Semitrailer rollover early warning method and system based on dynamic threshold prediction

Abstract

The invention provides a semi-trailer rollover early warning method and system based on dynamic threshold value prediction. The early warning method comprises the steps that the transverse acceleration, the yaw velocity, the vehicle speed, the suspension pressure and the front lane curvature of a semi-trailer are obtained in real time; according to the collected and calculated data and a linear trend extrapolation method, constructing a dynamic rollover risk threshold function and a comprehensive risk index function, and calculating a rollover risk threshold, a real-time risk value and a predicted risk value; the item with the higher value in the real-time risk value and the predicted risk value is selected and compared with a rollover risk threshold value, and whether early warning is triggered or not is judged, and a control instruction is issued; the early warning system comprises a semitrailer state monitoring unit, a lane curvature recognition unit, a rollover risk threshold calculation unit, a real-time risk value calculation unit, a predicted risk value calculation unit and an early warning control unit, and is used for realizing the early warning method, so that the response speed and the judgment accuracy of the semitrailer rollover early warning system under variable working conditions are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of road traffic safety, and particularly relates to a semi-trailer rollover early warning method and system based on dynamic threshold prediction. BACKGROUND

[0002] As the absolute main force of modern highway logistics transportation, semi-trailers provide important support for economic development. However, semi-trailers are prone to rollover accidents under complex working conditions such as high-speed turning, emergency lane changing or heavy load operation, which poses a serious threat to road traffic safety. To reduce the risk of rollover, traditional rollover early warning systems mostly construct early warning models based on the physical parameters of the vehicle itself, such as lateral acceleration, roll angle or load transfer ratio (LTR), and determine whether there is a rollover risk by comparing with a pre-set static threshold. However, due to the complex structure of semi-trailers, the height of the center of gravity is significantly affected by the change of load, and the dynamic response of the vehicle under different speeds, load states and turning radii is quite different. At this time, the fixed threshold judgment mechanism is difficult to accurately reflect the real-time risk level, which may lead to false positives or false negatives, reducing the practicality and reliability of the early warning system. The false positive rate of the traditional fixed threshold rollover early warning method under light load and low speed conditions is usually as high as 20-25%; while under the conditions of high speed, sharp turning and heavy load, the false negative rate may reach 12-20%, and the important reason is that the fixed threshold is too low under the conditions of low speed and light load, and the fixed threshold is too high under the conditions of high speed, sharp turning and heavy load, which leads to false positives and false negatives due to poor adaptability.

[0003] In addition, most existing methods mainly focus on the state monitoring and evaluation of the vehicle body, ignoring the direct influence of the road environment on the rollover risk, especially the role of lane curvature, a kind of geometric information, on the lateral stability of the vehicle in actual driving. For example, on a curved road section with a small radius, even if the vehicle operating parameters are within the normal range, the rollover risk may still increase due to the road structure. Existing early warning methods generally lack the ability to perceive and respond to such external factors, resulting in insufficient early warning accuracy in critical road conditions and posing a great safety hazard. At the same time, traditional rollover early warning models mostly use passive judgment mechanisms and nonlinear prediction models, relying on current state parameters for risk determination, lack of forward-looking prediction ability for risk evolution trend, and cannot provide sufficient response margin for safety control systems, limiting the prevention potential of the system. At the same time, the complexity of the nonlinear model itself makes it difficult to integrate with the existing automatic intervention system of semi-trailers, and the high computational requirement of the nonlinear model itself also makes the engineering implementation more difficult. SUMMARY

[0004] Based on the above, the purpose of the present application is to provide a rollover warning method and system that fuses vehicle dynamic state and road environment information, has dynamic adaptive threshold adjustment and short-term future risk prediction capability, to improve the response speed, judgment accuracy and practical application value of the semi-trailer rollover warning system under variable working conditions.

[0005] To achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0006] The semi-trailer rollover warning method based on dynamic threshold prediction comprises:

[0007] Real-time acquisition of lateral acceleration a of semi-trailer l , yaw rate , vehicle speed V and suspension pressure P, calculation of center of gravity height h based on the suspension pressure P g ;

[0008] Real-time acquisition of front lane curvature k;

[0009] Construction of a dynamic rollover risk threshold function based on the vehicle speed V, the lane curvature k and the center of gravity height h g , calculation of rollover risk threshold R l ;

[0010] Calculation of roll angle θ and load transfer ratio LTR based on the lateral acceleration a l and the center of gravity height h g , construction of a comprehensive risk index function based on the lateral acceleration a l , the roll angle θ and the load transfer ratio LTR as risk factors, calculation of real-time risk value R risk ;

[0011] Using a linear trend extrapolation method, based on the state changes in the current and past period of time, to calculate the predicted lateral acceleration , the predicted roll angle θ pre and the predicted load transfer ratio LTR pre , substituting the predicted lateral acceleration , the predicted roll angle θ pre and the predicted load transfer ratio LTR pre into the comprehensive risk index function to calculate the predicted risk value

[0012] Selecting the value of the real-time risk value R risk and the predicted risk value , comparing it with the rollover risk threshold R l to determine whether to trigger a warning and issue a control instruction.

[0013] As a preferred scheme of the semi-trailer rollover early warning method based on dynamic threshold prediction, the gravity center height h is calculated based on the suspension pressure P g , specifically:

[0014] A nonlinear fitting model is established to estimate the gravity center height of the current vehicle, and the nonlinear fitting model of the gravity center height can be expressed as:

[0015] h g =A·P 2 +B·P+C

[0016] wherein P is the suspension pressure, h g is the gravity center height; the fitting model parameters A, B and C are obtained by static loading experiment fitting.

