Method for controlling stability of transport vehicle

Abstract

The invention relates to a method for controlling the stability of a transport vehicle. Comprising the following steps that dynamic parameters of the speed, the acceleration, the steering angle and the sideslip angle of a vehicle are collected through a vehicle sensor comprising an accelerometer, a gyroscope and a wheel speed sensor; in combination with various sensors such as a vehicle-mounted radar, a laser radar and a camera, conditions of wet, slippery, rugged and uneven road surfaces are monitored in real time, and road surface friction coefficient parameters are generated through a data fusion algorithm; according to the dynamic state of the vehicle and the real-time road surface state, the stability of the vehicle is evaluated through a mathematical model, and the potential risk of current instability is calculated; a plurality of stability control modes are preset according to different driving scenes including expressways, mountain roads and wet and slippery roads, and different control parameters and algorithms are adopted in each mode; a sensor including a driver state monitoring system inside a vehicle and external environment data including meteorological data and traffic information are utilized to predict operation behaviors including sudden acceleration and sudden braking of a driver.

Description

Technical Field

[0001] The invention relates to a method for controlling vehicle stability, in particular to a method for controlling the stability of a transport vehicle. Background Art

[0002] It is currently widely used in transportation vehicle stability control methods, especially the technology mentioned in "A Vehicle Stability Integrated Control Method Based on Dual Stability Envelope" (Publication No. CN118701030A). Although it has improved vehicle stability to a certain extent, it still has certain shortcomings and disadvantages, which are mainly reflected in the reaction speed, model complexity, adaptability and scalability of stability control.

[0003] First, the control framework of the dual stability envelope can balance the stability performance of the vehicle and the calculation amount of the controller under certain conditions by introducing soft constraints on vehicle state and hard constraints on actuator stability, but the calculation burden of this method is heavy, especially in dynamic driving environments. Specifically, the mapping of vehicle instability conditions and the determination of the stability limit value of the actuator involve complex mathematical modeling and real-time data calculation, which places high demands on the processing power of the on-board computing platform. In practical applications, due to limited on-board computing resources, when the vehicle enters a complex driving environment or needs to adjust the control strategy in real time, the existing system may face the problem of excessive calculation, resulting in a delay in control response. This delay may directly affect the stability of the vehicle, especially in the case of high-speed driving or emergency braking, and effective stability intervention cannot be carried out quickly, thereby increasing the risk of instability. Secondly, the dynamic update method of wheel posture and the stability boundary correction function in the current technical solution, although the vehicle stability boundary is adjusted by considering the vehicle driving conditions, driver operation and control input of the active safety system, this process still relies on a fixed stability boundary description and does not fully consider the dynamic changes under different road conditions and different environmental conditions. In fact, external environmental factors such as road friction coefficient and climate change have a significant impact on vehicle stability, and these factors have not been fully fed back and dynamically adjusted in the existing control framework. For example, when driving on slippery or icy roads, the posture of the wheels and the stability boundary of the vehicle will change significantly. If the control system fails to perceive and adjust the boundary in time, it may cause the vehicle to become unstable, especially when making sharp turns or driving at high speeds. The current correction function may not take these extreme conditions into account. Therefore, the existing methods have certain limitations in adaptability and are difficult to handle changing driving environments and different road conditions, which limits their widespread application in actual traffic environments.

[0004] Furthermore, although the stability boundary correction function in the existing technical solution can be adjusted according to the driver's operation and the control input of the active safety system, its ability to cope with emergencies (such as extreme control or sudden changes in road conditions) is weak. For example, when the driver operates the steering wheel too vigorously or in emergency avoidance, the vehicle may suddenly lose control, and the existing system requires a long time of prediction and calculation to respond. In such emergencies, if the vehicle's stability control system cannot respond quickly, it may cause the vehicle to lose control or an accident. The current technology lacks sufficient real-time perception and rapid response capabilities, and cannot complete complex stability adjustments in a very short time, which affects its effectiveness and safety in emergency situations. Summary of the invention

[0005] The object of the present invention is to provide a method for controlling the stability of a transport vehicle, thereby solving some of the drawbacks and deficiencies pointed out in the background art.

[0006] The present invention solves the above-mentioned technical problems by adopting the following technical solution: comprising the following steps:

[0007] S1. Adaptive stability control based on vehicle kinematics and road perception:

[0008] S1.1. Collect the vehicle's speed, acceleration, steering angle, and sideslip angle dynamic parameters through vehicle sensors including accelerometers, gyroscopes, and wheel speed sensors;

[0009] S1.2, Combine multiple sensors such as vehicle-mounted radar, lidar, and camera to monitor the slippery, rugged, and uneven road conditions in real time, and generate road friction coefficient parameters through data fusion algorithms;

[0010] S1.3. Based on the dynamic state of the vehicle and the real-time road conditions, the stability of the vehicle is evaluated using a mathematical model to calculate the potential risk of instability;

[0011] S2, multi-mode stability control strategy based on intelligent algorithm:

[0012] S2.1. Multiple stability control modes are preset according to different driving scenarios, including highways, mountain roads, and slippery roads. Different control parameters and algorithms are used in each mode.

[0013] S2.2. Use sensors inside the vehicle, including the driver status monitoring system, and external environmental data, including meteorological data and traffic information, to predict the driver's operating behaviors, including sudden acceleration and sudden braking, and develop stability adjustment strategies for each behavior;

[0014] S3, Collaborative stability control based on Internet of Vehicles:

[0015] S3.1. Use vehicle-to-vehicle communication technology (V2V) to control the real-time information sharing between vehicles, including vehicle speed, acceleration, and dynamic data of vehicle spacing;

[0016] S3.2. Based on the data shared by the Internet of Vehicles, the stability risk of the entire fleet is evaluated through collaborative algorithms to predict group vehicle stability issues;

[0017] S3.3. During the driving process of the convoy, the stability control system of each vehicle is adjusted through the collaborative control algorithm, including adjusting the vehicle speed, increasing the following distance, and optimizing the braking system;

[0018] S4. Stability management based on predictive control:

[0019] S4.1. Combine vehicle dynamics models with real-time sensor data including GPS, speed sensors, and environmental sensors to build a dynamic predictive control system.

[0020] S4.2. Use the model predictive control (MPC) algorithm to predict in real time the stability risks that the vehicle is expected to encounter in the next few seconds or hundreds of meters, including skidding and loss of control.

[0021] S4.3. Take intervention measures in advance based on the stability risk prediction results, including automatic adjustment of steering angle, braking force, and traction distribution.

[0022] Furthermore, the adaptive stability control method based on vehicle kinematics and road perception includes:

[0023] By collecting the vehicle's dynamic state data including speed, acceleration, and sideslip angle in real time, as well as the real-time road state information including friction coefficient and road slope, the vehicle dynamics model is dynamically adjusted in combination with the machine learning algorithm; an adaptive stability model is constructed based on the vehicle's real-time state and road information; the vehicle stability index S(t) is expressed by the following formula:

[0024]

[0025] Where: S(t) is the stability index of the vehicle at time t; f(v(t), w(t), R(t)) is a function that represents the dynamic state of the vehicle and the road conditions; v(t) is the instantaneous speed of the vehicle, indicating the speed of the vehicle at time t; w(t) is the real-time load of the vehicle, representing the change in the current load of the vehicle; R(t) is the road friction coefficient, reflecting the friction intensity between the road surface and the tire; a(t) is the acceleration of the vehicle at time t, indicating the rate of change of the vehicle speed; T is the time period, indicating the dynamic state of the vehicle and the total duration of the evaluation period.

