A folding-prevention early-warning control system suitable for a semitrailer

Abstract

The application discloses a kind of suitable for semi-trailer's anti-folding early warning control system;The anti-folding early warning control system suitable for semi-trailer includes driver safety module and side car safety module.Driver safety module is collected to the occurrence of folding danger by driver information, vehicle driving information and road information prediction, and flashing warning light reminds driver and completes the operation of brake;Side car safety module is integrated by driver safety module Information and the collection of road side car information, judge the dangerous situation of driving at this time, warn side car driver by flashing warning light and remind lane change and brake;Road safety point is predicted by the analysis of driver safety module and side car safety module, give more safety information to side car;The anti-folding early warning control system suitable for semi-trailer is collected by big data acquisition module Data, calculate optimization using neural network to data, improve the accuracy of prediction probability.

Description

Technical fields:

[0001] This invention relates to the field of traffic accident early warning, specifically to an anti-folding early warning control system for semi-trailers. It is used to provide early warnings of potential folding hazards to semi-trailers by acquiring road environment and vehicle driving status information while the semi-trailer is in motion, thereby increasing driver vigilance and preventing traffic accidents. Background technology:

[0002] With the booming development of the express delivery industry and long-haul transportation, the market demand for semi-trailers is increasing, and the resulting traffic accident problems are becoming more and more serious. Due to the characteristics of semi-trailers, such as large loading capacity, high center of gravity, interlocking articulations, and rear amplification effect, they are prone to folding hazards.

[0003] Given the rapidly increasing demand for semi-trailers, relevant national regulations and corresponding safety protection measures have been gradually improved, from passive safety technology to active safety technology. These have all improved the safety of drivers, but these measures are based on the premise that the danger of folding has already occurred, which also puts forward requirements on the driver's ability. Summary of the Invention:

[0004] The purpose of this invention is to provide an anti-folding warning and control system suitable for semi-trailers. This system acquires road environment and vehicle driving status information to provide early warnings of potential folding hazards in semi-trailers. To achieve the above objective, the technical solution of this invention is as follows:

[0005] A folding prevention and control system for semi-trailers includes a driver safety module, a vehicle safety module, and a big data acquisition module; wherein the driver safety module includes a driver information acquisition submodule, a vehicle road information acquisition submodule, a folding hazard prediction submodule, and a driver safety execution submodule; wherein the vehicle safety module includes a road vehicle information acquisition submodule, a driver safety information assessment submodule, a vehicle safety area prediction submodule, and a vehicle safety execution submodule.

[0006] The driver safety module described in the technical solution is divided into a driver information collection submodule and a vehicle road information collection submodule. The driver information collection submodule collects driver behavior information via a camera built into the cockpit. This driver behavior information is divided into normal behavior information and historical violation information. Normal behavior information includes whether the driver is using a mobile phone for entertainment, whether both hands are on the steering wheel, and the number of eyelid twitches. Attention level is assessed based on this driver behavior information. The vehicle road information collection submodule collects vehicle driving information via onboard sensors, including current speed, yaw rate, lateral acceleration, distance to other vehicles, road surface adhesion coefficient, road slope, road curvature, and articulation angle change rate. The vehicle road information collection submodule also includes a road environment identifier, which identifies the current road conditions based on the road parameters collected by the vehicle road information collection submodule, including icy / snowy roads, low-adhesion roads, steep slope roads, and curvature roads.

[0007] The driver safety module described in the technical solution includes a folding hazard prediction submodule. This submodule processes and integrates information collected by the driver information collection submodule and the vehicle road information collection submodule, and designs a driver behavior bias coefficient and a vehicle road bias coefficient. The driver behavior bias coefficient is defined by the attention level assessment determined by the driver information collection submodule. The vehicle road bias coefficient is jointly determined by a vehicle stability weight coefficient, a road safety weight coefficient, and a folding hazard weight coefficient. The vehicle stability weight coefficient is determined using fuzzy control based on parameters such as current vehicle speed, yaw rate, lateral acceleration, and distance to other vehicles. The road safety weight coefficient is determined using fuzzy control based on parameters such as road adhesion coefficient, road slope, and road curvature. The folding hazard weight coefficient is determined using fuzzy control based on parameters such as the hinge angle change rate. Through these steps, the degree of folding hazard during driving is assessed.

[0008] The driver safety module described in the technical solution includes a driver safety execution submodule; the driver safety execution submodule includes a sensor warning submodule and a behavioral warning submodule; wherein the sensor warning submodule includes voice warnings and flashing lights; wherein the behavioral warning submodule includes braking prompts;

[0009] The roadside vehicle safety module described in the technical solution includes a roadside vehicle information collection submodule and a driver safety information assessment submodule. The roadside vehicle information collection submodule collects information on the number of vehicles beside the road, their speed, lateral distance, and longitudinal distance. The driver safety information assessment submodule receives parameter information calculated by the folding hazard prediction submodule within the driver safety module. Furthermore, the safety level of the vehicles beside the road is evaluated based on the parameter information collected by the roadside vehicle information collection submodule and the driver safety information assessment submodule.

[0010] The vehicle safety module described in the technical solution includes a vehicle safety execution submodule; the vehicle safety zone prediction submodule includes a sensor warning submodule and a behavioral warning submodule; wherein the sensor warning submodule includes voice warning and light flashing; wherein the behavioral warning submodule includes braking prompts and lane change prompts;

[0011] The vehicle safety module described in the technical solution includes a vehicle safety area prediction submodule; the vehicle safety area prediction submodule predicts the safe area for the vehicle to travel by using a dual unscented particle filter algorithm;

[0012] The big data acquisition module described in the technical solution uses a neural network algorithm to train the parameters collected by the driver safety module and the side vehicle safety module, which can more accurately estimate the probability of a folding accident.