[0017] As a preferred scheme of the semi-trailer rollover early warning method based on dynamic threshold prediction, the dynamic rollover risk threshold function is constructed according to the vehicle speed V, the lane curvature k and the gravity center height h g , and the rollover risk threshold R l is calculated, specifically:

[0018]

[0019] wherein R0 is the basic rollover risk threshold, k is the lane curvature, k ref is the normalized preset maximum lane curvature, V is the vehicle speed, V ref is the normalized preset maximum vehicle speed, h g is the gravity center height, h gref is the normalized preset maximum gravity center height; α, β and γ are the first empirical weight coefficients, reflecting the influence degree of each factor on the dynamic rollover risk threshold, and the first empirical weight coefficients are obtained according to simulation or real vehicle data, and then fitted by genetic algorithm or regression analysis.

[0020] As a preferred scheme of the semi-trailer rollover early warning method based on dynamic threshold prediction, the lateral acceleration a l and the gravity center height h g are used to calculate the roll angle θ and the load transfer rate LTR, specifically:

[0021]

[0022] wherein a l is the lateral acceleration, h g is the gravity center height, m is the mass of the semi-trailer, is the roll stiffness of the semi-trailer, g is the acceleration of gravity, and t is the vehicle span of the semi-trailer.

[0023] As a preferred scheme of the semi-trailer rollover early warning method based on dynamic threshold prediction, the comprehensive risk index function is constructed by the lateral acceleration a l , the roll angle θ and the load transfer rate LTR as risk factors, and the real-time risk value R is calculated risk , specifically:

[0024]

[0025] Wherein, a l is the lateral acceleration, a lref is the preset maximum lateral acceleration after normalization, θ is the roll angle, θ ref is the preset maximum roll angle after normalization, LTR is the load transfer rate, LTR ref is the preset maximum load transfer rate after normalization; ω1, ω2, ω3 are second empirical weight coefficients, reflecting the influence degree of each factor on the comprehensive risk index, and the second empirical weight coefficients are obtained by genetic algorithm or regression analysis fitting according to simulation or real vehicle data.

[0026] As a preferred scheme of the semi-trailer rollover early warning method based on dynamic threshold prediction, the linear trend extrapolation method is used to predict the predicted lateral acceleration a , the predicted roll angle θ pre and the predicted load transfer rate LTR pre in the short future based on the state changes in the current and the past period of time, specifically:

[0027] The prediction time window is set to define the time interval, the current time is t, the backtracking window length is N, the time step is Δt, and the forward prediction length is τ;

[0028] Data collection, real-time acquisition of lateral acceleration sequence {a l (t-NΔt)......a l (t-2Δt),a l (t-Δt),a l (t)} is carried out, and then the roll angle sequence {θ(t-NΔt)......θ(t-2Δt), θ(t-Δt), θ(t)} and the LTR sequence {LTR(t-NΔt)......LTR(t-2Δt), LTR(t-Δt), LTR(t)} are calculated;

[0029] Linear trend extrapolation, the value of the future time point is predicted by using the linear interpolation extrapolation formula:

[0030] Lateral acceleration prediction: Wherein the slope is

[0031] Roll angle prediction: where the slope is

[0032] LTR prediction: where the slope is

[0033] As a preferred scheme of the semi-trailer rollover early warning method based on dynamic threshold prediction, the predicted lateral acceleration a The predicted roll angle θ pre And the predicted load transfer rate LTR pre Substitute the comprehensive risk indicator function to calculate the predicted risk value R Specifically:

[0034]

[0035] As a preferred scheme of the semi-trailer rollover early warning method based on dynamic threshold prediction, the real-time risk value R risk And the predicted risk value R The value of the higher one is compared with the rollover risk threshold R l To determine whether to trigger an early warning and issue a control instruction, specifically:

[0036] The real-time risk value R risk And the predicted risk value R l , select the higher one and compare it with the rollover risk threshold R l : if It is determined that the risk is below the threshold, and no early warning is triggered; if It is determined that the risk is higher than the threshold, and an early warning is triggered and the vehicle speed is reduced.

[0037] The semi-trailer rollover early warning system based on dynamic threshold prediction is applied to any one of the semi-trailer rollover early warning methods based on dynamic threshold prediction, comprising:

[0038] A semi-trailer state monitoring unit is used to realize the real-time acquisition of the lateral acceleration a l , yaw rate Vehicle speed V and suspension pressure P, and the center of gravity height h g is calculated based on the suspension pressure P;

[0039] A lane curvature recognition unit is used to realize real-time acquisition of the front lane curvature k;

[0040] A rollover risk threshold calculation unit is used to realize the calculation of the rollover risk threshold R gThe dynamic rollover risk threshold function is jointly constructed, and a rollover risk threshold R is calculated l ;

[0041] The real-time risk value calculation unit is configured to calculate a roll angle θ and a load transfer ratio LTR based on the lateral acceleration a l and the height of the center of gravity h g , and to jointly construct a comprehensive risk index function based on the lateral acceleration a l , the roll angle θ and the load transfer ratio LTR as risk factors, and to calculate a real-time risk value R risk ;

[0042] The predicted risk value calculation unit is configured to implement the linear trend extrapolation method, to calculate a predicted lateral acceleration a predicted roll angle θ pre and a predicted load transfer ratio LTR pre based on the current and past state changes in a period of time, and to calculate a predicted risk value R by substituting the predicted lateral acceleration a , the predicted roll angle θ pre and the predicted load transfer ratio LTR pre into the comprehensive risk index function.