[0026] Furthermore, the adaptive stability control method based on vehicle kinematics and road perception includes:

[0027] Environmental variables including road friction coefficient, meteorological conditions, road type, and driver behavior characteristics are introduced; the instability risk of the vehicle is evaluated through a weighted combination of multi-dimensional factors; in the multi-dimensional stability assessment, the instability risk R risk (t) can be calculated by the following formula:

[0028]

[0029] Where: R risk (t) represents the instability risk at time t; D i (t) is the dynamic variable of the i-th dimension including vehicle steering angle, vehicle speed, and sideslip angle; α i is the weight coefficient of the i-th dynamic variable; It is the derivative of the vehicle stability index S(t) with respect to time t, indicating the rate of change of vehicle stability.

[0030] Furthermore, the adaptive stability control method based on vehicle kinematics and road perception includes:

[0031] Using deep learning and model predictive control MPC technology, we develop a real-time instability risk calculation and dynamic warning system. We monitor the dynamic state of the vehicle in real time, combine the current road state and environmental data, predict whether the vehicle will become unstable in the future unit time, and trigger the warning system in advance. The real-time instability risk prediction is achieved by calculating the real-time stability risk R risk (t) and the adjustment factor δ(τ) related to the driver's behavior are calculated by combining the following integral formula:

[0032] W(t)=∫0 t [R risk (τ)·δ(τ)]dτ

[0033] Where: W(t) is the cumulative instability risk up to time t; it represents the accumulated amount of instability risk of the vehicle in unit time; it is used to predict whether the vehicle is about to become unstable and trigger the early warning system; δ(τ) is a dynamic adjustment factor related to the driver's behavior, reflecting the driver's operating habits including sudden acceleration, sudden braking, and cornering; R risk (τ) is the real-time instability risk at time τ, which is calculated in real time based on the dynamic state of the vehicle, environmental variables and the driver’s operation.

[0034] Furthermore, the stability management method based on predictive control includes:

[0035] The model predictive control MPC algorithm is introduced to predict and optimize the future state, predict and intervene in advance the dynamic changes of the vehicle in the next few seconds or hundreds of meters; the vehicle state prediction formula is as follows:

[0036]

[0037] in, It indicates that the vehicle state vector at time t+k is predicted starting from time t; the state vector includes the vehicle's position, speed, acceleration, steering angle, and sideslip angle dynamic information; f represents the vehicle dynamics model function, which describes the relationship between the vehicle state and the control input and external disturbance; It represents the predicted value of the vehicle state at time t+k-1; u(t+k-1) is the control input of the vehicle at time t+k-1 including acceleration, braking and steering; w(t+k-1) is the external disturbance including road friction coefficient and meteorological conditions.

[0038] Furthermore, the stability management method based on predictive control includes:

[0039] Real-time calculation of the vehicle's stability risk index in a few seconds or hundreds of meters, timely detection of predicted stability problems; by combining the vehicle dynamics model, the external environment including road conditions, meteorological data and control inputs, MPC calculates the vehicle's stability risk in real time at future moments; stability risk assessment is performed using the following integral formula:

[0040] R(t+k)=∫ t t+k [α·||x(t)|| 2 +β·||u(t)|| 2 +γ·||w(t)|| 2 ]dt

[0041] Among them, R(t+k) represents the stability risk index from time t to time t+k, which quantifies the stability risk of the vehicle in this unit time; x(t) is the vehicle state vector, including important dynamic parameters such as vehicle speed, steering angle, and sideslip angle, which is used to measure the dynamic behavior of the vehicle; u(t) is the control input including acceleration, braking force, and steering angle; w(t) is the external disturbance including road friction coefficient and meteorological conditions; α, β, γ are weighting coefficients, which respectively represent the importance of vehicle state, control input, and external disturbance in stability risk assessment; the integral symbol Indicates a cumulative risk assessment for the next k seconds.

[0042] Furthermore, the stability management method based on predictive control includes:

[0043] MPC achieves vehicle stability control through multi-constraint optimization of multiple control variables and constraints; the optimization objective function is expressed as follows:

[0044]

[0045] Among them, u(t) is the vehicle's control input including acceleration and braking; x(t+k|t) represents the vehicle state prediction at the future time t+k; MPC will predict the vehicle state at each future time; w(t+k) represents external disturbances including road friction coefficient and meteorological conditions; λ1, λ2, λ3 are adjustment coefficients, which respectively represent the relative importance of vehicle state, control input and external disturbance in the optimization objective; N is the number of optimization time steps, which usually represents the prediction interval foreseen in the future; by optimizing the control input within the entire interval, MPC ensures that the vehicle is always in a stable state.

[0046] The transport vehicle stability control method of the present invention has the following specific beneficial effects:

[0047] 1. Identify and adjust the vehicle's motion trajectory in advance before the vehicle potentially loses stability. For example, in the case of sharp turns or slippery roads, the system can predict the risk of the vehicle sliding or losing control, and prevent the vehicle from entering an unstable state by adjusting control inputs such as vehicle speed, steering angle, acceleration and braking systems. Compared with traditional stability control methods, the present invention can intervene earlier and reduce the probability of loss of control.

[0048] 2. Comprehensively consider multiple control inputs, vehicle dynamics and external disturbances (such as road friction coefficient, weather conditions, etc.) to ensure the best driving state of the vehicle under different road conditions and environments. By optimizing the motion trajectory within the next few seconds or hundreds of meters through MPC, the system can adjust the control strategy in real time to balance safety, efficiency and comfort, thus effectively avoiding the limitations of traditional methods based on a single control target or simplified model. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The figure is a flow chart of the method for controlling the stability of a transport vehicle according to the present invention.

[0050] Figure 2 This is a flow chart of the adaptive stability control method based on vehicle kinematics and road surface perception of the present invention.

[0051] Figure 3 This is a flow chart of the stability management method based on predictive control of the present invention. DETAILED DESCRIPTION

[0052] The specific implementation modes of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0053] Combination Figure 1 The transportation vehicle stability control method of the present invention firstly carries out an in-depth design in terms of vehicle kinematics and road perception, aiming to adjust the vehicle's stability level in real time through adaptive stability control, ensuring that the vehicle remains stable and prevents loss of control under various complex road conditions. The key steps of the method involve using a series of high-precision sensors to obtain the dynamic state information of the vehicle, and at the same time combining road perception technology to generate a real-time road friction coefficient through a data fusion algorithm, and further calculate and predict the stability risk of the vehicle.

