a high-risk state.

[0047] The real-time calculated lateral centroid coordinates The absolute value divided by the lateral safety boundary The normalized lateral offset is obtained. The vertical centroid coordinates The absolute value divided by the longitudinal safety boundary The normalized longitudinal offset is obtained. The Euclidean norm is used to fuse the two normalized offsets to calculate the final normalized centroid offset. The specific calculation logic is as follows: calculate the normalized lateral offset. Square and normalized longitudinal offset The sum of the squares is then taken, and the square root of that sum is calculated. To ensure that the output value range is strictly within [0, 1], the calculation result is saturated; that is, if the calculated value is greater than 1, the centroid offset is normalized. Take 1.

[0048] Normalized centroid offset The closer the value is to 0, the closer the real-time centroid position of the float is to its ideal geometric center, the more balanced and stable the mass distribution of the vehicle is, and the lower the risk of quasi-static roll caused by internal load distribution.

[0049] Normalized centroid offset The closer the value is to 1, the greater the offset between the real-time centroid position and the geometric center of the float, or the greater the deviation from the preset safety boundary. This indicates a more uneven distribution of vehicle mass, making it more prone to or about to enter a high-risk tilting state dominated by internal load distribution. This suggests that subsequent risk assessment and intervention systems need to be on high alert or activated. This affects the normalized centroid offset of the final output value. The key input parameter is the dynamic axle load at the four corners. , , , Its influence mechanism is through the intermediate variable, the lateral centroid coordinate. and longitudinal centroid coordinates Conduction.

[0050] For the total axle load on the right side normalized centroid offset The impact; when other axle load parameters remain unchanged, the total axle load on the right side The increase will lead to changes in the lateral centroid coordinates. It monotonically increases in the positive direction (representing the right side), thus leading to the normalized centroid offset. Monotonically increasing. This is a positive correlation. According to the principle of static torque balance, the torques on a vehicle about its longitudinal axis must be balanced. When the load on the right axle increases, to maintain torque balance, the center of mass of the vehicle's total mass, which is the equivalent point of application, must shift to the right to generate a gravitational torque that counteracts the unbalanced torque of the axle load. This algorithm design accurately maps this fundamental physical law, ensuring the physical accuracy of the center of mass calculation.

[0051] For the total axle load on the left normalized centroid offset Impact: When other axle load parameters remain unchanged, the total axle load on the left side The increase will lead to changes in the lateral centroid coordinates. The absolute value increases monotonically in the negative direction (representing the left side), which in turn leads to the normalized centroid shift. Monotonically increasing. This is also a positive correlation. The increase in load on the left forces the center of mass to shift to the left to maintain moment balance. Regardless of whether the center of mass shifts to the left or right, its distance from the center increases, therefore the normalized center of mass offset, representing the overall risk, increases. And so it increases.

[0052] For the difference between front and rear axle loads normalized centroid offset Impact: An increase in the absolute value of the difference between the front and rear axle loads will lead to an increase in the longitudinal centroid coordinate. The absolute value of increases monotonically, leading to a normalized centroid shift. Monotonically increasing, indicating a positive correlation. Also based on the principle of torque balance around the vehicle's lateral axis, the greater the difference in axle load between the front and rear axles, the more significant the longitudinal shift of the center of gravity. The algorithm integrates lateral and longitudinal shifts using the Euclidean norm, ensuring that any deterioration in shift direction is factored into the final comprehensive risk assessment. To verify the effectiveness of this technology, six typical float load distribution scenarios were designed for testing. Preset vehicle parameters: Lateral safety boundary. The longitudinal safety boundary is 0.5 meters. The value is 1.0 meter, as shown in Table 1 below:

[0053] Table 1. Test table of load distribution on floats in various scenarios:

[0054]

[0055] The above experimental data clearly demonstrate the accurate sensing capability and quantitative evaluation effect of the "first centroid state sensing module" of this invention.

[0056] Comparing Scenario 1 and Scenario 2: In the initial equilibrium state (Scenario 1), the loads on the four corner axes are equal, and the calculated values ​​are... and All are 0, the final normalized centroid offset The value is also 0, accurately reflecting the ideal stable state of the vehicle. When the group of performers moves to the right (Scene 2), the total axle load on the right side is significantly greater than that on the left. The algorithm accurately calculates the center of mass shifting to the right by 0.344 meters and gives a risk assessment value of 0.710. This strongly demonstrates the algorithm's sensitivity and quantification accuracy to lateral load transfer, verifying the positive correlation conclusion regarding the total axle load on the right side in the first step of the analysis.

[0057] Comparing Scenario 2 and Scenario 3: Scenario 3 simulates the performer moving to the extreme left position. At this point, the total axle load on the left is much greater than on the right, and the algorithm calculates... The value is -0.489 meters, and its absolute value is close to the lateral safety boundary of 0.5 meters. The final output is the normalized centroid offset. The value is as high as 0.979, indicating that the system can accurately identify that the vehicle is in a critical roll risk state caused by internal load transfer. This is in complete agreement with the theoretical analysis in the first step, demonstrating the significant advantage of this invention in early warning of extreme risks.

[0058] Comparing Scenario 4 and Scenario 5: Scenario 4 demonstrates a pure longitudinal load transfer, where the algorithm accurately captures the longitudinal centroid shift and provides the corresponding risk value. Scenario 5, however, simulates a complex condition where the load concentrates on the right front diagonal, resulting in significant lateral and longitudinal shift risks. The algorithm of this invention, through Euclidean norm fusion, provides a comprehensive risk score of 0.978, demonstrating that this invention possesses a comprehensive ability to perceive and assess multi-dimensional, coupled centroid shift risks. Compared to existing technologies that can only monitor a single lateral or longitudinal indicator, this invention can more realistically and comprehensively reflect the overall stability of the vehicle, thereby avoiding risk omissions due to neglecting coupling effects.