[0043] The early warning control unit is configured to compare the real-time risk value R risk and the predicted risk value R with the rollover risk threshold R l , to determine whether to trigger early warning and issue a control instruction.

[0044] As a preferred solution of the semi-trailer rollover early warning method based on dynamic threshold prediction, the semi-trailer ABS anti-lock braking system or ESC electronic stability control system is further included and integrated with the early warning control unit, and the early warning control unit executes a hierarchical control strategy:

[0045] If R is much smaller than R l , the early warning control unit does not trigger early warning.

[0046] If R is close to R l , the early warning control unit triggers L1-level early warning and reminds the driver in the form of sound and light.

[0047] If R is greater than R l , the early warning control unit triggers L2-level early warning and triggers the ABS anti-lock braking system or the ESC electronic stability control system to control the semi-trailer.The pre-warning control unit triggers L2 level pre-warning, and sends a deceleration instruction to the ABS anti-lock braking system or the ESC electronic stability control system while giving an audible and visual prompt, and the engine torque is inhibited;

[0048] If R l The pre-warning control unit triggers L3 level pre-warning, and sends a higher deceleration instruction to the ABS anti-lock braking system or the ESC electronic stability control system while giving an audible and visual prompt, and the engine torque is inhibited and the braking force of the outer wheel is preferentially increased and the differential pressure is limited.

[0049] The beneficial effects of the present application are:

[0050] (1) By means of adaptive adjustment of the dynamic rollover risk threshold, the dynamic rollover risk threshold is closer to the real limit value, the threshold is increased to reduce false alarms when the semitrailer is lightly loaded at low speed, and the threshold is reduced to prevent missed alarms when the semitrailer is heavily loaded at high speed, thereby significantly reducing the false alarm and missed alarm probabilities of the rollover pre-warning system;

[0051] (2) By means of real-time acquisition of the curvature of the front lane, the rollover pre-warning system can adapt to various roads and working conditions, and the rollover risk identification capability is improved, especially in the case of driving on a curve;

[0052] (3) By means of the introduction of the prediction mechanism of "linear trend extrapolation", the rollover pre-warning system can give a warning before the danger occurs, and is particularly suitable for high-speed curves or emergency lane changes, and the linear trend extrapolation method has lower requirements on system computing power and is more conducive to engineering implementation;

[0053] (4) By means of dual-channel evaluation of "real-time risk value" and "predicted risk value", the more dangerous one is selected as the warning logic, and then the value is compared with the rollover risk threshold and the graded pre-warning control strategy is executed, so that the safety is not lost due to prediction failure;

[0054] (5) The system architecture has strong compatibility and can be integrated with the existing ABS system and ESC system of the semitrailer and other automatic intervention systems, thereby improving the application range of the system. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the description of the embodiments of the present application will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the contents of the embodiments of the present application and the drawings.

[0056] Figure 1is a schematic diagram of the overall structure of a semi-trailer rollover early warning system based on dynamic threshold prediction provided by the present application;

[0057] Figure 2 is a rollover risk threshold R l Typical relationship curve of vehicle lane curvature k changing with time. DETAILED DESCRIPTION

[0058] The present application will be further described below in conjunction with the accompanying drawings and examples. It should be understood that the specific examples described herein are intended to be illustrative only and are not in any way limiting of the present application. In addition, it should be noted that only those portions of the drawings that are pertinent to the present application have been included herein for ease of understanding.

[0059] In the description of the present application, unless specifically defined and limited otherwise, the terms "connected", "connected", "fixed" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrated; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or the internal communication of two elements or the interaction relationship between two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0060] In the present application, unless specifically defined and limited otherwise, the first feature "on" or "under" the second feature can include that the first and second features are in direct contact, or that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, the first feature "on", "above" and "above" the second feature includes that the first feature is directly above and obliquely above the second feature, or only indicates that the horizontal height of the first feature is higher than that of the second feature. The first feature "below", "below" and "below" the second feature includes that the first feature is directly below and obliquely below the second feature, or only indicates that the horizontal height of the first feature is less than that of the second feature.

[0061] In the description of the present embodiment, the terms "up", "down", "left", "right" and other orientation or position relationship are based on the orientation or position relationship shown in the drawings, which is only for the convenience of description and simplification of operation, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore cannot be understood as a limitation of the present application. In the description of the present application, unless otherwise stated, the meaning of "a plurality of" is two or more. In addition, the terms "first", "second" are only used to distinguish in the description, and have no special meaning.

[0062] The present application provides a semi-trailer rollover early warning method based on dynamic threshold prediction, as shown in Figure 1 including:

[0063] S1, acquiring the lateral acceleration a of the semitrailer in real time l , the yaw rate vehicle speed V and the suspension pressure P, and calculating the height of the center of gravity h based on the suspension pressure P g ;

[0064] S2, acquiring the curvature k of the front lane in real time

[0065] S3, constructing a dynamic rollover risk threshold function based on the vehicle speed V, the curvature k of the lane and the height of the center of gravity h g to calculate the rollover risk threshold R l ;

[0066] S4, calculating the roll angle θ and the load transfer ratio LTR based on the lateral acceleration a l and the height of the center of gravity h g , and constructing a comprehensive risk index function based on the lateral acceleration a l , the roll angle θ and the load transfer ratio LTR as risk factors to calculate the real-time risk value R risk ;

[0067] S5, using a linear trend extrapolation method to calculate the predicted lateral acceleration predicted roll angle θ pre and predicted load transfer ratio LTR pre , and substituting the predicted lateral acceleration predicted roll angle θ pre and predicted load transfer ratio LTR pre into the comprehensive risk index function to calculate the predicted risk value

[0068] S6, comparing the higher one between the real-time risk value R risk and the predicted risk value with the rollover risk threshold R l to determine whether to trigger a warning and issue a control instruction.