[0054] First, the system collects the basic dynamic parameters of the vehicle in real time through vehicle sensors (such as accelerometers, gyroscopes, wheel speed sensors, etc.). These sensors can provide key dynamic data such as vehicle speed, acceleration, steering angle, and sideslip angle, which are the basis for evaluating vehicle stability. For example, vehicle speed can reflect the driving state of the vehicle, steering angle reflects the change in the vehicle's driving direction, and sideslip angle is an important indicator for judging whether the vehicle is skidding or about to lose control. By continuously monitoring these parameters, the system can accurately judge the dynamic changes of the vehicle and thus detect potential stability problems in a timely manner.

[0055] At the same time, a variety of sensors such as on-board radar, lidar and cameras are used to monitor road conditions in real time. These sensors can capture information about the road surface, including wet, rugged, uneven road conditions, which have an important impact on the stability of the vehicle. For example, a wet road has a low friction coefficient, which can easily cause the vehicle to slip; while a rugged or uneven road surface may cause the vehicle to vibrate more, affecting the vehicle's driving stability. In order to accurately reflect these changes in road conditions, the system uses a data fusion algorithm to comprehensively process data from different sensors to generate accurate road friction coefficient parameters. This parameter is the key to evaluating the relationship between road friction and vehicle dynamic behavior. It can provide more accurate road state information under real-time changing environmental conditions. Next, the system combines the vehicle's dynamic state with the real-time road state and uses a mathematical model to comprehensively evaluate the vehicle's stability. The model takes into account vehicle dynamics (such as steering system, suspension system, etc.) and external environmental factors (such as road friction coefficient, etc.), and can accurately predict the vehicle's stability level under the current driving state. Through real-time calculations, the model can assess the vehicle's current potential risk of instability, especially when the vehicle is traveling fast or entering complex road conditions, and can accurately determine whether there are risks such as skidding and loss of control.

[0056] The multi-mode stability control strategy based on intelligent algorithms aims to adjust the vehicle's stability control strategy in real time according to different driving scenarios and environmental conditions in an intelligent way to cope with stability problems that may occur in various driving scenarios. First, the method presets stability control modes under various driving scenarios and designs different control parameters and algorithms for each mode, thereby achieving stability optimization under different circumstances. Specifically, for different road conditions such as highways, mountain roads, and slippery roads, each driving scenario has a corresponding control mode, which sets different control algorithms and parameters according to specific driving conditions (such as vehicle speed, steering angle, road friction coefficient, etc.) to achieve the best stability control effect. For example, on highways, vehicles may maintain a high speed, so the system will give priority to stability adjustment in a large range, while on mountain roads or slippery roads, the system will adjust to a more detailed control mode, emphasizing real-time monitoring and adjustment of small changes in the vehicle. In addition, the driver status monitoring system inside the vehicle will also provide crucial information for this step.

[0057] The system monitors the driver's behavior in real time through sensors to identify whether there are any operations that may affect the stability of the vehicle, such as sudden acceleration and sudden braking. For example, when the system recognizes that the driver is accelerating suddenly, it will immediately adjust the vehicle's power output according to the vehicle's current dynamic state and road conditions, and limit the acceleration change to avoid loss of control or skidding due to excessive acceleration; and when the system recognizes the operation of sudden braking, it will optimize the response of the braking system based on factors such as the current vehicle speed and road friction coefficient, and distribute the brakes in advance, thereby reducing the risk of wheel locking or instability during braking. In addition, external environmental data is also an important basis for the formulation of stability control strategies. These external data include meteorological information (such as rain and snow, wind speed, temperature, etc.) and traffic information (such as the distance to the vehicle ahead, traffic flow, road condition changes, etc.). This information is processed by intelligent algorithms, combined with the driver's behavior pattern, to predict possible dangerous situations and adjust the vehicle's stability control strategy in advance. For example, on slippery road conditions, if the system detects water or rain or snow ahead, it will pre-adjust the vehicle's dynamic response, reduce the speed, and adjust the steering and braking systems to ensure that the vehicle remains stable under changing road conditions and avoids loss of control.

[0058] The third step of the present invention is the collaborative stability control based on the Internet of Vehicles, which aims to achieve real-time information sharing and collaborative control between vehicles in the fleet through vehicle-to-vehicle (V2V) communication technology, thereby improving the stability and safety of the entire fleet. During the driving process of the fleet, each vehicle shares real-time information with other vehicles through on-board communication equipment. The information includes dynamic data such as vehicle speed, acceleration, and vehicle spacing. The real-time sharing of these data enables each vehicle in the fleet to understand the status and behavior changes of other vehicles in a timely manner, so as to make more accurate control decisions. Specifically, V2V communication technology enables each vehicle to send dynamic status data including its own speed, acceleration, steering angle, etc. to surrounding vehicles, and receive corresponding information from other vehicles at the same time. The exchange of these real-time data not only allows each vehicle to understand the driving status of the front and rear vehicles, but also obtains the changes in the vehicle spacing in real time. When a vehicle accelerates or decelerates, other vehicles can immediately obtain this information through V2V communication, so as to make adaptive adjustments while ensuring safety, avoiding accidental collisions or causing instability in the entire fleet. In addition, based on the data shared by the Internet of Vehicles and through advanced collaborative algorithms, the system can comprehensively evaluate the stability risks of the entire fleet and predict possible stability problems for all vehicles in the group.

[0059] For example, if a vehicle skids or loses control due to excessive speed, the risk information will be transmitted to the vehicle behind through V2V technology, and the vehicle behind can adjust its speed or change its driving path in real time to avoid the instability caused by following the vehicle too closely or too fast. Based on this information, the collaborative control algorithm can make a global assessment of the stability of the entire fleet to ensure that the stability risk of each vehicle is within the controllable range. Through this global assessment, the system can identify potential group vehicle instability risks in advance and take corresponding preventive measures. During the driving process of the fleet, the collaborative control algorithm can dynamically adjust the stability control system of each vehicle. For example, the system may increase the following distance between vehicles by slowing down according to the overall driving situation of the fleet, or optimize the braking system response of each vehicle when encountering complex road conditions to ensure that the entire fleet can respond in a coordinated manner when encountering emergencies, avoiding the overall loss of control of the fleet due to improper operation of individual vehicles. Through this collaborative control method, the control system of each vehicle in the fleet can not only make adjustments according to its own status, but also be dynamically optimized according to the dynamic information of the entire fleet to ensure that the fleet can maintain the best stability performance at any time.

[0060] The fourth step is stability management based on predictive control, which aims to build a dynamic predictive control system by combining the vehicle dynamics model and real-time perception data, so as to predict the stability risk of the vehicle in the future in real time and take timely intervention measures when risks occur.

[0061] The core of this method is to use the model predictive control (MPC) algorithm to predict the future motion trajectory of the vehicle, and to comprehensively manage the stability of the vehicle by combining a variety of real-time perception data including GPS, speed sensors, and environmental sensors. First, the system builds a dynamic vehicle state perception system by integrating data from various vehicle sensors, including vehicle position provided by GPS information, vehicle speed data provided by speed sensors, and external environmental data such as road friction coefficient and meteorological conditions provided by environmental sensors. GPS and speed sensors provide the system with accurate position information and vehicle speed data, while environmental sensors monitor road conditions, climate conditions, and other environmental factors that may affect vehicle stability in real time. Through these data, the system can obtain the vehicle's motion state and changes in the surrounding environment in real time, thereby providing sufficient input information for predictive control.