[0059] To further explain, the real-time grounding margin is a quantitative indicator that is proportional to the amplitude of the angular deceleration response generated under the action of a probing braking torque pulse. The amplitude of the angular deceleration response is determined by performing linear regression analysis on the angular velocity time series data collected during the pulse action and taking the absolute value of the slope of the fitted straight line. The "quantitative indicator that is proportional to the amplitude of the angular deceleration response" is defined as the normalized grounding margin, which is obtained by dividing the real-time calculated amplitude of the angular deceleration response by a preset maximum expected angular deceleration. The maximum expected angular deceleration is determined by an offline calibration method, which includes: applying a probing braking torque pulse under the maximum single-wheel load allowed by the vehicle design and measuring the corresponding amplitude of the angular deceleration response. This embodiment relates to the second grounding margin detection module in the device for actively and directly quantifying and evaluating the real-time grounding status of each wheel. The core function of this second grounding margin detection module is to precisely apply a standardized, driver-insensitive probing braking torque pulse to the brake of a selected wheel, and accurately capture the resulting dynamic angular velocity response of the wheel. Based on a defined rotational dynamics model, it calculates a quantitative index that is strongly positively correlated with the real-time vertical load borne by the wheel, namely the normalized grounding margin, which directly characterizes the tightness and stability of the wheel's contact with the ground. In this embodiment, the key parameters used to determine the real-time grounding margin of the wheel are defined as follows:

[0060] Detective braking torque pulse, its parameter symbol is Physically, this pulse is a brief, fixed-amplitude braking torque, measured in Newton-meters, actively applied to the brakes of the target wheel by the vehicle controller. The amplitude and duration of this pulse are set within a range that does not affect normal vehicle operation or passenger comfort; its function is to serve as an active excitation signal to detect the dynamic response characteristics of the wheel.

[0061] Instantaneous wheel angular velocity, its parameter symbol is Its physical meaning is the angular velocity of the wheel at time t, measured in radians per second; it describes the dynamic behavior of the wheel during the detection process. The angular deceleration response amplitude, its parameter symbol is... Its physical meaning is the rate of change of wheel angular velocity during the application of a probing braking torque pulse, i.e., the absolute value of angular acceleration, measured in radians per square second. Maximum expected angular deceleration, its parameter symbol is... Its physical meaning is the theoretical maximum angular deceleration response amplitude that can be generated when the same probing braking torque pulse is applied under the maximum single-wheel load allowed by the vehicle design.

[0062] Normalized grounding margin, its parameter symbol is Its physical meaning is a dimensionless quantitative index with a value range of [0, 1], used to characterize the real-time ground stability of the target wheel as determined by active detection. The larger the value, the closer the wheel is to the ground and the greater the vertical load it bears.

[0063] Detection braking torque pulse The acquisition method is as follows: the vehicle's electronic control unit (ECU) applies a precise current or hydraulic command to the brake caliper of the target wheel by controlling the hydraulic pump or motor of the braking system according to a preset program. This command corresponds to a preset braking torque amplitude and duration. In this embodiment, the preset braking torque amplitude is 50 Nm and the duration is 20 milliseconds. In this embodiment, the "control braking system" is an electro-hydraulic braking system (EHB).

[0064] Instantaneous wheel angular velocity The data is acquired through the wheel speed sensors of the vehicle's standard anti-lock braking system (ABS). To ensure accuracy, the controller continuously reads the sensor's output signal during the detection period at a high sampling frequency of 1000Hz.

[0065] angular deceleration response amplitude Based on fundamental kinematic principles, acceleration is the first derivative of velocity with respect to time. The actual calculation method is as follows: the controller captures instantaneous wheel angular velocity time-series data points within the time window of the probing braking torque pulse; then, the least squares method is used to perform linear fitting on these data points to obtain an optimally fitted straight line; the absolute value of the slope of this line is determined as the amplitude of the angular deceleration response. .

[0066] Maximum expected angular deceleration The determination method is an offline experimental calibration method: the float is placed on a test bench capable of accurately measuring and applying vertical loads on a single wheel; in this embodiment, a four-wheel alignment machine or a dedicated loading device is used; the vertical load on one wheel is gradually increased to its design limit load; under this limit load, standardized probing braking torque pulses are repeatedly applied. Repeatedly measure the corresponding angular deceleration response amplitude. The average value is taken as the maximum expected angular deceleration of the wheel. This information is stored as a constant in the ECU. Each wheel position can be calibrated independently.

[0067] Normalized grounding margin It originates from data standardization processing methods. The calculation method involves taking the real-time calculated angular deceleration response amplitude... Divide by the pre-calibrated and stored maximum expected angular deceleration For normalized grounding margin The calculation results are saturated; that is, if the output value is greater than 1, the final output value is 1.

[0068] This embodiment determines the maximum expected angular deceleration of a certain wheel through offline calibration. for During online detection, the controller applied a standard probing braking torque pulse and acquired a set of angular velocity data. The current angular deceleration response amplitude was calculated through linear fitting. for Then the normalized grounding margin at that moment. The calculation result is: 12.0 / 15.0 = 0.8. This result indicates that the current wheel's contact margin is 80% of its maximum safety margin.

[0069] The complete calculation process for implementing the function of the second ground margin detection module in this embodiment is executed alternately by the vehicle controller (ECU) for the inner and outer wheels every 100 milliseconds according to a preset period. The specific steps are as follows:

[0070] 2.1) Triggered by a periodic timer. The controller first sends a precise control command to the brake actuator of the target wheel to generate a probe braking torque pulse with a preset amplitude and duration. 2.2) Simultaneously with issuing the command, the controller initiates a high-frequency data acquisition program for the wheel speed sensor, continuously acquiring and buffering a series of instantaneous wheel angular velocities within a time window longer than the pulse duration. 2.3) After the data acquisition window ends, the controller performs a linear regression algorithm on the cached angular velocity time series data to calculate the slope of the best-fit line, and uses the absolute value of the slope as the amplitude of the angular deceleration response for this detection. 2.4) The controller reads the pre-calibrated maximum expected angular deceleration corresponding to the wheel from memory. The value. Then, the obtained real-time angular deceleration response amplitude. Divide by the maximum expected angular deceleration 2.5) Perform an upper limit saturation check on the division result; if the result is greater than 1, set it to 1. The final value in the interval [0, 1] is determined as the normalized grounding margin for the current period. It is then output to shared memory or broadcast via the CAN bus for use by other functional modules such as the risk source decoupling module. When the same probing braking torque pulse is applied... According to Newton's second law of rotation, the resulting angular deceleration is inversely proportional to the effective moment of inertia of the wheel. The effective moment of inertia of a wheel includes not only its own moment of inertia but also the rolling resistance generated by the contact between the tire and the ground, and potential slippage; both of these are positively correlated with the vertical load. Therefore, the greater the vertical load, the stronger the wheel's "resistance" to braking torque, and the greater the resulting angular deceleration.

[0071] Further explanation: For the risk source decoupling module, the expected grounding margin is calculated by inputting the normalized centroid offset into a preset benchmark model; the preset benchmark model is a lookup table established through offline experimental calibration that stores the mapping relationship between the normalized centroid offset and the corresponding expected normalized grounding margin; the predicted residual signal is a risk source characteristic spectrum separation degree, which is determined by calculating the absolute value of the difference between the normalized grounding margin and the expected normalized grounding margin.