[0069] According to the above method, in S1, the lateral acceleration a of the semitrailer in real time l , the yaw rate vehicle speed V and the suspension pressure P can be specifically acquired as follows: the lateral acceleration a l can be collected by an accelerometer; the yaw rate can be collected by a gyroscope; the lateral acceleration a l and the yaw rate The IMU (Inertial Measurement Unit) can also be used for unified collection; the vehicle speed V can be collected by using the GPS or the wheel speed meter; and the suspension pressure P can be collected by using the suspension air pressure sensor.

[0070] According to the above method, the center of gravity height h g Specifically,

[0071] The nonlinear fitting model is established to estimate the center of gravity height of the current vehicle, and the nonlinear fitting model of the center of gravity height can be expressed as:

[0072] h g =A·P 2 +B·P+C

[0073] wherein P is the suspension pressure, h g is the center of gravity height; the fitting model parameters A, B and C are obtained by fitting the static loading experiment. The TruckSim-Simulink joint simulation or the Adams / Car platform is used for

[0074] According to the above method, the lane curvature k in S2 is obtained in real time, which can be based on the high-precision map and the vehicle positioning system to extract the lane center line track, and the curvature is calculated by the circle fitting, the three-point method or the like; or the vehicle-mounted camera is used to collect images, and the lane boundary is recognized by the deep learning algorithm (such as LaneNet, YOLO, etc.), and the lane curvature k is estimated in combination with the IMU and GPS data. The recognition algorithm used in the present application is not limited to the above-mentioned manner, and can be flexibly adjusted according to the actual semi-trailer system configuration to meet the real-time and recognition accuracy requirements.

[0075] According to the above method, the vehicle speed V, the lane curvature k and the center of gravity height h g are used to jointly build a dynamic rollover risk threshold function to calculate the rollover risk threshold R l , specifically,

[0076]

[0077] wherein R0 is the basic rollover risk threshold, k is the lane curvature, k ref is the preset maximum lane curvature after normalization processing, V is the vehicle speed, V ref is the preset maximum vehicle speed after normalization processing, h g is the center of gravity height, h gref is the preset maximum center of gravity height after normalization processing; α, β and γ are the first empirical weight coefficients, reflecting the influence degree of each factor on the dynamic rollover risk threshold, and the first empirical weight coefficients are obtained according to the simulation or real vehicle data, and then fitted by using the genetic algorithm or regression analysis.

[0078] In the present embodiment, asFigure 2 As shown, multiple sets of vehicle speeds V and center of gravity height h are defined. g Changing the lane curvature k yields the rollover risk threshold R. l Typical relationship curves of lane curvature k: As lane curvature k increases, the rollover risk threshold R... l A decrease in center of gravity means an increased likelihood of rollover, affecting vehicle speed V and center of gravity height h. g Similarly, as the vehicle speed V and center of gravity height h increase... g The increase in the risk of rollover, the threshold R l It will also decrease, so the corresponding typical relationship curves are not listed here. Meanwhile, Figure 2 Used to demonstrate the effect of different parameters on the rollover risk threshold R l The influence trend, rather than a strict fitting curve, is considered in practical applications when the vehicle speed V and center of gravity height h... g When fixed, changes in lane curvature k will also cause changes in the slope of the line. Figure 2 The lines in the diagram are drawn to be nearly parallel only for ease of explanation and do not affect the fundamental principle of this invention.

[0079] Following the method described above, in S4, based on the lateral acceleration a l and center of gravity height h g Calculate the roll angle θ and load transfer rate LTR as follows:

[0080]

[0081] Among them, a l For lateral acceleration, h g Here, m is the height of the center of gravity, and m is the mass of the semi-trailer. Let g be the rolling stiffness of the semi-trailer, g be the acceleration due to gravity, and t be the vehicle span of the semi-trailer. The mass of the semi-trailer, m, is the sum of the mass of the semi-trailer itself and the load mass transported in each trip. The vehicle span and rolling stiffness are basic parameters of each semi-trailer and do not require additional measurement.

[0082] Following the method described above, in S4, based on the lateral acceleration a l The roll angle θ and load transfer rate LTR are used as risk factors to construct a comprehensive risk index function, and the real-time risk value R is calculated. risk Specifically:

[0083]

[0084] Among them, a l Let a be the lateral acceleration. lref The maximum lateral acceleration preset for normalization processing, θ is the roll angle, θ ref The maximum roll angle preset for normalization processing, LTR is the load transfer rate, LTRref The maximum load transfer rate preset by the normalization processing; ω1, ω2, ω3 are second empirical weight coefficients, reflecting the influence degree of each factor on the comprehensive risk index, and the second empirical weight coefficients are obtained by fitting through genetic algorithm or regression analysis according to simulation or real vehicle data.