[0062] Next, the system uses the model-based predictive control (MPC) algorithm to predict the vehicle's trajectory in the next few seconds or hundreds of meters. The core advantage of the MPC algorithm lies in its foresight. It not only depends on the current state of the vehicle, but also takes into account possible changes in the future, and predicts the future state of the vehicle by optimizing the control model. This process can take into account the dynamic characteristics of the vehicle, including the steering system, suspension system, traction distribution, etc. Combined with the current road conditions and driving conditions, the system can calculate the stability risks that may occur in the future, such as skidding and loss of control. For example, when the vehicle is about to enter a curve, the MPC algorithm can predict the risk of skidding and adjust the control parameters such as steering angle and vehicle speed in advance to prevent the vehicle from losing control or slipping. For possible loss of control, MPC can predict the dynamic behavior of the vehicle in an unstable state and intervene in time through the control system to avoid the safety hazards caused by loss of control. Based on the prediction results of these stability risks, the system can take intervention measures in advance to avoid the occurrence of instability. Specifically, the system will automatically adjust the vehicle's steering angle, braking force, traction distribution and other control parameters to ensure that the vehicle is always in the best stable state. Adjusting the steering angle can help the vehicle deal with stability issues that may arise when turning or avoiding obstacles. Optimizing the braking force can avoid vehicle loss of control caused by excessive braking, and adjusting the traction distribution can effectively deal with the impact of changes in the road friction coefficient or vehicle load on stability. Through this series of real-time intervention measures, the system can ensure that the vehicle always remains within a safe stability range and avoid instability caused by excessive or insufficient control. In general, the predictive control-based stability management method in this step can adjust the control parameters in real time during vehicle driving by acquiring real-time status information of the vehicle and the environment, combined with the MPC algorithm's prediction of future risks, to achieve precise management and effective intervention in vehicle stability.

[0063] Embodiment 1:

[0064] Combination process Figure 2 , assume that a truck is traveling on a highway, the current speed is 90 km / h, the vehicle weight is 10 tons, the road is dry, and the friction coefficient is 0.8. The vehicle's acceleration (at) is 2 m / s 2 , the vehicle is traveling on a relatively gentle slope. To further understand the adaptive stability control method based on vehicle kinematics and road perception, combined with the adaptive adjustment mechanism of real-time dynamic data and machine learning algorithms, a detailed example analysis can be performed in the following steps.

[0065] First, the vehicle's real-time status data and road status information are collected by the system in real time through various sensors. At this time, the vehicle's dynamic status information includes vehicle speed, acceleration and load, and the road information is monitored by sensors to measure the friction coefficient and road slope. According to the system design, the vehicle stability index (St) is expressed by the following formula:

[0066]

[0067] In this formula:

[0068] v(t) is the instantaneous speed of the vehicle, which is 90 km / h here, converted to meters per second is 25 m / s;

[0069] w(t) is the real-time load of the vehicle, which is 10 tons, or 10,000 kg;

[0070] R(t) is the road friction coefficient, which is set to 0.8 here, indicating a typical dry road surface;

[0071] a(t) is the acceleration of the vehicle, here it is 2m / s 2 ;

[0072] T is the time period, which is set to 10 seconds, that is, the total duration of the evaluation period is 10 seconds.

[0073] Next, the stability level of the current vehicle is evaluated by calculating the stability index S(t).

[0074] Step 1: Calculate the numerator of the stability index

[0075] First, determine the specific form of the function f(v(t), w(t), R(t)). This function is usually designed in combination with the vehicle's dynamic model and road conditions. The common model form is as follows:

[0076] f(v(t),w(t),R(t))=v(t)·w(t)·R(t)

[0077] Among them, v(t) is the instantaneous speed of the vehicle, w(t) is the load of the vehicle, and R(t) is the friction coefficient of the road surface. Substitute the current data:

[0078] f(v(t),w(t),R(t))=25m / s×10000kg×0.8=200000kg·m / s·friction

[0079] Step 2: Calculate the denominator of the stability index

[0080] The denominator is the square of the acceleration integrated over the time period. Since the vehicle's acceleration a(t) is 2m / s 2 , and assuming that the vehicle's acceleration is constant over 10 seconds, the integration becomes a simple multiplication:

[0081] ∫0 T (a(t)) 2 dt=∫0 10 (2) 2 dt=4·10=40m 2 / s 4

[0082] Step 3: Calculate the stability index S(t)

[0083] Substitute the above calculation results into the stability index formula:

[0084]

[0085] Therefore, the vehicle's stability index at this moment is 5000. This value indicates the vehicle's stability level under the current dynamic state and road conditions. A higher stability index indicates that the vehicle is in a relatively stable state, while a lower stability index may mean that the vehicle is at a higher risk of instability.

[0086] However, the stability of the vehicle is not always fixed. It will be affected by many factors, including changes in road conditions, changes in vehicle load, etc. Therefore, based on the machine learning algorithm, the system will dynamically adjust the vehicle's dynamic model to cope with different dynamic situations. For example, if the road friction coefficient decreases (such as when encountering a slippery road, the friction coefficient may drop to 0.4), the system will adjust the vehicle's stability control algorithm to increase the response sensitivity to vehicle acceleration, thereby better controlling vehicle stability.

[0087] After 10 seconds, the system detects that the road friction coefficient has changed to 0.4, and the vehicle speed has increased to 100 km / h (about 27.78 m / s), the load is still 10 tons, and the acceleration is 3 m / s 2 , then the vehicle's stability index will be recalculated.

[0088] The new calculation steps are as follows:

[0089] v(t)=27.78m / s

[0090] w(t)=10000kg

[0091] R(t)=0.4

[0092] a(t)=3m / s 2

[0093] Time period T = 10 seconds

[0094] The updated stability index formula is:

[0095] f(v(t),w(t),R(t))=27.78m / s×10000kg×0.4=111120kg·m / s·friction

[0096] ∫0 T (a(t)) 2 dt=∫0 10 (3) 2 dt=9·10=90m 2 / s 4

[0097] Therefore, the new stability index is:

[0098]

[0099] It can be seen that with the increase of vehicle speed and acceleration, and the decrease of road friction coefficient, the stability index of the vehicle decreases significantly, indicating that the stability risk of the vehicle increases. At this time, the system will automatically adjust the vehicle control system according to this stability index and dynamic model prediction, such as reducing the speed, adjusting the steering angle or optimizing the braking system to maintain the stability of the vehicle. Through the above examples, we can clearly see the practical application of the adaptive stability control method based on vehicle kinematics and road perception. In this process, the real-time collected vehicle dynamic data and road state information are effectively combined, and the control strategy is adjusted in real time using machine learning algorithms and dynamic models to ensure the stability of the vehicle under complex dynamic conditions.

[0100] The truck is driving on a mountain road at a current speed of 70 km / h and a load of 10 tons. The road surface is slippery with a friction coefficient of 0.6. The weather conditions are rainy, and the slippery conditions reduce the friction coefficient. There is some water on the road surface. The driver is making a sharp turn with a steering angle of 20 degrees and there is a slight acceleration during the turn. The vehicle's acceleration is 1.5 m / s 2, and the side slip angle is 3 degrees when turning. At this time, the vehicle stability index S(t) is calculated as a certain value (for example, 5000) by the formula as mentioned above, but considering more dimensions of the environment and driver behavior, the stability risk R risk (t) needs to be calculated using the following formula:

[0101]

[0102] in:

[0103] R risk (t) represents the instability risk at time t.