[0072] The calculation steps for the risk source feature spectrum separation degree also include: comparing the "absolute value of the difference" with a preset insensitivity threshold; only when the absolute value is greater than the insensitivity threshold is it output, otherwise the output is zero. The core function of the risk source decoupling module is to construct a comparison framework based on model prediction and physical measurement; specifically, using the normalized centroid offset output by the first centroid state perception module, which characterizes the macroscopic mass distribution of the vehicle, the expected normalized grounding margin, determined only by internal load transfer, is calculated through a preset benchmark model. Subsequently, this expected value is compared in real time with the normalized grounding margin actively detected and output by the second grounding margin detection module, which reflects the actual physical contact state of the wheel. The innovation of the "risk source decoupling module" lies in generating a novel quantitative index with a strictly limited value range through a robust norm calculation, denoted as the risk source feature spectrum separation degree. The risk source feature spectrum separation degree can accurately and stably quantify the inconsistency between the expected state and the actual state. Its value directly maps the contribution of unpredictable external risk sources, thereby achieving clear decoupling and feature separation of internal and external overturning risk sources. In this embodiment, the key parameters used to generate the prediction residual signal are defined as follows:

[0073] The expected normalized grounding margin, with parameter sign as follows: Its physical / logical meaning is a dimensionless index taking values ​​in the interval [0, 1]. It characterizes the ground stability of the target wheel under the current normalized centroid offset, based on the inherent physical characteristics of the vehicle. This parameter is a theoretical prediction based on internal risk sources. Internal risk sources include mass transfer;

[0074] Risk source characteristic spectrum separation degree, its parameter sign is Its physical / logical meaning is a dimensionless index that takes values ​​in the interval [0, 1], serving as the final output of this risk source decoupling module. It is used to quantify the degree of deviation between the expected normalized grounding margin and the actual measured normalized grounding margin. It is the core criterion for distinguishing whether overturning risk originates from internal predictable factors or external unforeseen factors.

[0075] The insensitivity threshold, whose parameter symbol is: Its physical / logical meaning is a preset, small positive value close to zero. In this example, the value is 0.05, which is used to filter out residual fluctuations that are meaningless due to sensor noise or small model errors, so as to enhance the stability of the output signal.

[0076] Expected normalized grounding margin The method for obtaining this is by calling a preset benchmark model for calculation. This benchmark model is essentially a normalized centroid offset of the input. Expected normalized ground margin mapped to the output The model is established using an offline experimental calibration method: In this embodiment, under controlled experimental conditions, the position of the vehicle's center of gravity is systematically changed by precisely moving the counterweights inside the float, thus adjusting the normalized center of gravity offset. Iterate from 0 to 1.

[0077] At each set normalized centroid offset Under these conditions, with the vehicle stationary and the road surface flat, external risk sources are eliminated. The second grounding margin detection module is used to perform multiple measurements on the target wheel, and the normalized grounding margin is obtained. The stable average value.

[0078] A series Data pairs are recorded, and a one-dimensional lookup table is constructed using polynomial fitting or piecewise linear interpolation. This lookup table serves as the pre-defined baseline model and is stored in the memory of the electronic control unit (ECU). Risk source characteristic spectrum separation. Derived from the norm, or Manhattan distance, and incorporating thresholding logic, this invention improves upon it to ensure the robustness and explicit physical meaning of the output. The calculation method is as follows: calculate the expected normalized ground margin. With normalized grounding margin The absolute value of the difference between them; secondly, the absolute value is compared with a preset insensitivity threshold. The absolute value is compared with the insensitive threshold. If the absolute value is less than or equal to the insensitive threshold, the output of the risk source feature spectrum separation degree is 0; otherwise, the output is the absolute value.

[0079] Risk source feature spectrum separation The closer the value is to 0, the better the measured grounding condition matches the prediction based on the internal mass distribution. The system determines that the current overturning risk mainly originates from internal, predictable load transfer. Specifically, this indicates that the physical cause of the current overturning risk mainly stems from the transfer of loads within the vehicle, and this risk source has the characteristics of gradual change and predictability. The transfer of loads within the vehicle (such as performer movement) includes situations where the performer moves.

[0080] Risk source feature spectrum separation The closer the value is to 1, the more serious the deviation between the measured grounding state and the theoretical prediction, and the more the system focuses on determining that the vehicle is being disturbed by external risk sources. In this embodiment, external risk sources include one side of the wheel driving onto the shoulder, getting stuck in a pothole, or encountering strong crosswinds. These risk sources have the characteristics of being instantaneous and unpredictable.

[0081] In this embodiment, the data of the reference model (LUT) stored in the vehicle controller ECU at a certain input point is: when the normalized centroid offset is... When it is 0.7, the corresponding expected normalized grounding margin is The threshold value is 0.8. This is the preset insensitivity threshold. It is 0.05.

[0082] Scenario 1 is an endogenous risk: the output of the first centroid state perception module. The second ground clearance detection module outputs detection data for the wheels on the same side. The controller queries the baseline model to obtain... Calculate the absolute value of the difference: .because Therefore, the final output risk source feature spectrum separation degree It is 0.

[0083] Scenario 2 is an external risk: the output of the first centroid state perception module. However, because the wheel instantly drove onto a shoulder, the second ground clearance detection module detected and output... The controller queries the baseline model to obtain... Calculate the absolute value of the difference: .because Therefore, the final output risk source feature spectrum separation degree It is 0.55.

[0084] In this embodiment, the risk source decoupling module is executed synchronously by the vehicle control unit (ECU) after receiving a new round of perception data. The specific steps are as follows:

[0085] 3.1) The input consists of two parallel real-time data streams: the latest normalized centroid offset obtained from the first centroid state sensing module. And the latest normalized grounding margin for a specific target wheel obtained from the second grounding margin detection module. Simultaneously, the preset baseline model and insensitivity threshold are read from the controller memory. 3.2) Normalize the input centroid offset. As a query index, it is used to perform lookups or interpolation calculations in the baseline model to obtain the corresponding expected normalized ground margin. 3.3) Calculate the normalized grounding margin. With expected normalized grounding margin 3.4) Calculate the arithmetic difference between the two values ​​and take its absolute value to obtain a temporary residual value. Then, compare the obtained temporary residual value with a preset insensitivity threshold. The comparison is performed. If the temporary residual value is less than or equal to the insensitivity threshold, the final output value is set to 0. If the temporary residual value is greater than the insensitivity threshold, the temporary residual value is directly used as the final output value. The determined value is used as the risk source feature spectrum separation degree for the current period. The output is sent to a shared memory area or broadcast via the CAN bus for use by the subsequent braking control strategy decision module. The output value is formatted as a floating-point number in the range [0, 1]. The core fusion method used in this embodiment is model-based residual analysis. Specifically, it utilizes the macroscopic state perception based on the centroid, represented by the information channel, to establish an "internal model" of the expected state of the physical world and generate the expected normalized grounding margin. Then, another independent channel of microstate sensing information, obtained through direct physical detection and active probing, is used as the "external truth," which is the normalized ground margin. This embodiment abandons the normalization method in traditional residual calculation, which may lead to numerical instability due to division by near-zero predicted values. Instead, it calculates the absolute value of the difference and combines it with threshold denoising. This ensures that the absolute value of the difference naturally falls within the [0, 1] interval, thereby directly generating a risk source feature spectrum separation degree with clear physical meaning, numerical stability, and no need for further normalization. .