[0085] Preferably, the genetic algorithm of the second empirical weight coefficients ω1, ω2, ω3 of the comprehensive risk index function includes the following steps in the training process:

[0086] (1) Training data preparation: collect or simulate the key dynamic parameters of the semitrailer under different operating conditions, including lateral acceleration a l , vehicle speed V, lane curvature k, roll angle θ, load transfer rate LTR, and label information of whether the actual rollover occurs. The data source can be a vehicle dynamics simulation platform (such as TruckSim-Simulink joint simulation), or can come from real vehicle road test;

[0087] (2) Initial population generation: randomly generate several groups of second empirical weight coefficients ω 10 , ω 20 , ω 30 as the initial solution of the comprehensive risk index function;

[0088] (3) Definition of fitness function: set the objective function as “maximize the matching accuracy rate of the real-time risk value R risk calculated by the comprehensive risk index function and the actual rollover event”, for example, when the real-time risk value R risk exceeds the threshold value, the system predicts “high risk”, and the statistical accuracy rate is compared with the real rollover event;

[0089] (4) Iterative optimization: in each generation, the population is selected, crossed and mutated according to the fitness function to generate new candidate solutions. After multiple iterations, the population gradually converges to the optimal weight combination;

[0090] (5) Output optimal weight: obtain a set of optimal second empirical weight coefficients (for example, ω1* = 0.35, ω2* = 0.42, ω3* = 0.23) and embed them into the comprehensive risk index function for risk assessment in actual operation.

[0091] (6) Online update: optionally, during the long-term operation of a semitrailer, the system can feed back the new operation data to the training module through periodic retraining or incremental learning mechanism to continuously optimize the weight parameters and improve the adaptive ability of the model.

[0092] Through the above steps, the genetic algorithm can automatically find the optimal parameter combination under multiple working conditions and multiple vehicle types, which significantly improves the universality and prediction accuracy of the risk assessment model compared with fixed weights, and the first empirical weight coefficient of the dynamic rollover risk threshold function can also be obtained by this method.

[0093] According to the above method, the linear trend extrapolation method is used in S5 to predict the predicted lateral acceleration in the short future based on the state changes in the current and past period of time Predicted roll angle θ pre And predicted load transfer rate LTR pre By introducing the linear trend extrapolation method based on real-time evaluation for predicting the trend of risk indicator changes in the short future, in this embodiment, the short future specifically refers to within 0.5s, the setting of 0.5s can both maintain real-time and trigger early warning in advance before danger occurs, leaving response margin for automatic intervention system, and since the prediction time window is short, the nonlinear change or linear change of the dynamic parameters of the vehicle can be approximated as linearization processing, the linear fitting method can be used to realize early warning judgment, which uses the sensor data at the current time and past time, uses linear fitting once, predicts the key dynamic parameters at the future time, and calculates the predicted risk value Specifically:

[0094] The prediction time window is set to define the time interval, the current time is t, the backtracking window length is N, the time step is Δt, and the forward prediction length is τ; in this embodiment, N=3 (i.e. using the last 3 sampling points), Δt=0.1s, and τ=0.5s;

[0095] Data collection, real-time acquisition of lateral acceleration sequence {a l (t-2Δt),a l (t-Δt),a l (t)} is calculated, and the roll angle sequence {θ(t-2Δt), θ(t-Δt), θ(t)} and the LTR sequence {LTR(t-2Δt), LTR(t-Δt), LTR(t)} are calculated;

[0096] Linear trend extrapolation, using linear interpolation extrapolation formula to predict the value at the future time point:

[0097] Lateral acceleration prediction: Where the slope is

[0098] Roll angle prediction: Where the slope is

[0099] LTR prediction: Where the slope is

[0100] According to the above method, the predicted lateral acceleration a The predicted roll angle θ pre And the predicted load transfer rate LTR pre Substitute the comprehensive risk index function to calculate the predicted risk value Specifically:

[0101]

[0102] According to the above method, the real-time risk value R risk And the predicted risk value The value of the higher one is compared with the rollover risk threshold R l To determine whether to trigger an early warning and issue a control command, specifically:

[0103] Compare the real-time risk value R risk And the predicted risk value R l , select the higher one and compare it with the rollover risk threshold R l : if The risk is lower than the threshold value, no early warning is triggered; if The risk is higher than the threshold value, triggering an early warning and reducing the vehicle speed.

[0104] The application also provides a semi-trailer rollover early warning system based on dynamic threshold prediction, which is applied to any one of the semi-trailer rollover early warning methods based on dynamic threshold prediction, comprising:

[0105] The semi-trailer state monitoring unit contains various sensors for real-time acquisition of the lateral acceleration a l , yaw rate Vehicle speed V and suspension pressure P, based on the suspension pressure P to calculate the height of the center of gravity h g ;

[0106] The lane curvature recognition unit realizes real-time acquisition of the front lane curvature k based on the high-precision map or the vehicle-mounted camera and the positioning system.

[0107] The rollover risk threshold calculation unit is used to construct a dynamic rollover risk threshold function according to the vehicle speed V, the lane curvature k and the height of the center of gravity h g As risk factors, the rollover risk threshold R l is calculated.