[0104] D i (t) is the dynamic variable of the i-th dimension, including the vehicle speed, steering angle, sideslip angle, etc.

[0105] α i is the weight coefficient of the ith dynamic variable, which is used to reflect the impact of different factors on stability risk.

[0106] It represents the derivative of the vehicle stability index S(t) with respect to time, reflecting the rate of change of vehicle stability.

[0107] Based on the real-time collected data, there are dynamic variables in the following dimensions:

[0108] Vehicle speed D1(t) = 70 km / h ≈ 19.44 m / s

[0109] Steering angle D2(t) = 20°

[0110] Side slip angle D3(t) = 3°

[0111] Next, a weight coefficient α is assigned to each dimension i , reflecting the influence of each variable on the risk of instability. Usually, vehicle speed and steering angle have greater influence, so they are given higher weights:

[0112] α1=0.5 (weight coefficient of vehicle speed)

[0113] α2 = 0.3 (weight coefficient of steering angle)

[0114] α3=0.2 (weight coefficient of sideslip angle)

[0115] The vehicle stability index S(t) = 5000 has been obtained. Next, we need to calculate the rate of change of the stability index relative to time. This needs to be estimated through the real-time status of the vehicle (such as acceleration, steering angle, sideslip angle, etc.) and the vehicle behavior predicted by the machine learning model. Assuming that in the current situation, the rate of change of the stability index is (Unit: change in stability index / second), which means that the stability of the vehicle is decreasing during acceleration.

[0116] Calculate the risk of instability R risk (t)

[0117] Substitute all known data into the formula:

[0118]

[0119] First, calculate the weighted combination of the dimensions:

[0120] α1·D1(t)=0.5·19.44=9.72

[0121] α2·D2(t)=0.3·20=6

[0122] α3·D3(t)=0.2·3=0.6

[0123] The sum of the weighted combinations is:

[0124] 9.72+6+0.6=16.32

[0125] Next, substitute the stability index change rate into the formula to calculate the final instability risk:

[0126] R risk (t) = 16.32 (-100) = -1632

[0127] Instability risk R risk The calculation result of (t) is negative, which means that the current vehicle has a high risk of instability under certain environmental conditions and needs to take appropriate intervention measures. Based on this result, the system will adjust the control strategy in real time, for example:

[0128] Slow down: As vehicle speed has a significant impact on stability, the system will instruct the driver to reduce speed.

[0129] Optimize steering: Too large a steering angle may result in loss of control, so the steering needs to be adjusted appropriately to ensure that the vehicle does not skid during the turn.

[0130] Increase following distance: Since the current road surface is slippery, increasing the distance between vehicles can help reduce the risk of collision.

[0131] When the instability risk reaches -1632, the system will adjust the vehicle's stability control system according to this warning value. Specific measures may include:

[0132] Vehicle deceleration: Reduce the vehicle speed to 50km / h (13.89m / s). At this time, the weighted combination of vehicle speeds will reduce the risk of instability.

[0133] Steering Angle Adjustment: Reduce the steering angle during sharp turns to keep the vehicle on a smoother trajectory.

[0134] Increase the following distance: adjust the distance according to the change of the distance between vehicles, so as to minimize the risk of collision between vehicles during driving. Through these real-time adjustments and predictions, the stability of the vehicle can be effectively guaranteed, thus avoiding possible dangerous situations such as skidding and loss of control.

[0135] Assume that the truck encounters complex road and weather conditions during driving, and the vehicle stability is facing potential threats. In order to achieve more accurate instability risk warning, this solution introduces deep learning and model predictive control (MPC) technology, combined with real-time instability risk calculation and dynamic warning system to ensure timely intervention and prevent loss of control.

[0136] The truck is still on a slippery mountain road, and the speed is maintained at 80km / h (about 22.22m / s). At this time, the on-board radar and camera have detected a sharp turn ahead, and the road is slippery, with a friction coefficient of R(t) = 0.4. In addition, the driver is accelerating sharply, and the on-board sensor detects that the driver is accelerating when passing the curve, the steering angle is 30 degrees, and the vehicle acceleration is 1.8m / s 2 , the sideslip angle is 4 degrees. The real-time stability index at this time is S(t) = 4500, but in order to improve the stability assessment accuracy of the system, an instability risk prediction system combined with deep learning is used.

[0137] Step 1: Calculate the real-time instability risk R risk (t)

[0138] First, the system uses deep learning models and MPC technology to monitor and predict the dynamic state of the vehicle (vehicle speed, acceleration, steering angle, sideslip angle, etc.) in real time, and combines environmental data (such as the road friction coefficient R(t)) to calculate the real-time instability risk. risk (t) is calculated to be 1200.

[0139] The formula is:

[0140]

[0141] Among them, dynamic variables such as vehicle speed D1(t) = 22.22m / s, steering angle D2(t) = 30°, side slip angle D3(t) = 4°, stability change rate (Based on acceleration and steering angle changes). Based on this data, the system calculates R risk (t) = 1200, which indicates that in the current state, the risk of vehicle instability is relatively high.

[0142] Step 2: Calculate the dynamic adjustment factor δ(τ) related to driver behavior

[0143] Next, consider the impact of driver behavior on vehicle stability, especially the habits of sudden acceleration, sudden braking, and cornering. The system analyzes the driver's behavior through sensors and machine learning models and finds that the driver has a strong habit of sudden acceleration. Therefore, in the current case, the dynamic adjustment factor δ(τ) is calculated as 1.3 (this value reflects the enhanced effect of the driver's sudden acceleration on the risk of instability). If the driver brakes or turns sharply, the adjustment factor may increase, otherwise it will decrease.

[0144] Step 3: Calculate the cumulative instability risk W(t)

[0145] Based on the real-time instability risk and adjustment factor, the system uses an integral formula to calculate the cumulative instability risk W(t) up to time t, which is a key indicator for predicting whether the vehicle is about to become unstable. The formula is as follows:

[0146] W(t)=∫0 t [R risk (τ)·δ(τ)]dτ

[0147] The calculation period is set to 5 seconds (i.e. t = 5), and the instability risk R is set to risk (τ) slightly rises to 1300 (due to dynamic changes in the environment and fluctuations in vehicle status). Therefore, the calculation process of the cumulative instability risk is:

[0148] w(5)=∫0 5 [1300·1.3]dτ=1300W1.3·5=8450

[0149] This result shows that in the next 5 seconds, the cumulative risk of instability of the vehicle is very high and close to the critical value. The system will immediately trigger a warning and automatically adjust the control strategy.

[0150] Based on the above calculations, the system has identified that the risk of instability is about to reach the threshold (8450), so it will automatically adjust the vehicle control system to avoid loss of control. Specific intervention measures may include:

[0151] Deceleration: Automatically reduce the vehicle speed to 60km / h (16.67m / s) to reduce the instability caused by acceleration and sudden braking.