[0086] Insensitivity threshold The threshold is pre-set after offline statistical analysis of the system noise level to balance detection sensitivity and signal stability. Determining the insensitivity threshold involves: acquiring sensor output under static conditions with no effective input signal, and statistically analyzing the mean and standard deviation of the output; multiplying the standard deviation by a preset multiplier factor as the initial threshold; verifying the stability and sensitivity of the initial threshold in a simulation environment, and correcting it based on real-vehicle test results to finally determine the insensitivity threshold, where the multiplier factor is between 2 and 3. Expected normalized ground margin. With normalized centroid offset There is a positively correlated nonlinear relationship between them. The further the center of mass shifts towards the target side, the greater the load on the wheel on that side, and the higher its expected ground contact margin. The final output is the risk source characteristic spectrum separation degree. The output shows a piecewise linear positive correlation with the absolute value of the predicted residual. When the absolute value of the residual is less than or equal to the insensitivity threshold, the output is 0; when it is greater than the insensitivity threshold, the output is equal to the absolute value of the residual. This indicates that only significant and meaningful deviations will be identified as external risk signals.

[0087] Further details regarding the risk source decoupling module: Factors affecting the final output risk source feature spectrum separation degree. The key input parameter is the normalized centroid offset. and normalized grounding margin The specific analysis and reasons are as follows:

[0088] For normalized centroid offset Separation of risk source characteristic spectrum Impact Analysis: Normalized Centroid Shift Separation from the risk source characteristic spectrum There is no direct monotonic relationship between them. Normalized centroid offset. Its function is to serve as input to the baseline model, used to generate the expected normalized ground margin that characterizes the "expected baseline". Regardless of the normalized centroid offset Whether it's large or small, it represents the measured normalized grounding margin, regardless of whether the internal mass distribution of the float is uniform. With the normalized centroid offset Corresponding expected normalized grounding margin Matching, the final risk source feature spectrum separation degree All of these will approach 0. This ensures that the algorithm focuses on the "deviation between expectation and reality," rather than the magnitude of any single state variable. This accurately achieves the technical objective of this invention: to isolate the sources of risk, rather than simply assessing the magnitude of the risk.

[0089] For normalized grounding margin Separation of risk source characteristic spectrum Impact analysis: Normalized centroid shift Normalized grounding margin when unchanged Separation from the risk source characteristic spectrum There exists a normalized grounding margin based on expectations. The relationship is a "V"-shaped nonlinearity centered on the grounding margin. That is, when the normalized grounding margin... From the expected normalized grounding margin When deviating in either 0 or 1, the separation degree of the risk source characteristic spectrum All are monotonically increasing. Expected normalized grounding margin. This represents the "normal" grounding state under the current internal load distribution. Any measured value deviating from this normal state—whether it's an abnormal decrease in grounding margin (including but not limited to wheels being lifted off the ground) or an abnormal increase (including but not limited to wheels hitting obstacles causing a sudden surge in vertical force)—indicates the occurrence of an external anomaly. This "V"-shaped relationship ensures that the algorithm has equal sensitivity to any form of grounding state anomaly caused by external factors. To verify the effectiveness of this technical feature, six representative float driving scenarios were designed for testing. A preset insensitivity threshold was used. The value is 0.05; specific test data are shown in Table 2 below:

[0090] Table 2. Test Table for Parade Float Driving Scenarios:

[0091]

[0092] In scenarios one, two, and four, the calculated absolute values ​​of the residuals are 0.02, 0.03, and 0.02, respectively, all less than or equal to the preset insensitivity threshold of 0.05. Therefore, the final output risk source feature spectrum separation degree is... The result was processed to 0.000. The above experimental data strongly demonstrates the outstanding substantive features and significant progress of the "Risk Source Decoupling Module" of this invention in accurately identifying and quantifying risks from different sources.

[0093] 1. Comparing Scenario 2 and Scenario 3: In both scenarios, the vehicle's internal mass distribution is exactly the same, and the normalized center of mass offset is... Both are 0.60, therefore the model predicts the expected normalized grounding margin. The same applies. However, in scenario two, the measured normalized grounding margin... The system accurately outputs the separation degree of the risk source characteristic spectrum, which is in high agreement with the expected value. With a value of 0.000, it can be accurately determined that the risk originates solely from internal load transfer.

[0094] In scenario three, the normalized grounding margin is reduced because the right wheel drives onto the shoulder. The value plummeted to 0.15, and the system immediately calculated a risk source characteristic spectrum separation of 0.330. This comparison decisively demonstrates that the present invention can accurately isolate and quantify the additional risks introduced by external road surface disturbances within the same internal risk context.

[0095] 2. Compare Scenario 5 and Scenario 6: Both scenarios involve severe external impacts, such as a revolver getting stuck in a dent.

[0096] In scenario five, the internal loads are balanced (normalized centroid offset). (0.10), the risk source feature spectrum separation degree calculated by the system. The value of 0.600 accurately quantifies this pure and intense external risk.

[0097] In scenario six, the vehicle also experiences moderate internal load transfer ( (0.60) and external shocks. The system output risk source characteristic spectrum separation degree. The value is 0.480. This figure clearly demonstrates that even when internal risks already exist, the system can still identify and quantify the significant "additional" risks brought about by external events. It accurately reflects the gap between reality and "expectations based on the current internal state," which is the core value of this invention in decoupling risk sources and provides precise quantitative input for subsequent hybrid control strategies.

[0098] Further explanation: The adaptive selection module for control strategy is specifically used to: calculate a control mode authority factor based on the predicted residual signal, and use this control mode authority factor to weight and fuse the outputs of the feedforward stable control strategy and the feedback stable control strategy to generate a final synthesized stable torque; the control mode authority factor is calculated by comparing the predicted residual signal with a preset authority factor activation threshold and a preset authority factor saturation threshold, and then using a piecewise linear mapping function.