[0108] The real-time risk value calculation unit is used to calculate the roll angle θ and the load transfer rate LTR according to the lateral acceleration a l And the height of the center of gravity h g , according to the lateral acceleration a lThe lateral inclination angle θ and the load transfer rate LTR jointly construct a comprehensive risk index function, and a real-time risk value R is calculated risk ;

[0109] A prediction risk value calculation unit is configured to implement a linear trend extrapolation method to calculate a predicted lateral acceleration based on the current and past state changes in a period of time A predicted lateral inclination angle θ pre And a predicted load transfer rate LTR pre The predicted lateral acceleration The predicted lateral inclination angle θ pre And the predicted load transfer rate LTR pre are substituted into the comprehensive risk index function to calculate a predicted risk value

[0110] A pre-warning control unit is configured to implement a comparison between the real-time risk value R risk And the predicted risk value with a higher value, and a rollover risk threshold R l to determine whether to trigger a pre-warning and issue a control instruction;

[0111] Preferably, the rollover pre-warning system further comprises an ABS anti-lock braking system or an ESC electronic stability control system (hereinafter referred to as an ABS / ESC system) of the semitrailer, and is integrated with the pre-warning control unit. When the pre-warning system of the present application is combined with the automatic intervention of the ABS / ESC system of the semitrailer, the 0.5s short-time future risk prediction provided by the linear trend extrapolation method can provide a control margin of 0.2-0.3s for the ABS / ESC system, which is sufficient to cover the delay of sensor acquisition, controller operation, CAN communication and hydraulic execution, thereby significantly improving the effectiveness of the rollover prevention intervention in the critical state. At the same time, the pre-warning control unit implements a hierarchical control strategy:

[0112] If R is much smaller than R l , the pre-warning control unit does not trigger a pre-warning;

[0113] If R is close to R l , the pre-warning control unit triggers an L1-level pre-warning, and reminds the driver in the form of sound and light;

[0114] If R exceeds R l , the pre-warning control unit triggers an L2-level pre-warning, and issues a deceleration instruction to the ABS anti-lock braking system or the ESC electronic stability control system while suppressing the engine torque;

[0115] If R is much larger than R lWhen the early warning control unit triggers L3 early warning, it sends higher deceleration instructions to ABS anti-lock braking system or ESC electronic stability control system, suppresses engine torque, and preferentially increases outside wheel braking force and differential pressure limit.

[0116] Preferably, when the early warning system is integrated with the ABS / ESC system of the semitrailer, real-time and low computational power requirements need to be met. Linear trend extrapolation method is more conducive to engineering implementation than more complex nonlinear models. When the linear assumption fails due to low adhesion mutation on the road surface, the early warning system still evaluates the real-time risk value and the predicted risk value through the double-channel evaluation, and the early warning logic takes the more dangerous one of the two, ensuring that safety is not lost due to prediction failure.

[0117] Illustratively, taking a three-axle semitrailer as an example, the curb weight is 10t, the maximum load is 30t, the vehicle center of gravity height is about 1.7m (obtained by static loading test fitting) when fully loaded, the vehicle span t=2.05m, and the vehicle roll stiffness is The bend radius is 150m, the lateral acceleration a l =0.335g, the roll angle θ and the load transfer rate LTR are calculated; the sampling frequency is set to 200Hz, the time step Δt=0.2s, and the forward prediction length τ=0.5s.

[0118] The dynamic rollover risk threshold function is constructed as follows:

[0119] The basic rollover risk threshold R0 and each normalized factor jointly determine the dynamic threshold

[0120] R l =R0·F k (k n )·F v (v n )·F h (h n )

[0121] Where F k , F v and F h are decay functions in the interval [0,1] (i.e., the larger the lane curvature k, vehicle speed V and center of gravity height h g , the smaller F)

[0122] F k (k n )=1-αk n

[0123] F v (v n )=1-βv n

[0124] F k (h n )=1-γh n

[0125]

[0126] Reference existing driving data, determine the normalized reference value, for distinguishing normal driving conditions and limit driving conditions, in the assurance dimensionless and in reasonable range:

[0127] k ref =0.0125m -1 ,v ref =27.78m / s,h ref =1.9m

[0128] The first empirical weight coefficient: [αβγ] = [0.10, 0.08, 0.06] (obtained by genetic algorithm, example in the following calculation of the second empirical weight coefficient)

[0129] When the lane curvature k, vehicle speed V and the height of the center of gravity h g All are reference values or small, the basic rollover risk threshold R0 takes 0.75.

[0130] Construct a comprehensive risk index function:

[0131] (1) Training data construction: collect 200 groups of vehicle dynamic data under different load, speed and curvature, label "whether to overturn" as label;

[0132] (2) Initial model setting: the comprehensive risk index function is set as:

[0133]

[0134] The second empirical weight coefficient [ω1ω2ω3] is randomly generated;

[0135] (3) Genetic algorithm training: set the objective function as "maximize risk prediction accuracy", population size 50, iteration 100 generations, crossover probability 0.8, mutation probability 0.05;

[0136] (4) Optimal solution output: after iteration optimization, the optimal weight combination is obtained:

[0137] [ω1ω2ω3] = [0.15, 0.40, 0.45]

[0138] Comprehensive risk index function (weight after genetic algorithm optimization):

[0139] R risk =ω1a ln +ω2θ n+ ω3 LTR n

[0140]

[0141] With reference to existing driving data, a normalized reference value is determined for distinguishing between normal driving conditions and extreme driving conditions, which, after ensuring dimensionless and within a reasonable range, can be obtained as:

[0142] a lref = 0.5g, θ ref = 0.14 rad, LTR ref = 1

[0143] Second empirical weight coefficient [ω1 ω2 ω3] = [0.15, 0.40, 0.45]

[0144] Condition A: entering a 150m radius curve at 80km / h (smooth entry)

[0145] Measured state

[0146] v = 22.22m / s, R = 150m, k = 0.00667, a l = 0.335g, θ = 0.058 rad, LTR = 0.56

[0147] Time step Δt = 0.2s gives slope

[0148]

[0149] 0.5s linear extrapolation

[0150]

[0151] Real-time / predicted risk indicator

[0152]

[0153] R risk = 0.67*0.15 + 0.4936*0.4 + 0.56*0.45 = 0.55

[0154] Dynamic threshold calculation

[0155] k n = 0.00667 / 0.0125 = 0.533, v n = 22.22 / 27.78 = 0.799, h n = 1.7 / 1.9 = 0.895

[0156] R l = 0.629

[0157] Decision

[0158]

[0159] Risk is lower than threshold, but Approaching R l , triggering L1 level early warning, reminding the driver in the form of sound and light.