[0013] The driver behavior information acquired by the driver information collection submodule in the driver safety module is used to define the attention level evaluation. The attention level evaluation is divided into two levels: focused and inattentive. The evaluation method includes the following steps:

[0014] 1) Calculate the focus level rating using the proposed formula;

[0015] Z x =w1S1+w2S2

[0016] Where S1 and S2 are the normal behavior judgment coefficient and historical violation judgment coefficient, respectively, w1 is the conditional weight coefficient of S1, w2 is the conditional weight coefficient of S2, and Z... x Attention level rating;

[0017] 2) The definition method of S1 is as follows:

[0018] a) If the driver is using a mobile phone for entertainment, S1 is defined as 1; otherwise, S1 is defined as 0.

[0019] b) If the driver does not hold the steering wheel with both hands, S1 is defined as 1; otherwise, S1 is defined as 0.

[0020] c) If the driver's eyelids twitch less than 20 times within half a minute, the driver is considered to be in a state of fatigue, and S1 is defined as 1; otherwise, S1 is defined as 0.

[0021] S2 is defined as follows:

[0022] If a driver has a history of violations and the period is within six months, then S2 is defined as 1; otherwise, S2 is defined as 0.

[0023] 3) The determination method for conditional weight coefficients w1 and w2 adopts fuzzy control, and the specific execution method is as follows: The input of this fuzzy control method is the normal behavior judgment coefficient and the historical violation judgment coefficient. Its fuzzy control subset is defined as S1 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}, and S2 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}; the output is the conditional weight coefficients w1 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}, and w2 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}; where S1 and S2 are the normal behavior judgment coefficient and the historical violation judgment coefficient, respectively, and w1 and w2 are the conditional weight coefficients of S1 and S2, respectively.

[0024] 4) If 1≤Z x If the score is ≤2, the driver's attention level is defined as lazy.

[0025] If 0 ≤ Z x If the score is less than 1, then the driver's attention level is defined as attention.

[0026] The road environment recognition device acquires a view of the road environment ahead via the vehicle's front-facing camera and, based on road surface adhesion coefficient, road slope, and road curvature information collected by the vehicle's road information acquisition submodule, alerts the driver in the cockpit to upcoming road conditions, including icy and snowy roads, low-adhesion roads, steep slopes, and curvature roads. The road condition recognition steps are as follows:

[0027] The road condition is identified by distinguishing the proportion of different road conditions based on the parameter information collected by the vehicle road information collection submodule.

[0028]

[0029] When 0.8≤E≤1, the parameter information fed back by the road environment identifier is 0.1, and the road surface is identified as an icy or snowy road surface.

[0030] When 0.6≤E<0.8, the parameter information fed back by the road environment identifier is 0.3, and the road surface is identified as a low-adhesion road surface.

[0031] When 0.3≤E<0.6, the parameter information fed back by the road environment identifier is 0.5, and the road surface is identified as a steep slope.

[0032] When 0 ≤ E < 0.3, the parameter information fed back by the road environment identifier is 0.8, and the road surface is identified as a curved road surface.

[0033] Where E is the pavement proportion distribution coefficient, δ1 is the ice and snow pavement factor, δ2 is the low-adhesion pavement factor, δ3 is the steep slope pavement factor, and δ4 is the curvature pavement factor.

[0034] The folding hazard prediction submodule processes and integrates information collected by the driver information collection submodule and the vehicle road information collection submodule to determine the degree of folding hazard; the calculation method for the degree of folding hazard is as follows:

[0035] 1)

[0036] Where N z X is the stable value for folding hazards. p C represents the driver's behavioral bias coefficient. p This refers to the vehicle's road deviation coefficient.

[0037] 2)X p Defined by focus level assessment

[0038] If the focus rating is Z x If it is laziness, then X p =1;

[0039] If the focus rating is Z x For focus, then X p =0;

[0040]

[0041] Where w3 is the vehicle stability weighting coefficient, w4 is the road safety weighting coefficient, w5 is the folding hazard weighting coefficient, u is the current vehicle speed, r is the current vehicle yaw rate, and a y L represents the vehicle's current lateral acceleration. f L is the distance to the car in front. r Where φ is the following distance, φ is the road adhesion coefficient, i is the road slope, R is the road curvature, and θ is the hinge angle. Let Δu be the rate of change of the hinge angle, Δr be the instantaneous velocity increment, and Δa be the instantaneous yaw rate increment. y Δθ represents the instantaneous lateral acceleration increment, and Δθ represents the instantaneous hinge angle increment.