[0099] The specific method of weighted fusion is as follows: multiply the output of the feedforward stabilization control strategy by the difference between "1" and the control mode authority factor, multiply the output of the feedback stabilization control strategy by the control mode authority factor, and add the two products together; the output of the feedforward stabilization control strategy is based on the parallel calculation of the real-time centroid position, and the output of the feedback stabilization control strategy is based on the parallel calculation of the real-time grounding margin.

[0100] Further explanation: The output of the feedforward stabilization control strategy is obtained by acquiring the offset and rate of change of the real-time centroid position, and performing lookup and interpolation calculations in a pre-calibrated feedforward control lookup table; the output of the feedback stabilization control strategy is obtained by calculating the error between the minimum real-time grounding margin and a preset safety target value, and inputting this error into a proportional-integral-derivative controller for calculation; the braking torque distribution execution module first determines the dangerous wheel with the smallest real-time grounding margin, and then determines the diagonal wheel of the dangerous wheel as the target braking wheel; the braking torque distribution execution module will calculate the target braking torque based on the final synthesized stabilizing torque and a preset braking torque distribution ratio, and send it to the target braking wheel.

[0101] The core innovation of the adaptive control strategy selection module and the braking torque allocation execution module lies in constructing a smooth weighted fusion control framework based on a control mode authority factor. Specifically, the adaptive control strategy selection module calculates a continuously varying control mode authority factor in the [0, 1] interval in real time, based on the risk source feature spectrum separation degree, which characterizes the contribution of external risks, provided by the upstream module. This control mode authority factor serves as a dynamic weight for the smooth weighted fusion of a parallel feedforward stable control torque based on the real-time centroid position and a feedback stable control torque based on the real-time grounding margin, thereby generating a final synthesized stable torque. The braking torque allocation execution module further applies this final synthesized stable torque precisely to the diagonal wheel opposite the wheel with the worst grounding state, according to the principle of optimal vehicle dynamics and a preset allocation ratio. This design elevates the discrete strategy "selection" to a continuous strategy "fusion," achieving a shock-free and smooth transition from pure feedforward to pure feedback control, and ensuring that the braking intervention force and the nature and intensity of the risk source are always optimally matched. In this embodiment, the key parameters and strategy implementation methods used to achieve adaptive selection and execution of the control strategy are defined as follows:

[0102] Control mode permission factor, its parameter symbol is Its physical / logical meaning is a dimensionless weighting coefficient that takes values ​​in the interval [0, 1]. It is used to determine the contribution weight of the feedback stability control strategy in the final synthesized stability torque. When its value is 0, it indicates that the system is completely in feedforward control mode; when its value is 1, it indicates that the system is completely in feedback control mode.

[0103] The activation threshold for the permission factor, whose parameter symbol is: Its physical / logical meaning is a pre-defined characteristic that suggests external risks can be identified. The lower limit value is 0.25 in this embodiment.

[0104] The saturation threshold for the permission factor, whose parameter sign is... Its physical / logical meaning is a pre-defined risk source characteristic spectrum separation degree that characterizes an external risk level that has reached an extremely dangerous level. The upper limit value is 0.55 in this embodiment.

[0105] The final resultant stabilizing moment has the following parameter symbol: Its physical / logical meaning is the total braking torque that needs to be executed in the current control cycle after weighted fusion of the feedforward and feedback basic torques.

[0106] Dangerous wheel index, its parameter symbol is Its physical / logical meaning is the unique identifier corresponding to the wheel with the smallest normalized grounding margin among all current wheels.

[0107] Diagonal wheel index, its parameter symbol is Its physical / logical meaning is based on the dangerous wheel index. A unique identifier for the wheel located on the geometric diagonal of the vehicle, determined by looking up a table.

[0108] Braking torque distribution ratio, its parameter symbol is Its physical / logical meaning is a preset proportional coefficient greater than 0.5 and less than or equal to 1. In this embodiment, the value is 0.85.

[0109] Furthermore, 4.1) The specific implementation of the feedforward stability control strategy is as follows: The core objective of the feedforward stability control strategy in this embodiment is to calculate and output a feedforward control base torque to counteract the expected roll moment in advance, smoothly, based on predictable internal risk sources. .

[0110] Normalized centroid shift rate of change, its parameter sign is Its physical / logical meaning is the rate of change of the normalized centroid offset over time, characterizing the speed of internal load transfer. Its value range is normalized to the interval [-1, 1], with positive values ​​indicating accelerated centroid offset to one side and negative values ​​indicating decelerated offset to that side or offset to the opposite side. This parameter is obtained by measuring the normalized centroid offset over two consecutive control cycles. The value is calculated by performing backward difference and then dividing by the control cycle duration.

[0111] Feedforward control lookup table, its parameter symbols are Its physical / logical meaning is a two-dimensional data table stored in the controller's non-volatile memory. The two input axes of this table are the discretized normalized centroid offset. and normalized centroid shift rate of change The values ​​stored in the table are the calibrated feedforward control base torques required to maintain vehicle stability under the corresponding input combinations. In this embodiment, the feedforward control lookup table is generated and calibrated through offline modeling and simulation testing using professional vehicle dynamics simulation software such as CarSim or TruckSim, ensuring the accuracy and reliability of the data.

[0112] The feedforward control base torque, its parameter symbol is Its physical / logical meaning is the braking torque value finally output by the feedforward stabilization control strategy of this embodiment within the current control cycle;

[0113] The calculation process of the feedforward stabilization control strategy is as follows: At the beginning of each control cycle, obtain the normalized centroid offset at the current moment. Read the normalized centroid offset from memory from the previous control cycle. The value is obtained by differential calculation to obtain the current normalized centroid shift rate of change. .

[0114] The current normalized centroid offset and normalized centroid shift rate of change As coordinates, in the feedforward control lookup table The query is performed within the range of [the data type]. A bilinear interpolation algorithm is used to calculate an accurate, smoothly transitioning torque value based on the torque values ​​at the four nearest grid points surrounding the coordinate point. The interpolated torque value is then used as the feedforward control baseline torque for this cycle. Output is then generated. By simultaneously considering the "position" and "velocity" of the center of mass, the tendency to tilt in the very short term can be accurately predicted, thus allowing a moderately large and gently applied stabilizing torque to be applied in advance. The beneficial effects are: because the intervention is early and gentle, the float passengers will hardly notice the braking system's intervention; and efficient energy utilization avoids energy waste caused by overreaction or delayed response.