[0152] Optimize steering: Reduce the steering angle to prevent oversteering and skidding.

[0153] Increase following distance: Increase the distance from the vehicle in front to reduce the risk of rear-end collision due to slippery roads.

[0154] After the early warning and intervention measures are implemented, the dynamic state of the vehicle will change, and the new instability risk calculation and adjustment factors will be re-evaluated. The system will monitor these changes in real time and continue to optimize the control strategy based on the new state to ensure the stability of the vehicle. After the vehicle is adjusted, the instability risk is reduced to 800, and the system will continue to monitor and make further adjustments if necessary.

[0155] Embodiment 2:

[0156] Combination process Figure 3 , assume that a transport truck is being processed and is passing a curve. At this time, the road surface on which the vehicle is driving is slippery, the road friction coefficient R = 0.4, the weather conditions are bad, and the wind speed is high. The driver's operation suddenly changes, and he accelerates sharply and tries to pass the curve quickly. At this time, the dynamic state information of the vehicle (such as speed, acceleration, steering angle, sideslip angle, etc.) is collected in real time through the on-board sensors. The current state of the vehicle is:

[0157] Vehicle current position: x(t) = 2000m;

[0158] Vehicle speed: v(t) = 22.22 m / s (80 km / h);

[0159] Vehicle acceleration: a(t) = 2.0 m / s 2 ;

[0160] Vehicle steering angle: θ(t) = 15°;

[0161] Vehicle sideslip angle: φ(t) = 3°;

[0162] Road friction coefficient: R(t)=0.4 (wet road surface).

[0163] The goal is to use the model predictive control (MPC) algorithm to predict the dynamic changes of the vehicle in the next few seconds and intervene in advance to avoid vehicle instability (such as skidding or loss of control).

[0164] The core of model predictive control (MPC) is the prediction of future states, based on the current vehicle state (x(t)), and taking into account control inputs (such as acceleration, braking, steering) and external disturbances (such as road friction coefficient, weather conditions, etc.). The formula for predicting the future state of the vehicle is:

[0165]

[0166] in:

[0167] It means starting from time t, predicting the state vector of the vehicle at time t+k;

[0168] is the predicted value at time t+k-1;

[0169] u(t+k-1) is the control input (acceleration, braking, steering, etc.) of the vehicle at time t+k-1;

[0170] w(t+k-1) is the external disturbance, such as road friction coefficient and meteorological conditions.

[0171] A simplified vehicle dynamics model is used to describe the changing relationship of the vehicle state. For the convenience of calculation, the future time prediction period is set to 3 seconds. Based on the state information at the current time t=0, the vehicle state in the future is predicted. The control input (such as acceleration, braking) and external disturbance (such as friction coefficient, weather) are assumed to change little in a short time.

[0172] Current status:

[0173] Speed ​​v(0) = 22.22 m / s;

[0174] Acceleration a(0) = 2.0 m / s 2 ;

[0175] Steering angle θ(0) = 15°;

[0176] Side slip angle φ(0) = 3°;

[0177] Road friction coefficient R = 0.4;

[0178] External disturbance w(0) = weather (affects steering and friction).

[0179] According to the simplified model prediction function f, the future state can be obtained through iterative calculation. For example, if the acceleration is set to be constant (u = 2.0m / s 2 ), and the road condition is stable, the predicted state can be calculated by the following formula:

[0180]

[0181] The change of vehicle status can be calculated as follows:

[0182] Velocity prediction: v(1) = v(0) + a(0) · Δt = 22.22 + 2.0 · 1 = 24.22 m / s

[0183] Side slip angle prediction: φ(1) = φ(0) + angularrateadjustment·Δt = 3° + 0.3° = 3.3°

[0184] Similarly, in this way, the system can iteratively calculate the state at each future moment (for example, t+2, t+3).

[0185] Based on the future state prediction, the MPC algorithm calculates the optimal control strategy based on the vehicle state, control input (such as acceleration, steering, etc.) and external disturbances to avoid vehicle instability. The system is set to detect that the vehicle may increase the risk of instability due to the increase in sideslip angle in 3 seconds (i.e. If the sideslip angle in the vehicle reaches a critical value), the system will optimize the control input in the following ways: Reduce acceleration: If the sideslip angle is predicted to be too large, the system will reduce acceleration to reduce the speed of the vehicle; Adjust steering angle: Based on future steering needs, the system may adjust the steering angle to reduce the risk caused by oversteering.

[0186] By optimizing the control input, the system can adjust the vehicle's driving state in advance to ensure that the vehicle remains stable during driving. Based on the optimization results of MPC, the system will adjust the control input in real time in the next few seconds. For example, if the prediction shows that the sideslip angle will continue to increase, the system may intervene in the following ways:

[0187] Automatic throttle reduction;

[0188] Adjust the braking force;

[0189] Optimizes the steering system to avoid oversteering.

[0190] For example, after 3 seconds, the MPC optimization will output a new control command: reduce the acceleration to 1.2m / s 2 , the steering angle is fine-tuned to 14° to ensure that the vehicle can smoothly pass the curve. A key advantage of the MPC algorithm is its dynamic adjustment capability. As the vehicle's state changes during driving (such as speed, steering angle, acceleration, etc.), the MPC system will continuously update the prediction model and adjust the control strategy based on the real-time state. For example, if the road conditions change suddenly (such as changes in the friction coefficient), the MPC system will re-evaluate the dynamic behavior of the vehicle and adjust the control input accordingly to ensure that the vehicle is always in the optimal stable state.

[0191] Suppose that a transport truck is being processed and is driving on a slippery mountain road. At the current time point t=0, the dynamic state data of the vehicle is collected in real time through sensors, including speed, steering angle, acceleration, etc. The current road condition is slippery, the weather conditions are bad, and the wind speed is high. The following is the initial state of the vehicle:

[0192] Vehicle current position: x(t) = 1500m;

[0193] Vehicle speed: v(t) = 20 m / s (72 km / h);

[0194] Vehicle acceleration: a(t) = 0.5 m / s 2 ;

[0195] Vehicle steering angle: θ(t) = 12°;

[0196] Vehicle sideslip angle: φ(t) = 2°;

[0197] Road friction coefficient: R(t) = 0.3 (wet road);

[0198] External weather conditions: strong wind, wind speed w wind =15m / s.

[0199] In order to assess the future stability risk of the vehicle, the MPC algorithm needs to use the vehicle's state information, control input and external disturbance information to calculate the stability risk index R(t+k) in real time. The risk assessment formula is as follows:

[0200] R(t+k)=∫ t t+k [α·||x(t)|| 2 +β·||u(t)|| 2 +γ·||w(t)|| 2 ]dt

[0201] in:

[0202] x(t) represents the state vector of the vehicle, including dynamic parameters such as vehicle speed, steering angle, and sideslip angle;

[0203] u(t) is the control input of the vehicle, including acceleration, braking, steering, etc.;

[0204] w(t) represents external disturbance, including road friction coefficient, meteorological conditions, etc.;

[0205] α, β, and γ are weighting coefficients of vehicle state, control input, and external disturbance, respectively, which are used to weigh their impact on stability risk assessment.