[0160] Working condition B: entering a radius of 120m curve at 85km / h (close to the limit)

[0161] Measured state

[0162] v=23.61m / s, R=120m, k=0.0083, a l =0.473g, θ=0.082rad, LTR=0.7845

[0163] Time step Δt=0.2s, slope

[0164]

[0165] 0.5s linear extrapolation

[0166]

[0167] Real-time / predicted risk index

[0168]

[0169] R risk =1.046*0.15+0.643*0.4+0.8675*0.45=0.804

[0170] Dynamic threshold calculation

[0171] k n =0.0083 / 0.0125=0.667, v n =22.22 / 27.78=0.85, h n =1.7 / 1.9=0.895

[0172] R l =0.617

[0173] Judgment

[0174]

[0175] Risk is higher than threshold, triggering L2 level early warning, sound and light prompt at the same time, issuing deceleration instruction to the ABS anti-lock braking system or the ESC electronic stability control system, and inhibiting engine torque.

[0176] The weight coefficients [ω1ω2ω3] and [αβγ] and reference dimension can be obtained by static loading experiment and simulation data fitting; the numerical values listed in this embodiment are preferred examples, and actual deployment can adaptively update the function structure and parameters through regression, genetic algorithm or incremental learning to adapt to different vehicle models and sensor configurations.

[0177] Other embodiments of the disclosure will be apparent to those of ordinary skill in the art from a consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses or adaptations of the disclosure other than those expressly indicated by the claims. The specification and examples are to be regarded as illustrative only, and the true scope and spirit of the disclosure are indicated by the following claims.

[0178] It should be understood that the present disclosure is not limited to the precise structures herein described and illustrated in the drawings, and that various modifications and changes can be made without departing from its scope. The scope of the present disclosure is limited only by the claims that follow.

Claims

1. A semi-trailer rollover early warning method based on dynamic threshold prediction, characterized in that, include: Real-time acquisition of the lateral acceleration a of the semi-trailer l yaw rate The vehicle speed V and suspension pressure P are used to calculate the center of gravity height h based on the suspension pressure P. g ; Real-time acquisition of the curvature k of the lane ahead; Based on the vehicle speed V, the lane curvature k, and the center of gravity height h g We jointly construct a dynamic rollover risk threshold function and calculate the rollover risk threshold R. l ; According to the lateral acceleration a l and the height of the center of gravity h g Calculate the roll angle θ and load transfer rate LTR based on the lateral acceleration a. l The tilt angle θ and the load transfer rate LTR are used as risk factors to construct a comprehensive risk index function, and the real-time risk value R is calculated. risk ; Using a linear trend extrapolation method, based on the current and past state changes, the predicted lateral acceleration for the short term is calculated. Predicting the roll angle θ pre and predicted load transfer rate (LTR) pre The predicted lateral acceleration The predicted roll angle θ pre and the predicted load transfer rate LTR pre Substitute the values ​​into the comprehensive risk index function to calculate the predicted risk value. Select the real-time risk value R risk and the predicted risk value The higher of the values ​​is related to the rollover risk threshold R. l The system compares data to determine whether an alert has been triggered and issues control commands.

2. The semi-trailer rollover early warning method based on dynamic threshold prediction according to claim 1, characterized in that, The center of gravity height h is calculated based on the suspension pressure P. g Specifically: A nonlinear fitting model is established to estimate the current center of gravity height of the vehicle. The nonlinear fitting model for the center of gravity height can be expressed as: h g =A·P 2 +B·P+C Where P is the suspension pressure, h g The height of the center of gravity is given; the fitting model parameters A, B, and C were obtained through static loading experiments.

3. The semi-trailer rollover early warning method based on dynamic threshold prediction according to claim 1, characterized in that, The information is based on the vehicle speed V, the lane curvature k, and the center of gravity height h. g We jointly construct a dynamic rollover risk threshold function and calculate the rollover risk threshold R. l Specifically: Where R0 is the basic rollover risk threshold, and k is the lane curvature, k ref The maximum lane curvature preset for normalization processing, V is the vehicle speed, V ref The maximum vehicle speed preset for normalization processing, h g h is the height of the center of gravity. gref The maximum center of gravity height is preset for normalization processing; α, β, and γ are the first empirical weight coefficients, reflecting the degree of influence of each factor on the dynamic rollover risk threshold. The first empirical weight coefficients are obtained by fitting simulation or real vehicle data through genetic algorithm or regression analysis.

4. The semi-trailer rollover early warning method based on dynamic threshold prediction according to claim 1, characterized in that, According to the lateral acceleration a l and the height of the center of gravity h g Calculate the roll angle θ and load transfer rate LTR as follows: Among them, a l Let h be the lateral acceleration. g The height of the center of gravity is m, and the mass of the semi-trailer is m. Let g be the rolling stiffness of the semi-trailer, g be the gravitational acceleration, and t be the vehicle span of the semi-trailer.