[0115] 4.2) The specific implementation of the feedback stabilization control strategy is as follows: The core objective of the feedback stabilization control strategy in this embodiment is to respond as quickly as possible when the vehicle encounters an unpredictable external risk source that causes the wheels to tend to lift off the ground, and to calculate and output the feedback control base torque for emergency attitude correction. Its specific implementation employs a proportional-integral-derivative (PID) closed-loop control algorithm; the key parameters of the feedback stability control strategy are defined as follows:

[0116] Minimum normalized grounding margin, its parameter symbol is Its physical / logical meaning is the normalized grounding margin of all wheels within the current control cycle. The minimum value selected from the array. This parameter directly reflects the ground contact status of the vehicle's most dangerous wheel.

[0117] The grounding margin safety target value, its parameter symbol is: Its physical / logical meaning is a preset minimum grounding margin lower limit that the system expects to maintain. It is a constant greater than 0 and close to 1. In this embodiment, the value is 0.9, which represents an absolutely safe grounding state.

[0118] Grounding margin error, its parameter symbol is Its physical / logical meaning is the grounding margin safety target value. With minimum normalized grounding margin The difference between the two values. This error is the direct input to the PID controller.

[0119] proportional gain Integral gain Differential gain The physical / logical meaning of these parameters refers to the three core control parameters of the PID controller. These parameters are systematically calibrated offline on a professional hardware-in-the-loop (HIL) test platform using the Ziegler-Nichols tuning method to ensure that the control system has excellent dynamic performance with fast response, small overshoot, and low steady-state error.

[0120] Feedback control of the basic torque, its parameter symbol is Its physical / logical meaning is the final output braking torque value calculated by this feedback stabilization control strategy within the current control cycle. The calculation process of the feedback stabilization control strategy is as follows: Obtain the normalized ground clearance margin of all wheels. An array is used, and the array is iterated through to determine the minimum normalized ground margin. Use the preset grounding margin safety target value. minus The current grounding margin error is obtained. .

[0121] grounding margin error Multiply by proportional gain This yields a control component proportional to the current error, used for rapid response. The grounding margin error for this cycle is then calculated. Add this to the total historical error, then multiply by the integral gain. This yields the control components used to eliminate the system's steady-state error. The grounding margin error between the current cycle and the previous cycle is calculated. The difference, multiplied by the differential gain This yields a control component related to the rate of change of error, which is used to suppress overshoot and oscillation.

[0122] The proportional, integral, and derivative components are added together to obtain the original control torque value. This value is then compared to the maximum / minimum physical torque that the vehicle's braking system can provide, and output saturation processing is performed to prevent the command from exceeding the actuator's capability. The torque value after saturation processing is used as the feedback control base torque for this cycle. Output the results.

[0123] Control Mode Permission Factor It originates from the piecewise linear mapping function in control theory. Its calculation method is as follows: [The text abruptly shifts to a different topic] ...the separation degree of the risk source characteristic spectrum... With permission factor activation threshold and permission factor saturation threshold Compare. If ,but ;like ,but If it falls between the two, then the control mode permission factor is... The value is equal to .

[0124] Control Mode Permission Factor As the value changes from 0 to 1, the technical effect is a smooth transfer of control from feedforward to feedback control. The final result is a stable torque. It originates from the linear weighted sum in signal processing. Its calculation method is as follows: [The text abruptly shifts to a different topic] ...the feedforward control base torque... Multiply by (1 minus the control mode permission factor) Simultaneously, feedback control of the basic torque will be applied. Multiply by control mode permission factor Finally, add the two products together.

[0125] Permission factor activation threshold With permission factor saturation threshold The value was determined through statistical analysis of a large amount of vehicle-in-the-loop simulation data. (This embodiment...) , .enter Parallel computation yields , Calculate the control mode permission factor. Calculate the final resultant stabilizing moment. .

[0126] The complete calculation process for implementing the adaptive selection and execution of the control strategy in this embodiment is executed by the vehicle controller (ECU) in each control cycle, and the specific steps are as follows:

[0127] The initial input consists of three parallel real-time data streams: risk source feature spectrum separation. Normalized centroid offset and the normalized grounding margin of all wheels. Array.

[0128] Define path A as the execution path of the feedforward stable control strategy: obtain the normalized centroid offset. Calculate its normalized centroid shift rate of change. Normalized centroid offset and normalized centroid shift rate of change As coordinates, a lookup table is used in the feedforward control. Bilinear interpolation is performed to calculate the feedforward control base torque. .

[0129] Define path B as the execution of the feedback stability control strategy: from the normalized ground margin Filter the array to find the minimum normalized ground margin ; Calculate its grounding margin safety target value Grounding margin error ; this grounding margin error The input is fed to a proportional-integral-derivative (PID) controller, and after calculation and limiting, the feedback control base torque is obtained. .

[0130] Separation of risk source feature spectrum The input is fed into a piecewise linear mapping function, and an activation threshold is set based on a preset permission factor. With permission factor saturation threshold The control mode permission factor for the current cycle is calculated. Applying the linear weighted sum formula, the obtained feedforward control base torque is... and feedback control basic torque With the obtained control mode permission factor By performing fusion calculations, the final composite stabilizing torque is obtained. In normalized grounding margin In the array, identify the wheel corresponding to the minimum value and record its dangerous wheel index. Based on the vehicle geometry mapping table, query and determine the diagonal wheel index. The final resultant stabilizing torque will be... Multiplied by the preset braking torque distribution ratio The target braking torque value is obtained and sent to the diagonal wheel index. The brake actuator for the specified wheel. This embodiment introduces a continuously varying control mode permission factor. By linearly combining the outputs of two control strategies, a smooth, shock-free transition from one dominant control mode to the other is achieved. This elevates the discrete "selection" to a continuous "ratio," improving the robustness, smoothness, and precision of the control system. The core "weight" in the fusion process is the control mode authority factor. Its value is calculated dynamically and in real time. The activation threshold of the permission factor determines its calculation behavior. With permission factor saturation threshold The braking torque distribution ratio is determined through model-based system design and data-driven optimization. The determination of the braking torque distribution ratio is based on the first principles of vehicle dynamics, and is achieved through theoretical analysis and simulation verification to determine the proportional value that maximizes the generation of stable yaw torque. The determination of the braking torque distribution ratio specifically includes:

[0131] (1) Establish a dynamic model based on the vehicle's yaw and side-slip motion to obtain the equivalent yaw torque expression for the vehicle under different tire braking forces; (2) Based on the dynamic model, select yaw rate error and ground clearance as vehicle stability indicators; (3) Perform parameter scanning on the braking torque distribution ratio within a preset range, and use vehicle dynamics simulation to calculate the stability index response under different parameters; In this embodiment, the braking torque distribution ratio is... Scanning is performed in the range of 0.0 to 0.5; (4) The braking torque distribution ratio that makes the vehicle stability index meet the preset optimization target is determined as the candidate value; In this embodiment, the ratio with the smallest yaw rate deviation and positive grounding margin under high-speed lane change conditions is selected as the candidate value; (5) The candidate value is verified and corrected in the actual vehicle test to obtain the final braking torque distribution ratio.