[0042] 3) The determination method for the vehicle stability weight coefficient w3, road safety weight coefficient w4, and folding hazard weight coefficient w5 adopts fuzzy control, and the specific execution method is as follows: The inputs to this fuzzy control are the vehicle's current speed, current yaw rate, current lateral acceleration, distance to the vehicle in front, distance to the vehicle behind, road adhesion coefficient, road slope, road curvature, and hinge angle change rate. Its fuzzy subsets are defined as u = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive middle, positive large}, r = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive middle, positive large}, a y ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, L f ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, L r ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, φ = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, i = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, R = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}. That is, {zero, positive, positive small, positive center, positive large}; the output is the vehicle stability weight coefficient w3 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}, the road safety weight coefficient w4 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}, and the folding danger weight coefficient w5 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}; where u is the current vehicle speed, r is the current vehicle yaw rate, and a y L represents the vehicle's current lateral acceleration. f L is the distance to the car in front. r Where φ is the following distance, φ is the road surface adhesion coefficient, i is the road surface slope, and R is the road surface curvature. The rate of change of the hinge angle;

[0043] 4) When the focus level is rated as lazy, X p When N is 1, if z When the value is ≥√2, the probability of the folding danger occurring is determined to be 0.75≤P≤1;

[0044] When the focus rating is "focused", X p When the value is 0, if 1≤N z When the value is less than √2, the probability of the folding hazard occurring is determined to be 0.5 ≤ P < 0.75;

[0045] When the focus rating is "focused", Xp When the value is 0, if 0 ≤ N z When the value is less than 1, the probability of the folding danger occurring is determined to be 0 ≤ P < 0.5.

[0046] The driver safety module estimates the stable value N of the folding hazard based on the folding hazard prediction submodule. z Perform coordinated control of the sensor-based warning submodule and the behavior-based warning submodule;

[0047] 1) When the probability of a folding hazard occurring is 0 ≤ P < 0.5, the hazard level is relatively safe; a yellow light flashes in the cockpit until the probability of a folding hazard is 0 ≤ Z. x <1, at this time the driver's attention level is assessed as focused, the yellow light stops flashing, and the alarm is off;

[0048] 2) When the probability of a folding hazard occurring is 0.5 ≤ P < 0.75, the hazard level is classified as general hazard; a red light flashes in the cockpit, and a voice prompt instructs the driver to apply the brakes until the probability of a folding hazard is 0 ≤ Z. x When u ≤ 30 km / h, the red light stops flashing and the alarm is off;

[0049] 3) When the probability of a folding hazard occurring is 0.75≤P≤1, the hazard level is an emergency hazard; red lights flash simultaneously inside and outside the cockpit, and the driver is prompted by voice to change lanes, brake and pull over to the side of the road until the operation is completed and the alarm is lifted.

[0050] The side vehicle safety module receives information from the folded hazard prediction submodule of the driver safety module, and further determines the degree of danger of the side vehicles by collecting information on the number of side vehicles, their speed, lateral distance, and longitudinal distance. The specific calculation steps are as follows:

[0051] 1)

[0052] Where D p w6 is the hazard level factor for the adjacent vehicle, w6 is the hazard weight coefficient for the adjacent vehicle, n is the number of adjacent vehicles, u1……u n u is the speed of the vehicle next to it. max P is the maximum speed of the vehicle next to it. h P is the lateral distance between adjacent vehicles. z S is the longitudinal distance between the two vehicles, and S is the area of ​​the region with a radius of 100m centered on the center of gravity of the semi-trailer.

[0053] 2) The method for determining the hazard weight coefficient w6 of the adjacent vehicle adopts fuzzy control, and the specific implementation method is as follows: The input of this fuzzy control is the number of adjacent vehicles, the average speed of the adjacent vehicles, the lateral distance between adjacent vehicles, and the longitudinal distance between adjacent vehicles. Its fuzzy subset is defined as n = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive middle, positive large}. That is, {zero, positive, positive small, positive medium, positive large}, P h ={ZE, PO, PS, PM, PB}, which is {zero, positive, positive small, positive medium, positive large}, P z ={ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive medium, positive large}, S = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive medium, positive large}, the output is the side vehicle hazard weight coefficient w6 = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive medium, positive large}, where n is the number of side vehicles. P represents the average speed of the vehicles alongside. h P is the lateral distance between adjacent vehicles. z S is the longitudinal distance between the two vehicles, and S is the area of ​​the region with a radius of 100m centered on the center of gravity of the semi-trailer.

[0054] 3) When N z When ≥1, D p ≥2, at this point, emergency control is required;

[0055] When N z When ≥1, 0≤D p <2, this is simple control;

[0056] When 0≤N z When <1, 0≤D p <2, in this case, no control is applied;

[0057] The vehicle safety module described in the technical solution also includes a vehicle safety execution submodule, whose control mode is as follows:

[0058] 1) When N z When ≥1, D p ≥2 indicates emergency control mode. The sensor warning submodule flashes a red light, and a voice prompt instructs the driver of the adjacent vehicle to brake early and change lanes, until 0≤N. z <1, 0≤D p When the time is less than 2, the red light stops flashing, and the voice warning system stops broadcasting.

[0059] 2) When N z When ≥1, 0≤D p <2, this is the simple control mode. The induction warning light flashes yellow, and a voice prompt tells the driver of the adjacent vehicle to brake until 0≤N z When the yellow light stops flashing after 1 hour, the voice warning system will also stop broadcasting.

[0060] 3) When 0 ≤ N z When <1, 0≤D p <2, this is the no-control mode, and the driver of the adjacent vehicle is driving normally;

[0061] The vehicle safety module described in the technical solution includes a vehicle safety area prediction submodule; the vehicle safety area prediction submodule predicts the safe area for the vehicle to travel by using a dual unscented particle filter algorithm;

[0062] The safe point state prediction equation is as follows:

[0063] x a =f(x) a-1 m a-1 )+n a-1

[0064] Where, x a Let x be the coordinates of the vehicle's safe point position at time a. a = [x0, y0] is the estimated safe zone coordinates, f(x a-1 m a-1 ) represents the state parameters from the previous time step, n a-1 This represents the vibration noise from the previous moment.