5. The semi-trailer rollover early warning method based on dynamic threshold prediction according to claim 4, characterized in that, According to the lateral acceleration a l The tilt angle θ and the load transfer rate LTR are used as risk factors to construct a comprehensive risk index function, and the real-time risk value R is calculated. risk Specifically: Among them, a l Let a be the lateral acceleration. lref The maximum lateral acceleration preset for normalization processing, θ is the roll angle, θ ref The maximum roll angle preset for normalization processing, LTR is the load transfer rate, LTR ref The maximum load transfer rate preset by the normalization process; ω1, ω2, and ω3 are the second empirical weight coefficients, reflecting the degree of influence of each factor on the comprehensive risk index. The second empirical weight coefficients are obtained by fitting simulation or real vehicle data through genetic algorithm or regression analysis.

6. The semi-trailer rollover early warning method based on dynamic threshold prediction according to claim 1, characterized in that, The method employs linear trend extrapolation to estimate the predicted lateral acceleration in the short term based on current and past state changes. Predicting the roll angle θ pre and predicted load transfer rate (LTR) pre Specifically: The prediction time window is set by defining the time interval, with the current time being t, the backtracking window length being N, the time step being Δt, and the prediction forward duration being τ. Data collection, real-time acquisition of lateral acceleration sequences {a l (t-NΔt)......a l (t-2Δt),a l (t-Δt),a l (t)}, and then calculate the tilt angle sequence {θ(t-NΔt)......θ(t-2Δt),θ(t-Δt),θ(t)} and the LTR sequence {LTR(t-NΔt)......LTR(t-2Δt),LTR(t-Δt),LTR(t)}; Linear trend extrapolation uses linear interpolation formulas to predict values ​​at future time points: Lateral acceleration prediction: Where the slope is Roll angle prediction: Where the slope is LTR Prediction: Where the slope is 7. The semi-trailer rollover early warning method based on dynamic threshold prediction according to claim 6, characterized in that, The predicted lateral acceleration The predicted roll angle θ pre and the predicted load transfer rate LTR pre Substitute the values ​​into the comprehensive risk index function to calculate the predicted risk value. Specifically:

8. The semi-trailer rollover early warning method based on dynamic threshold prediction according to claim 1, characterized in that, The selection of the real-time risk value R risk and the predicted risk value The higher of the values ​​is related to the rollover risk threshold R. l The comparison is performed to determine whether an alert is triggered and to issue a control command. Specifically: Compare the real-time risk value R risk and the predicted risk value R l Select the one with the higher value and compare it with the rollover risk threshold R. l Comparison: If If the risk is determined to be below the threshold, no warning will be triggered; if If the risk is deemed to be higher than the threshold, a warning will be triggered and the vehicle speed will be reduced.

9. A semi-trailer rollover early warning system based on dynamic threshold prediction, applied to the semi-trailer rollover early warning method based on dynamic threshold prediction as described in any one of claims 1-8, characterized in that, include: The semi-trailer status monitoring unit is used to achieve real-time acquisition of the semi-trailer's lateral acceleration 'a'. l yaw rate The vehicle speed V and suspension pressure P are used to calculate the center of gravity height h based on the suspension pressure P. g ; The lane curvature recognition unit is used to obtain the curvature k of the lane ahead in real time. The rollover risk threshold calculation unit is used to calculate the risk based on the vehicle speed V, the lane curvature k, and the center of gravity height h. g We jointly construct a dynamic rollover risk threshold function and calculate the rollover risk threshold R. l ; The real-time risk value calculation unit is used to implement the calculation based on the lateral acceleration a. l and the height of the center of gravity h g Calculate the roll angle θ and load transfer rate LTR based on the lateral acceleration a. l The tilt angle θ and the load transfer rate LTR are used as risk factors to construct a comprehensive risk index function, and the real-time risk value R is calculated. risk ; The predicted risk value calculation unit is used to calculate the predicted lateral acceleration in the short term based on the current and past state changes using the linear trend extrapolation method. Predicting the roll angle θ pre and predicted load transfer rate (LTR) pre The predicted lateral acceleration The predicted roll angle θ pre and the predicted load transfer rate LTR pre Substitute the values ​​into the comprehensive risk index function to calculate the predicted risk value. The early warning control unit is used to select the real-time risk value R. risk and the predicted risk value The higher of the values ​​is related to the rollover risk threshold R. l The system compares data to determine whether an alert has been triggered and issues control commands.

10. The semi-trailer rollover early warning system based on dynamic threshold prediction according to claim 9, characterized in that, It also includes the semi-trailer's ABS anti-lock braking system or ESC electronic stability control system, integrated with the warning control unit, which executes a tiered control strategy: like Much smaller than R l The warning control unit does not trigger a warning; like Approaching R l The warning control unit triggers an L1 level warning, using audible and visual cues to alert the driver; like More than R l The warning control unit triggers an L2 level warning, providing audible and visual alerts, while simultaneously sending a deceleration command to the ABS anti-lock braking system or the ESC electronic stability control system and suppressing engine torque. like Much larger than R l When the warning control unit triggers an L3 level warning and provides audible and visual alerts, it sends a higher deceleration command to the ABS anti-lock braking system or the ESC electronic stability control system to suppress engine torque and prioritize increasing the braking force of the outer wheels and limiting the differential pressure.

Images