[0132] Final stabilizing torque With feedforward control base torque There is a negatively correlated weighted relationship between them, with a weight of . With feedback control basic torque There is a positively correlated weighted relationship between them; in this embodiment, the weights are controlled by the permission factor of the control mode. Characterization.

[0133] Control Mode Permission Factor Separation from the risk source characteristic spectrum There is a piecewise linear positive correlation between them, ensuring that the investment of control resources is precisely matched with the actual needs of risk. When the control mode authority factor... As the value approaches 0, the characterization system tends to judge the current roll risk source as being more predictable, favoring internal load transfer; ultimately, the resulting stabilizing moment... The calculated weight allocation will increasingly approximate the feedforward control base torque. This allows the system to prioritize smooth, energy-efficient, and model-based predictive interventions to ensure ride comfort as the primary objective.

[0134] When control mode permission factor The closer the value is to 1, the more the characterization system judges the current roll risk source to be an unpredictable external sudden disturbance; the final synthesized stabilizing moment... The calculated weight allocation will increasingly approximate the feedback control base torque. This causes the system to switch to an emergency correction mode characterized by the fastest response speed and the greatest intervention force, with ensuring ultimate safety as the sole objective. The core input parameter influencing the final decision is the separation degree of the risk source feature spectrum. Its control mode permission factor The impact analysis is as follows:

[0135] Risk source feature spectrum separation With control mode permission factor There exists a piecewise linear positive correlation between them. When the separation degree of the risk source characteristic spectrum... exist When the control mode permission factor is monotonically increasing within the interval, It also increases monotonically from 0 to 1.

[0136] Risk source feature spectrum separation The physical meaning of "contribution to external risk" is "contribution to external risk." This positive correlation design maps to the fundamental logic of vehicle control: as unknown external risks increase, the control system must gradually abandon reliance on model-based "prediction" and instead rely more on real-time measurement "feedback." This invention utilizes a control mode authority factor. This continuously changing permission factor avoids vehicle shaking or shock caused by sudden changes in control strategy near the switching point, ensuring that the control strategy ratio is always optimal under any risk mixing ratio. This is the core improvement of the smoothness and stability of the vehicle control system in this embodiment. To verify the beneficial effects of the technical solution of this invention, a series of typical working conditions were constructed for testing. Table 3 below shows the changes in key parameters and final output results of the core algorithm module of this invention under different risk source inputs. Table 3 aims to simulate three representative vehicle operation scenarios: (i) only internal load transfer exists (such as a performer moving in the vehicle); (ii) only encountering sudden external disturbances (such as strong crosswinds or uneven road surfaces); (iii) a composite working condition with both internal load transfer and moderate external disturbances. By recording two sets of independent test data under each scenario, the decision-making and execution behavior of this invention under different risk inputs is compared and analyzed to verify the effectiveness of its risk decoupling and strategy fusion, as shown in Table 3 below:

[0137] Table 3. Key Parameter Changes and Final Output Results:

[0138]

[0139] The threshold processing in this embodiment is already reflected in the control mode permission factor. In the calculation. In scenario one, the separation degree of the risk source feature spectrum. The values ​​were 0.15 and 0.16 respectively; both were below the permission factor activation threshold. The control mode permission factor is 0.25. It is processed to 0. In scenario two, the separation degree of the risk source feature spectrum is... The values ​​of 0.80 and 0.82, respectively, are both higher than the saturation threshold of the permission factor, which is 0.55. Therefore, control mode permission factor The value was processed to 1. The above experimental data strongly demonstrates the outstanding substantive features and significant progress of this invention:

[0140] Comparing the data from Scenario 1 and Scenario 2, it is evident that this invention can clearly distinguish the sources of risk. In Scenario 1, although there is a significant centroid shift (normalized centroid shift), Greater than 0.6), but because the system identifies the risk source as internal (risk source feature spectrum separation degree) (less than 0.25), therefore the control mode permission factor With a value of 0, the system relies entirely on feedforward control, outputting a smooth and stable torque of 650 Nm. However, in scenario two, although the center of mass hardly moves (normalized center of mass offset...), the system... (less than 0.05), but the system sensitively detects the sudden drop in ground margin caused by external disturbances (minimum normalized ground margin). The risk level was 0.65, and the risk was determined to originate from external sources (risk source characteristic spectrum separation). (Greater than 0.55). Therefore, the control mode permission factor Switching to 1, the system decisively calls all resources to execute feedback control, outputting a powerful emergency stabilizing torque of 2500Nm.

[0141] The data from Scenario 3 directly reflects the core innovation of this invention. Under this complex risk condition, the separation degree of the risk source characteristic spectrum... The value is 0.40, which is within the transition range. The system calculates the control mode permission factor. It is 0.50. At this point, the final resultant stabilizing torque is... The value is 760 Nm, which is neither simply the feedforward torque of 520 Nm nor simply the feedback torque of 1000 Nm, but rather a precise weighted sum of the two. This contrasts sharply with traditional control strategies based on hard threshold switching. The smooth fusion mechanism of this invention ensures a precise match between control strength and the actual composition of risk, achieving an optimal balance between safety and comfort. To transform the quantified output of this invention into standardized, executable control actions, based on the principle of "expert experience + data analysis," the final synthesized stable torque is... The output range is used for application partitioning. The partitioning criterion is based on the final synthesized stable torque. The physical meaning of is that its numerical value directly reflects the braking torque intensity required to maintain vehicle stability, which is also the overall risk intensity perceived by the system. Through statistical cluster analysis of a large amount of experimental data, combined with the experience of vehicle dynamics experts, three torque interval boundaries with clear physical meanings were identified. The definitions and corresponding operations for each interval are shown in Table 4 below.

[0142] Table 4. Output Interval Division Table:

[0143]

[0144] It should be noted that all calculation formulas in this application employ regression analysis, including but not limited to machine learning algorithms, to deeply analyze the collected parameters and identify their natural trends and interrelationships. Specialized software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Then, cross-validation and other methods are used to objectively evaluate the model performance, and continuous feedback and optimization are combined to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their effectiveness and accuracy. In all calculation formulas in this application, the parameters in each formula undergo dimensionless processing within a consistent range to ensure that different physical quantities are compared on the same scale; dimensionless processing techniques include, but are not limited to, min-max-normalization and Z-score standardization.