[0206] In this example, the following coefficient value ranges can be selected:

[0207] α = 0.5 (indicates that the vehicle dynamic behavior has a greater impact on stability risk);

[0208] β = 0.3 (indicates that the influence of control input is moderate);

[0209] γ = 0.2 (indicating that external disturbances have little impact on risk).

[0210] Next, the contribution of each item to the stability risk is calculated separately based on the actual dynamic data of the vehicle.

[0211] Contribution of vehicle state x(t):

[0212] The vehicle's state vector x(t) = [v(t), θ(t), φ(t)] includes vehicle speed, steering angle, and sideslip angle. First, calculate the sum of the squares of each dynamic parameter:

[0213] ||x(t)|| 2 =v(t) 2 +θ(t) 2 +φ(t) 2 =(20) 2 +(12°) 2 +(2°) 2

[0214]

[0215] ||x(t)|| 2 ≈400.154

[0216] Contribution of control input u(t):

[0217] Control input u(t) = [a(t), θ(t)], where acceleration a(t) = 0.5 m / s 2 , steering angle θ(t) = 12°, calculate the sum of the squares of the control inputs:

[0218] ||u(t)|| 2 =a(t) 2 +θ(t) 2 =(0.5) 2 +(12°) 2

[0219]

[0220] ||u(t)|| 2 ≈0.399

[0221] Contribution of external disturbance w(t):

[0222] External disturbances include the road friction coefficient R(t) = 0.3 and the wind speed w wind =15m / s. Therefore, the sum of the squares of the disturbances is:

[0223]

[0224] ||w(t)|| 2 =0.09+225=225.09

[0225] Now, substitute the calculated results into the risk assessment formula:

[0226] R(t+k)=∫tt +k [α·||x(t)|| 2+β·||u(t)|| 2 +γ·||w(t)|| 2 ]dt

[0227] Substituting the calculation results into each item, we get:

[0228] R(t+k)=∫ t t+k [0.5·400.154+0.3·0.399+0.2·225.09]dt

[0229] Compute the product of each coefficient and its squared value:

[0230] R(t+k)=∫ t t+k [200.077+0.12+45.018]dt

[0231] R(t+k)=∫ t t+k 245.215dt

[0232] If the time period k is set to 3 seconds, the stability risk index is:

[0233] R(t+k)=245.215×3=735.645

[0234] Based on the calculated stability risk index R(t+k), the system can timely predict the stability risk of the vehicle in the next few seconds. If the value exceeds the set threshold (e.g. 700), the system will trigger an early warning and automatically adjust the vehicle's control input according to the MPC optimization results, such as slowing down the speed, optimizing the steering angle, increasing the braking force, etc., to avoid vehicle instability.

[0235] Suppose a heavy truck is being handled, driving on a slippery uphill road. At the current time t=0, the dynamic data of the vehicle is collected in real time and enters the MPC control system, including vehicle speed, acceleration, steering angle, road friction coefficient, etc. In the current scenario, there is a risk of vehicle instability, especially on uphill and slippery roads. In order to ensure stable operation of the vehicle, MPC needs to adjust the vehicle's control input (such as acceleration, braking) and future state prediction through the following optimization objective function to minimize the risk of instability.

[0236] Assume the following is the initial state of the vehicle:

[0237] Current vehicle speed: v(t) = 18 m / s (64.8 km / h);

[0238] Current acceleration: a(t) = 0.3 m / s 2 ;

[0239] Current steering angle: θ(t) = 6°;

[0240] Current sideslip angle: φ(t) = 1°;

[0241] Road friction coefficient: R(t) = 0.4 (wet road);

[0242] External weather conditions: slight wind speed, wind speed w wind =10m / s;

[0243] Prediction time step: N = 5 (5 time steps in the future, i.e., prediction within 5 seconds).

[0244] The optimization objective function of MPC is expressed by the following formula:

[0245]

[0246] in:

[0247] u(t) is the control input, including acceleration, braking, etc.;

[0248] x(t+k|t) represents the vehicle state prediction at the future time t+k, including vehicle speed, steering angle, sideslip angle, etc.;

[0249] w(t+k) represents external disturbances, including road friction coefficient and meteorological conditions;

[0250] λ1, λ2, and λ3 are adjustment coefficients, which respectively control the importance of vehicle state, control input, and external disturbance in the optimization;

[0251] N is the number of prediction steps, which indicates the prediction range of future moments.

[0252] In this example, the following coefficients are selected:

[0253] λ1=0.6(the vehicle state has a more important influence on stability);

[0254] λ2 = 0.3 (moderate influence of control input);

[0255] λ3=0.1(external disturbance has little effect on stability);

[0256] The prediction time step is N=5 (5 seconds prediction period).

[0257] During the MPC optimization process, the future state of the vehicle must first be predicted. This step is done through the vehicle dynamics model, where the state vector x(t+k|t) includes dynamic parameters such as vehicle speed, steering angle, and sideslip angle. It is assumed that within the next 5 seconds, the vehicle's acceleration and steering angle will have the following changes:

[0258] Acceleration: a(t+1)=0.5m / s 2 , a(t+2)=0.6m / s 2 , a(t+3)=0.7m / s 2 , a(t+4)=0.8m / s 2 , a(t+5)=0.9m / s 2 ;

[0259] Steering angle: θ(t+1)=8°, θ(t+2)=9°, θ(t+3)=10°, θ(t+4)=11°, θ(t+5)=12°.

[0260] Based on the vehicle's dynamic model, MPC can predict the vehicle's state within the next 5 seconds. For example, the vehicle's speed is predicted to be v(t+1)=18.5m / s at the first second, v(t+2)=19m / s at the second second, and so on.

[0261] The key to MPC is to minimize the future state prediction error, control input, and the impact of external disturbances by optimizing the objective function. In the next 5 seconds, MPC will ensure that the vehicle is in the best stable state by optimizing the control input u(t).

[0262] For the control input u(t), MPC will optimize the acceleration and braking force so that the vehicle can drive stably according to the road conditions and dynamic state. After setting MPC optimization, the following control input is obtained:

[0263] u(t+1)=0.4m / s 2 (slight acceleration);

[0264] u(t+2)=0.5m / s 2 (moderate acceleration);

[0265] u(t+3)=0.6m / s 2 (accelerated increase);

[0266] ut+4=0.5"{m / s}^2 (moderate acceleration);

[0267] u(t+5)=0.4m / s 2 (Slight acceleration).

[0268] MPC also needs to consider the impact of external disturbances (such as road friction coefficient and meteorological conditions). In this example, slippery roads and high wind speeds will affect the stability of the vehicle. MPC will optimize by predicting the external disturbance w(t). In this scenario, the wind speed is assumed to fluctuate slightly in the next 5 seconds:

[0269] w(t+1)=10.5m / s;

[0270] w(t+2)=11m / s;

[0271] w(t+3)=10.8m / s;

[0272] w(t+4)=11.2m / s;

[0273] w(t+5)=10.6m / s.