[0065] The big data acquisition module described in the technical solution utilizes a neural network algorithm to train the parameters collected by the driver safety module and the side vehicle safety module, thereby more accurately estimating the probability of a folding accident. This neural network includes two sub-neural networks: a driver sub-neural network and a side vehicle sub-neural network. Each sub-neural network is further divided into three layers: an input layer, a hidden layer, and an output layer. For the driver sub-neural network, information collected by the driver safety module is used for training, including driver behavior information and vehicle road information. The driver behavior information includes whether the driver is using a mobile phone for entertainment, etc. Whether the driver is holding the steering wheel with both hands, the number of eyelid twitches, and historical violation information are collected. Vehicle road information includes current speed, yaw rate, lateral acceleration, distance between vehicles, road surface adhesion coefficient, road slope, road curvature, and hinge angle change rate. For the side vehicle sub-neural network, information collected by the side vehicle safety module is used to train the side vehicle sub-neural network. The side vehicle safety information includes the number of side vehicles, the speed of side vehicles, the lateral distance between side vehicles, and the longitudinal distance between side vehicles. The output layers of the driver sub-neural network and the side vehicle sub-neural network output probability values ​​between 0 and 100% to determine the probability of a dangerous accident.

[0066] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0067] 1. The above solution establishes a driver safety module and a side vehicle safety module, which monitors the road environment and the vehicle's own condition, reducing the possibility of folding hazards and improving the driver's driving safety.

[0068] 2. While some existing active safety devices can ensure driving safety to a certain extent, the activation of these safety devices is based on the premise that the folding hazard has already occurred. This invention provides warnings and reminders to the driver and drivers of adjacent vehicles before the folding hazard may occur, thus reducing the possibility of traffic accidents from a higher level. Attached image description:

[0069] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0070] Figure 1 This is a system composition block diagram of the present invention; Detailed implementation method:

[0071] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0072] See Figure 1 , Figure 1 The system composition block diagram of the present invention shows that the anti-folding warning control system for semi-trailers includes a driver safety module, a vehicle safety module, and a big data acquisition module; wherein the driver safety module includes a driver information acquisition submodule, a vehicle road information acquisition submodule, a folding hazard prediction submodule, and a driver safety execution submodule; wherein the vehicle safety module includes a road vehicle information acquisition submodule, a driver safety information assessment submodule, a vehicle safety area estimation submodule, and a vehicle safety execution submodule.

[0073] The driver safety module is divided into a driver information collection submodule and a vehicle road information collection submodule. The driver information collection submodule collects driver behavior information through a camera built into the cockpit. The driver behavior information is divided into normal behavior information and historical violation information. Normal behavior information includes whether the driver is using a mobile phone for entertainment, whether both hands are on the steering wheel, and the number of eyelid twitches. Attention level is assessed based on the driver behavior information. The vehicle road information collection submodule collects vehicle driving information through onboard sensors, including current speed, yaw rate, lateral acceleration, distance to other vehicles, road surface adhesion coefficient, road slope, road curvature, and articulation angle change rate. The vehicle road information collection submodule also includes a road environment identifier, which identifies the current road conditions based on the road parameters collected by the vehicle road information collection submodule, including icy and snowy roads, low-adhesion roads, steep slope roads, and curved roads.

[0074] The driver safety module includes a folding hazard prediction submodule. This submodule processes and integrates information collected by the driver information collection submodule and the vehicle road information collection submodule, and designs driver behavior bias coefficients and vehicle road bias coefficients. The driver behavior bias coefficient is defined by the attention evaluation level proposed by the driver information collection submodule, while the vehicle road bias coefficient is jointly determined by the vehicle stability weight coefficient, road safety weight coefficient, and folding hazard weight coefficient. The vehicle stability weight coefficient is determined using fuzzy control based on parameters such as current vehicle speed, yaw rate, lateral acceleration, and distance to other vehicles. The road safety weight coefficient is determined using fuzzy control based on parameters such as road adhesion coefficient, road slope, and road curvature. The folding hazard weight coefficient is determined using fuzzy control based on parameters such as the hinge angle change rate. Through these steps, the degree of folding hazard during driving is assessed.

[0075] The driver safety module includes a driver safety execution submodule; the driver safety execution submodule includes a sensor warning submodule and a behavioral warning submodule; wherein the sensor warning submodule includes voice warnings and flashing lights; wherein the behavioral warning submodule includes braking prompts.

[0076] The aforementioned vehicle safety module includes a roadside vehicle information collection submodule and a driver safety information assessment submodule. The roadside vehicle information collection submodule collects information on the number of vehicles beside the road, their speed, lateral distance, and longitudinal distance. The driver safety information assessment submodule receives parameter information calculated by the folding hazard prediction submodule within the driver safety module. Furthermore, the safety level of the vehicles beside the road is evaluated based on the parameter information collected by the roadside vehicle information collection submodule and the driver safety information assessment submodule.

[0077] The vehicle safety module includes a vehicle safety execution submodule; the vehicle safety execution submodule includes a sensor warning submodule and a behavior warning submodule; wherein the sensor warning submodule includes voice warning and light flashing; wherein the behavior warning submodule includes braking prompts and lane change prompts.

[0078] The vehicle safety module also includes a vehicle safety area prediction submodule; the vehicle safety area prediction submodule predicts the safe area for the vehicle to travel in by using a dual unscented particle filter algorithm.