[0145] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A dynamic braking element arrangement and control device for preventing the tilting of parade floats, characterized in that, Specifically, it includes: The first center of mass state perception module is used to acquire dynamic axle load data at multiple wheel positions on the float in real time, and determine the real-time center of mass position of the float based on the dynamic axle load data. The second grounding margin detection module is used to determine the real-time grounding margin of the wheel by applying a preset probing braking torque pulse to the brake of at least one wheel and monitoring the angular velocity response of the wheel. The risk source decoupling module is used to calculate the expected grounding margin based on the real-time centroid position determined by the first centroid state sensing module through a preset benchmark model, and compare the expected grounding margin with the real-time grounding margin determined by the second grounding margin detection module to generate a prediction residual signal characterizing the difference between internal and external risk sources. The adaptive selection module for control strategy is used to adaptively select between a preset feedforward stabilization control strategy based on the real-time centroid position and a feedback stabilization control strategy based on the real-time grounding margin, according to the characteristics of the predicted residual signal. The braking torque distribution execution module is used to calculate and issue braking torque commands to the braking elements of the float according to the selected control strategy, so as to generate a stabilizing torque to suppress tilting.

2. The dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 1, characterized in that: The first centroid state perception module includes: the dynamic axle load data is the four-corner dynamic axle load collected by sensors installed at the four corner suspensions; specifically, based on the four-corner dynamic axle load and preset geometric parameters, the lateral centroid coordinates and longitudinal centroid coordinates of the float are calculated in real time by solving the static torque balance equation; the lateral centroid coordinates and longitudinal centroid coordinates are combined and analyzed to obtain the normalized centroid offset.

3. The dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 2, characterized in that: The preset geometric parameters include the vehicle wheelbase and the vehicle track width; Before solving the static torque balance equation, the first centroid state perception module also includes calculating the total mass of the vehicle; The steps for calculating the lateral centroid coordinates specifically include: calculating the difference between the total dynamic axle loads on the left and right sides, multiplying the difference by half the vehicle wheelbase, and then dividing the result by the product of the total mass of the vehicle and the gravitational acceleration. The steps for calculating the longitudinal centroid coordinates specifically include: calculating the difference in total dynamic axle load between the front and rear axles, multiplying the difference by half the vehicle wheelbase, and then dividing the result by the product of the vehicle's total mass and gravitational acceleration. Pre-set the lateral and longitudinal safety boundaries for the allowable offset of the vehicle's center of gravity; The absolute value of the real-time calculated lateral centroid coordinates is divided by the lateral safety boundary to obtain the normalized lateral offset; the absolute value of the longitudinal centroid coordinates is divided by the longitudinal safety boundary to obtain the normalized longitudinal offset; the normalized lateral offset and normalized longitudinal offset are fused using the Euclidean norm to calculate the final normalized centroid offset. The smaller the normalized centroid offset value, the closer the real-time centroid position of the float is to its ideal geometric center, the more uniform the vehicle mass distribution, and the lower the risk of tilting. The larger the normalized centroid offset value, the greater the offset between the real-time centroid position and the geometric center of the float, the more uneven the vehicle mass distribution, and the higher the risk of tilting.

4. The dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 3, characterized in that: The real-time grounding margin is a quantitative indicator that is proportional to the amplitude of the angular deceleration response generated under the action of the probing braking torque pulse; The amplitude of the angular deceleration response is determined by performing linear regression analysis on the angular velocity time series data collected during the pulse action and taking the absolute value of the slope of the fitted line. The "quantitative index that is proportional to the amplitude of the angular deceleration response" is defined as the normalized grounding margin. The normalized grounding margin is used to characterize the real-time grounding stability of the target wheel as determined by active detection. The larger the value, the tighter the contact between the wheel and the ground, and the greater the vertical load it bears.

5. The dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 4, characterized in that: For the risk source decoupling module, the expected grounding margin is calculated by inputting the normalized centroid offset into a preset benchmark model; The preset benchmark model is established through offline experimental calibration and a lookup table stores the mapping relationship between the normalized centroid offset and the corresponding expected normalized ground margin. The predicted residual signal is determined by calculating the absolute value of the difference between the normalized ground margin and the expected normalized ground margin, and the predicted residual signal is characterized as the risk source characteristic spectrum separation degree. The output of the risk source feature spectrum separation degree is defined as follows: the absolute value of the difference is compared with a preset insensitivity threshold, and the risk source feature spectrum separation degree is output only when the absolute value of the difference is greater than the insensitivity threshold; otherwise, the output is zero.

6. A dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 5, characterized in that: The smaller the risk source characteristic spectrum separation value, the higher the match between the measured grounding state and the prediction based on the internal mass distribution, and the more likely the current overturning risk is to originate from internal load transfer. The larger the risk source characteristic spectrum separation value, the more serious the deviation between the measured grounding state and the theoretical prediction, and the greater the disturbance of the vehicle from the external risk source.

7. The dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 6, characterized in that: The adaptive selection module for the control strategy is specifically used to: calculate the control mode authority factor based on the predicted residual signal, and use the control mode authority factor to perform weighted fusion of the output of the feedforward stable control strategy and the output of the feedback stable control strategy to generate the final synthesized stable torque; The control mode permission factor is calculated by comparing the predicted residual signal with a preset permission factor activation threshold and a preset permission factor saturation threshold, and then using a piecewise linear mapping function. The output of the feedforward stabilization control strategy is the feedforward control base torque obtained by parallel calculation of the real-time centroid position, and the output of the feedback stabilization control strategy is the grounding margin calculated in parallel.

8. A dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 7, characterized in that: The output of the feedforward stabilization control strategy is the feedback control base torque obtained by acquiring the offset and rate of change of the real-time centroid position, and performing lookup and interpolation calculations in a pre-calibrated feedforward control lookup table. The output of the feedback stability control strategy is obtained by calculating the error between the minimum real-time grounding margin and the preset safety target value, and inputting the error into the proportional-integral-derivative controller for calculation. The braking torque distribution execution module first determines the dangerous wheel with the smallest real-time grounding margin, and then determines the diagonal wheel of the dangerous wheel as the target braking wheel; The braking torque distribution execution module will calculate the target braking torque based on the preset braking torque distribution ratio of the final synthesized stable torque, and send it to the target braking wheel.

9. A dynamic braking element arrangement and control device for preventing tilting of parade floats according to claim 8, characterized in that: The smaller the control mode authority factor, the more the characterization system tends to predict the source of the current roll risk from the internal load transfer; the weight allocation of the final composite stabilizing moment will be closer to the feedforward control base moment. When the control mode authority factor is larger, the characterization system tends to judge the current roll risk source as more likely to be an unpredictable external sudden disturbance; the final weight allocation of the calculated stable torque will be closer to the feedback control base torque.

Images