[0274] Through the results of MPC optimization, an optimal control input sequence is finally obtained, which can ensure that the vehicle's stability in the next 5 seconds reaches the best level. During the optimization process, MPC will simultaneously consider the dynamic state of the vehicle, the control input and the impact of external disturbances on vehicle stability. Through optimized control inputs, the vehicle can achieve smooth driving and avoid instability or other dangers. For example, after MPC optimization, the control system decides to increase acceleration moderately in the next few seconds to overcome the lack of power caused by going uphill, and at the same time avoid skidding by adjusting the steering angle. Under the influence of slippery roads and strong winds, the control system will fine-tune the braking force to ensure the stability of the vehicle.

Claims

1. A method for controlling the stability of a transport vehicle, characterized in that The following steps are involved: S1. Adaptive stability control based on vehicle kinematics and road perception: S1.

1. Collect the vehicle's speed, acceleration, steering angle, and sideslip angle dynamic parameters through vehicle sensors including accelerometers, gyroscopes, and wheel speed sensors; S1.2, Combine multiple sensors such as vehicle-mounted radar, lidar, and camera to monitor the slippery, rugged, and uneven road conditions in real time, and generate road friction coefficient parameters through data fusion algorithms; S1.

3. Based on the dynamic state of the vehicle and the real-time road conditions, the stability of the vehicle is evaluated using a mathematical model to calculate the potential risk of instability; S2, multi-mode stability control strategy based on intelligent algorithm: S2.

1. Multiple stability control modes are preset according to different driving scenarios, including highways, mountain roads, and slippery roads. Different control parameters and algorithms are used in each mode. S2.

2. Use sensors inside the vehicle, including the driver status monitoring system, and external environmental data, including meteorological data and traffic information, to predict the driver's operating behaviors, including sudden acceleration and sudden braking, and develop stability adjustment strategies for each behavior; S3, Collaborative stability control based on Internet of Vehicles: S3.

1. Use vehicle-to-vehicle communication technology (V2V) to control the real-time information sharing between vehicles, including vehicle speed, acceleration, and dynamic data of vehicle spacing; S3.

2. Based on the data shared by the Internet of Vehicles, the stability risk of the entire fleet is evaluated through collaborative algorithms to predict group vehicle stability issues; S3.

3. During the driving process of the convoy, the stability control system of each vehicle is adjusted through the collaborative control algorithm, including adjusting the vehicle speed, increasing the following distance, and optimizing the braking system; S4. Stability management based on predictive control: S4.

1. Combine vehicle dynamics models with real-time sensor data including GPS, speed sensors, and environmental sensors to build a dynamic predictive control system. S4.

2. Use the model predictive control (MPC) algorithm to predict in real time the stability risks that the vehicle is expected to encounter in the next few seconds or hundreds of meters, including skidding and loss of control. S4.

3. Take intervention measures in advance based on the stability risk prediction results, including automatic adjustment of steering angle, braking force, and traction distribution.

2. The method for controlling the stability of a transport vehicle according to claim 1, characterized in that The adaptive stability control method based on vehicle kinematics and road perception includes: By collecting the vehicle's dynamic state data including speed, acceleration, and sideslip angle in real time, as well as the real-time road state information including friction coefficient and road slope, the vehicle dynamics model is dynamically adjusted in combination with the machine learning algorithm; an adaptive stability model is constructed based on the vehicle's real-time state and road information; the vehicle stability index S(t) is expressed by the following formula: Where: S(t) is the stability index of the vehicle at time t; f(v(t), w(t), R(t)) is a function that represents the dynamic state of the vehicle and the road conditions; v(t) is the instantaneous speed of the vehicle, indicating the speed of the vehicle at time t; w(t) is the real-time load of the vehicle, representing the change in the current load of the vehicle; R(t) is the road friction coefficient, reflecting the friction intensity between the road surface and the tire; a(t) is the acceleration of the vehicle at time t, indicating the rate of change of the vehicle speed; T is the time period, indicating the dynamic state of the vehicle and the total duration of the evaluation period.

3. The method for controlling the stability of a transport vehicle according to claim 2, characterized in that The adaptive stability control method based on vehicle kinematics and road perception includes: introducing environmental variables including road friction coefficient, meteorological conditions, road type and driver behavior characteristics; and evaluating the instability risk of the vehicle through a weighted combination of multi-dimensional factors.

4. The method for controlling the stability of a transport vehicle according to claim 3, characterized in that The adaptive stability control method based on vehicle kinematics and road perception includes: Using deep learning and model predictive control MPC technology, we develop a real-time instability risk calculation and dynamic warning system. We monitor the dynamic state of the vehicle in real time, combine the current road state and environmental data, predict whether the vehicle will become unstable in the future unit time, and trigger the warning system in advance. The real-time instability risk prediction is achieved by calculating the real-time stability risk R risk (t) and the adjustment factor δ(τ) related to the driver's behavior are calculated by combining the following integral formula: Where: W(t) is the cumulative instability risk up to time t; it represents the accumulated amount of instability risk of the vehicle in unit time; it is used to predict whether the vehicle is about to become unstable and trigger the early warning system; δ(τ) is a dynamic adjustment factor related to the driver's behavior, reflecting the driver's operating habits including sudden acceleration, sudden braking, and cornering; R risk (τ) is the real-time instability risk at time τ, which is calculated in real time based on the dynamic state of the vehicle, environmental variables and the driver’s operation.

5. The method for controlling the stability of a transport vehicle according to claim 1, characterized in that The stability management method based on predictive control includes: The model predictive control MPC algorithm is introduced to predict and optimize the future state, predict and intervene in advance the dynamic changes of the vehicle in the next few seconds or hundreds of meters; the vehicle state prediction formula is as follows: in, It indicates that the vehicle state vector at time t+k is predicted starting from time t; the state vector includes the vehicle's position, speed, acceleration, steering angle, and sideslip angle dynamic information; f represents the vehicle dynamics model function, which describes the relationship between the vehicle state and the control input and external disturbance; It represents the predicted value of the vehicle state at time t+k-1; u(t+k-1) is the control input of the vehicle at time t+k-1 including acceleration, braking and steering; w(t+k-1) is the external disturbance including road friction coefficient and meteorological conditions.

6. The method for controlling the stability of a transport vehicle according to claim 5, characterized in that The stability management method based on predictive control includes: real-time calculation of the stability risk index of the vehicle in a few seconds or hundreds of meters, and timely discovery of predicted stability problems; by combining the vehicle dynamics model, the external environment including road conditions and meteorological data, and control input, MPC calculates the stability risk of the vehicle in the future in real time.

7. The method for controlling the stability of a transport vehicle according to claim 6, characterized in that The stability management method based on predictive control includes: MPC achieves vehicle stability control through multi-constraint optimization of multiple control variables and constraints; the optimization objective function is expressed as follows: Among them, u(t) is the vehicle's control input including acceleration and braking; x(t+k|t) represents the vehicle state prediction at the future time t+k; MPC will predict the vehicle state at each future time; w(t+k) represents external disturbances including road friction coefficient and meteorological conditions; λ1, λ2, λ3 are adjustment coefficients, which respectively represent the relative importance of vehicle state, control input and external disturbance in the optimization objective; N is the number of optimization time steps, which usually represents the prediction interval foreseen in the future; by optimizing the control input within the entire interval, MPC ensures that the vehicle is always in a stable state.

Images