[0079] The big data acquisition module uses a neural network algorithm to train the parameters collected by the driver safety module and the side vehicle safety module, which can more accurately estimate the probability of a folding accident.

[0080] The driver information collection submodule in the driver safety module uses driver behavior information to define the attention level evaluation. The attention level evaluation is divided into two levels: focused and inattentive. The evaluation method includes the following steps:

[0081] 1) Calculate the focus level rating using the proposed formula;

[0082] Z x =w1S1+w2S2

[0083] Where S1 and S2 are the normal behavior judgment coefficient and historical violation judgment coefficient, respectively, w1 is the conditional weight coefficient of S1, w2 is the conditional weight coefficient of S2, and Z... x Attention level rating;

[0084] 2) The definition method of S1 is as follows:

[0085] a) If the driver is using a mobile phone for entertainment, S1 is defined as 1; otherwise, S1 is defined as 0.

[0086] b) If the driver does not hold the steering wheel with both hands, S1 is defined as 1; otherwise, S1 is defined as 0.

[0087] c) If the driver's eyelids twitch less than 20 times within half a minute, the driver is considered to be in a state of fatigue, and S1 is defined as 1; otherwise, S1 is defined as 0.

[0088] S2 is defined as follows:

[0089] If a driver has a history of violations and the period is within six months, then S2 is defined as 1; otherwise, S2 is defined as 0.

[0090] 3) The determination method for conditional weight coefficients w1 and w2 adopts fuzzy control, and the specific execution method is as follows: The input of this fuzzy control method is the normal behavior judgment coefficient and the historical violation judgment coefficient. Its fuzzy control subset is defined as S1 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}, S2 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}; the output is the conditional weight coefficients w1 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}, and w2 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}; where S1 and S2 are the normal behavior judgment coefficient and the historical violation judgment coefficient, respectively, and w1 and w2 are the conditional weight coefficients of S1 and S2, respectively; 4) If 1≤Z x If the score is ≤2, the driver's attention level is defined as lazy.

[0091] If 0 ≤ Z xIf the score is less than 1, then the driver's attention level is defined as attention.

[0092] The road environment recognition device acquires a view of the road environment ahead via the vehicle's front-facing camera and, based on road surface adhesion coefficient, road slope, and road curvature information collected by the vehicle's road information acquisition submodule, alerts the driver in the cockpit to upcoming road conditions, including icy and snowy roads, low-adhesion roads, steep slopes, and curvature roads. The road condition recognition steps are as follows:

[0093] The road condition is identified by distinguishing the proportion of different road conditions based on the parameter information collected by the vehicle road information collection submodule.

[0094]

[0095] When 0.8≤E≤1, the parameter information fed back by the road environment identifier is 0.1, and the road surface is identified as an icy or snowy road surface.

[0096] When 0.6≤E<0.8, the parameter information fed back by the road environment identifier is 0.3, and the road surface is identified as a low-adhesion road surface.

[0097] When 0.3≤E<0.6, the parameter information fed back by the road environment identifier is 0.5, and the road surface is identified as a steep slope.

[0098] When 0 ≤ E < 0.3, the parameter information fed back by the road environment identifier is 0.8, and the road surface is identified as a curved road surface.

[0099] Where E is the pavement proportion distribution coefficient, δ1 is the ice and snow pavement factor, δ2 is the low-adhesion pavement factor, δ3 is the steep slope pavement factor, and δ4 is the curvature pavement factor.

[0100] The folding hazard prediction submodule processes and integrates information collected by the driver information collection submodule and the vehicle road information collection submodule to determine the degree of folding hazard; the calculation method for the degree of folding hazard is as follows:

[0101] 1)

[0102] Where N z X is the stable value for folding hazards. p C represents the driver's behavioral bias coefficient. p This refers to the vehicle's road deviation coefficient.

[0103] 2)X p Defined by focus level assessment

[0104] If the focus rating is Z x If it is laziness, then X p =1;

[0105] If the focus rating is Z x For focus, then X p =0;

[0106]

[0107] Where w3 is the vehicle stability weighting coefficient, w4 is the road safety weighting coefficient, w5 is the folding hazard weighting coefficient, u is the current vehicle speed, r is the current vehicle yaw rate, and a y L represents the vehicle's current lateral acceleration. f L is the distance to the car in front. r Where φ is the following distance, φ is the road adhesion coefficient, i is the road slope, R is the road curvature, and θ is the hinge angle. Let Δu be the rate of change of the hinge angle, Δr be the instantaneous velocity increment, and Δa be the instantaneous yaw rate increment. y Δθ represents the instantaneous lateral acceleration increment, and Δθ represents the instantaneous hinge angle increment.

[0108] 3) The determination method for the vehicle stability weight coefficient w3, road safety weight coefficient w4, and folding hazard weight coefficient w5 adopts fuzzy control, and the specific execution method is as follows: The inputs to this fuzzy control are the vehicle's current speed, current yaw rate, current lateral acceleration, distance to the vehicle in front, distance to the vehicle behind, road adhesion coefficient, road slope, road curvature, and hinge angle change rate. Its fuzzy subsets are defined as u = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive middle, positive large}, r = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive middle, positive large}, a y ={ZE, PO, PS, PM, PB}, which is {zero, positive, positive small, positive medium, positive large}, L f ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, L r ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, φ = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, i = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, R = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}. That is, {zero, positive, positive small, positive center, positive large}; the output is the vehicle stability weight coefficient w3 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}, the road safety weight coefficient w4 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}, and the folding danger weight coefficient w5 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}; where u is the current vehicle speed, r is the current vehicle yaw rate, and a y L represents the vehicle's current lateral acceleration. f L is the distance to the car in front. r Where φ is the following distance, φ is the road surface adhesion coefficient, i is the road surface slope, and R is the road surface curvature. The rate of change of the hinge angle;

[0109] 4) When the focus level is rated as lazy, X p When N is 1, if z When the value is ≥√2, the probability of the folding danger occurring is determined to be 0.75≤P≤1;

[0110] When the focus rating is "focused", X p When the value is 0, if 1≤N z When the value is less than √2, the probability of the folding hazard occurring is determined to be 0.5 ≤ P < 0.75;

[0111] When the focus rating is "focused", X p When the value is 0, if 0 ≤ N z When the value is less than 1, the probability of the folding danger occurring is determined to be 0 ≤ P < 0.5.

[0112] The driver safety module estimates the stable value N of the folding hazard based on the folding hazard prediction submodule. z Perform coordinated control of the sensor-based warning submodule and the behavior-based warning submodule;

[0113] 1) When the probability of a folding hazard occurring is 0 ≤ P < 0.5, the hazard level is relatively safe; a yellow light flashes in the cockpit until the probability of a folding hazard is 0 ≤ Z. x <1, at this time the driver's attention level is assessed as focused, the yellow light stops flashing, and the alarm is off;

[0114] 2) When the probability of a folding hazard occurring is 0.5 ≤ P < 0.75, the hazard level is classified as general hazard; a red light flashes in the cockpit, and a voice prompt instructs the driver to apply the brakes until the probability of a folding hazard is 0 ≤ Z. x When u ≤ 30 km / h, the red light stops flashing and the alarm is off;

[0115] 3) When the probability of a folding hazard occurring is 0.75≤P≤1, the hazard level is an emergency hazard; red lights flash simultaneously inside and outside the cockpit, and the driver is prompted by voice to change lanes, brake and pull over to the side of the road until the operation is completed and the alarm is lifted.

[0116] The side vehicle safety module receives information from the folded hazard prediction submodule of the driver safety module, and further determines the degree of danger of the side vehicles by collecting information on the number of side vehicles, their speed, lateral distance, and longitudinal distance. The specific calculation steps are as follows:

[0117] 1)

[0118] Where D p w6 is the hazard level factor for the adjacent vehicle, w6 is the hazard weight coefficient for the adjacent vehicle, n is the number of adjacent vehicles, u1……u n u is the speed of the vehicle next to it. max P is the maximum speed of the vehicle next to it. h P is the lateral distance between adjacent vehicles. z S is the longitudinal distance between the two vehicles, and S is the area of ​​the region with a radius of 100m centered on the center of gravity of the semi-trailer.

[0119] 2) The method for determining the hazard weight coefficient w6 of the adjacent vehicle adopts fuzzy control, and the specific implementation method is as follows: The input of this fuzzy control is the number of adjacent vehicles, the average speed of the adjacent vehicles, the lateral distance between adjacent vehicles, and the longitudinal distance between adjacent vehicles. Its fuzzy subset is defined as n = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive middle, positive large}. That is, {zero, positive, positive small, positive medium, positive large}, P h ={ZE, PO, PS, PM, PB}, which is {zero, positive, positive small, positive medium, positive large}, P z ={ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive medium, positive large}, S = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive medium, positive large}, the output is the side vehicle hazard weight coefficient w6 = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive medium, positive large}, where n is the number of side vehicles. P represents the average speed of the vehicles alongside. h P is the lateral distance between adjacent vehicles. z S is the longitudinal distance between the two vehicles, and S is the area of ​​the region with a radius of 100m centered on the center of gravity of the semi-trailer.

[0120] 3) When N z When ≥1, D p ≥2, at this point, emergency control is required;

[0121] When N z When ≥1, 0≤D p<2, this is simple control;

[0122] When 0≤N z When <1, 0≤D p <2, in this case, no control is applied;

[0123] The vehicle safety module includes a vehicle safety execution submodule, whose control mode is as follows:

[0124] 1) When N z When ≥1, D p ≥2 indicates emergency control mode. The sensor warning submodule flashes a red light, and a voice prompt instructs the driver of the adjacent vehicle to brake early and change lanes, until 0≤N. z <1, 0≤D p When the time is less than 2, the red light stops flashing, and the voice warning system stops broadcasting.

[0125] 2) When N z When ≥1, 0≤D p <2, this is the simple control mode. The induction warning light flashes yellow, and a voice prompt tells the driver of the adjacent vehicle to brake until 0≤N z When the yellow light stops flashing after 1 hour, the voice warning system will also stop broadcasting.

[0126] 3) When 0 ≤ N z When <1, 0≤D p <2, this is the no-control mode, and the driver of the adjacent vehicle is driving normally;

[0127] The vehicle safety module includes a vehicle safety area prediction submodule; the vehicle safety area prediction submodule predicts the safe area for the vehicle to travel in using a dual unscented particle filter algorithm.

[0128] The safe point state prediction equation is as follows:

[0129] x a =f(x) a-1 m a-1 )+n a-1

[0130] Where, x a Let x be the coordinates of the vehicle's safe point position at time a. a = [x0, y0] is the estimated safe zone coordinates, f(x a-1 m a-1 ) represents the state parameters from the previous time step, n a-1 This represents the vibration noise from the previous moment.

[0131] The big data acquisition module utilizes a neural network algorithm to train the parameters collected by the driver safety module and the adjacent vehicle safety module, thereby more accurately estimating the probability of a folding accident. This neural network includes two sub-neural networks: a driver sub-neural network and an adjacent vehicle sub-neural network. Each sub-neural network has three layers: an input layer, a hidden layer, and an output layer. The driver sub-neural network is trained using information collected by the driver safety module, including driver behavior information and vehicle road information. Driver behavior information includes whether the driver is using a mobile phone for entertainment, whether they are holding the steering wheel with both hands, the number of eyelid twitches, and historical violations. Vehicle road information includes current vehicle speed, yaw rate, lateral acceleration, distance between vehicles, road surface adhesion coefficient, road slope, road curvature, and hinge angle change rate. The adjacent vehicle neural network is trained using information collected by the adjacent vehicle safety module, including the number of adjacent vehicles, their speed, lateral distance, and longitudinal distance. The output layers of both the driver and adjacent vehicle neural networks output probability values ​​between 0% and 100% to determine the probability of a folding accident.

[0132] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and equivalent substitutions without departing from the principles of the present invention, and these improvements and equivalent substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A folding prevention and warning control system suitable for semi-trailers, characterized in that: The anti-folding warning control system for semi-trailers includes a driver safety module, a side vehicle safety module, and a big data acquisition module. The driver safety module includes a driver information collection submodule, a vehicle road information collection submodule, a hazard prediction submodule, and a driver safety execution submodule. The side vehicle safety module includes a roadside vehicle information collection submodule, a driver safety information assessment submodule, a side vehicle safety zone estimation submodule, and a side vehicle safety execution submodule. The driver safety module includes a driver information collection submodule that acquires driver behavior information through a camera built into the cockpit. The driver behavior information is divided into normal behavior information and historical violation information. Normal behavior information includes whether there is mobile phone entertainment, whether both hands are on the steering wheel, and the number of eyelid twitches. Attention level is assessed based on driver behavior information; The driver safety module includes a vehicle road information acquisition submodule that collects vehicle driving information through onboard sensors, including current vehicle speed, yaw rate, lateral acceleration, distance to surrounding vehicles, road surface adhesion coefficient, road slope, road curvature, and hinge angle change rate. The vehicle road information acquisition submodule also includes a road environment identifier, which identifies the current road surface condition based on the road parameter information collected by the vehicle road information acquisition submodule, including icy and snowy roads, low-adhesion roads, steep slope roads, and curvature roads. The driver safety module includes a foldable hazard prediction submodule, which processes and integrates the information collected by the driver information collection submodule and the vehicle road information collection submodule, and is designed with driver behavior bias coefficient and vehicle road bias coefficient. The driver behavior bias coefficient is defined by the attention rating obtained from the driver information collection submodule. The vehicle road bias coefficient is jointly determined by the vehicle stability weight coefficient, road safety weight coefficient, and folding hazard weight coefficient. The vehicle stability weight coefficient is determined by fuzzy control using parameters such as current vehicle speed, yaw rate, lateral acceleration, and distance to other vehicles. The road safety weight coefficient is determined by fuzzy control using parameters such as road adhesion coefficient, road slope, and road curvature. The folding hazard weight coefficient is determined by fuzzy control using parameters such as the hinge angle change rate. Through the above steps, the degree of folding hazard during driving is assessed. The driver safety module includes a driver safety execution submodule; the driver safety execution submodule includes a sensor warning submodule and a behavior warning submodule; The sensor-activated warning submodule includes voice alerts and flashing lights; The behavioral warning submodule includes braking prompts; The roadside vehicle safety module includes a roadside vehicle information collection submodule and a driver safety information assessment submodule; The roadside vehicle information collection submodule collects information on the number of vehicles beside the road, their speed, lateral distance, and longitudinal distance. The driver safety information assessment submodule receives parameter information calculated by the folded hazard prediction submodule in the driver safety module; Furthermore, the safety level of the vehicles beside the road is evaluated by the parameter information collected by the roadside vehicle information collection submodule and the driver safety information assessment submodule. The vehicle safety module includes a vehicle safety execution submodule; The vehicle safety execution submodule includes a sensor warning submodule and a behavior warning submodule; The sensor-activated warning submodule includes voice alerts and flashing lights; The behavioral warning submodule includes braking warnings and lane change warnings; The vehicle safety module includes a vehicle safety area prediction submodule; the vehicle safety area prediction submodule predicts the safe area for the vehicle to travel in using a dual unscented particle filter algorithm. The aforementioned anti-folding warning control system for semi-trailers includes a big data acquisition module. The big data acquisition module uses a neural network algorithm to train the parameters collected by the driver safety module and the adjacent vehicle safety module, which can more accurately estimate the probability of a folding accident.

2. The anti-folding warning control system for semi-trailers as described in claim 1, characterized in that... The driver information collection submodule in the driver safety module uses driver behavior information to define the attention level evaluation. The attention level evaluation is divided into two levels: focused and inattentive. The evaluation method includes the following steps: 1) Calculate the focus level rating using the proposed formula; WITH x =w1S1+w2S2 Where S1 and S2 are the normal behavior judgment coefficient and historical violation judgment coefficient, respectively, w1 is the conditional weight coefficient of S1, w2 is the conditional weight coefficient of S2, and Z... x Attention level rating; 2) The definition method of S1 is as follows: a) If the driver is using a mobile phone for entertainment, S1 is defined as 1; otherwise, S1 is defined as 0. b) If the driver does not hold the steering wheel with both hands, S1 is defined as 1; otherwise, S1 is defined as 0. c) If the driver's eyelids twitch less than 20 times within half a minute, the driver is considered to be in a state of fatigue, and S1 is defined as 1; otherwise, S1 is defined as 0. S2 is defined as follows: If a driver has a history of violations and the period is within six months, then S2 is defined as 1; otherwise, S2 is defined as 0. 3) The determination method for conditional weight coefficients w1 and w2 adopts fuzzy control, and the specific execution method is as follows: The input of this fuzzy control method is the normal behavior judgment coefficient and the historical violation judgment coefficient. Its fuzzy control subset is defined as S1 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}, and S2 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}; the output is the conditional weight coefficients w1 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}, and w2 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive medium, positive large}; where S1 and S2 are the normal behavior judgment coefficient and the historical violation judgment coefficient, respectively, and w1 and w2 are the conditional weight coefficients of S1 and S2, respectively. 4) If 1≤Z x If the score is ≤2, the driver's attention level is defined as lazy. If 0 ≤ Z x If the score is less than 1, then the driver's attention level is defined as attention.

3. The anti-folding warning control system for semi-trailers as described in claim 1, characterized in that... The driver safety module includes a road environment identifier. The road environment identifier obtains a view of the road environment ahead through the vehicle's front-facing camera and, based on the road adhesion coefficient, road slope, and road curvature information collected by the vehicle road information collection submodule, alerts the driver in the cockpit to the road conditions that are about to be encountered, including icy and snowy roads, low-adhesion roads, steep slope roads, and curvature roads. The road condition recognition steps are as follows: The road condition is identified by distinguishing the proportion of different road conditions based on the parameter information collected by the vehicle road information collection submodule. When 0.8≤E≤1, the parameter information fed back by the road environment identifier is 0.1, and the road surface is identified as an icy or snowy road surface. When 0.6≤E<0.8, the parameter information fed back by the road environment identifier is 0.3, and the road surface is identified as a low-adhesion road surface. When 0.3≤E<0.6, the parameter information fed back by the road environment identifier is 0.5, and the road surface is identified as a steep slope. When 0 ≤ E < 0.3, the parameter information fed back by the road environment identifier is 0.8, and the road surface is identified as a curved road surface. Where E is the pavement proportion distribution coefficient, δ1 is the ice and snow pavement factor, δ2 is the low-adhesion pavement factor, δ3 is the steep slope pavement factor, and δ4 is the curvature pavement factor.

4. The anti-folding warning control system for semi-trailers as described in claim 1, characterized in that... The folding hazard prediction submodule determines the degree of folding hazard by processing and integrating information collected by the driver information collection submodule and the vehicle road information collection submodule; the calculation method for the degree of folding hazard is as follows: 1) Where N z X is the stable value for folding hazards. p C represents the driver's behavioral bias coefficient. p This refers to the vehicle's road deviation coefficient. 2)X p Defined by focus level assessment If the focus rating is Z x If it is laziness, then X p =1; If the focus rating is Z x For focus, then X p =0; Where w3 is the vehicle stability weighting coefficient, w4 is the road safety weighting coefficient, w5 is the folding hazard weighting coefficient, u is the current vehicle speed, r is the current vehicle yaw rate, and a y L represents the vehicle's current lateral acceleration. f L is the distance to the car in front. r Where φ is the following distance, φ is the road adhesion coefficient, i is the road slope, R is the road curvature, and θ is the hinge angle. Let Δu be the rate of change of the hinge angle, Δr be the instantaneous velocity increment, and Δa be the instantaneous yaw rate increment. y Δθ represents the instantaneous lateral acceleration increment, and Δθ represents the instantaneous hinge angle increment. 3) The determination method for the vehicle stability weight coefficient w3, road safety weight coefficient w4, and folding hazard weight coefficient w5 adopts fuzzy control, and the specific execution method is as follows: The inputs to this fuzzy control are the vehicle's current speed, current yaw rate, current lateral acceleration, distance to the vehicle in front, distance to the vehicle behind, road adhesion coefficient, road slope, road curvature, and hinge angle change rate. Its fuzzy subsets are defined as u = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive middle, positive large}, r = {ZE, PO, PS, PM, PB}, i.e., {zero, positive, positive small, positive middle, positive large}, a y ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, L f ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, L r ={ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, φ = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, i = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}, R = {ZE,PO,PS,PM,PB}, which is {zero, positive, positive small, positive medium, positive large}. That is, {zero, positive, positive small, positive center, positive large}; the output is the vehicle stability weight coefficient w3 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}, the road safety weight coefficient w4 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}, and the folding danger weight coefficient w5 = {ZE, PO, PS, PM, PB}, that is, {zero, positive, positive small, positive center, positive large}; where u is the current vehicle speed, r is the current vehicle yaw rate, and a y L represents the vehicle's current lateral acceleration. f L is the distance to the car in front. r Where φ is the following distance, φ is the road surface adhesion coefficient, i is the road surface slope, and R is the road surface curvature. The rate of change of the hinge angle; 4) When the focus level is rated as lazy, X p When it is 1, like At this time, the probability of the folding danger occurring is determined to be 0.75≤P≤1; When the focus rating is "focused", X p When it is 0, like At this point, the probability of the folding hazard occurring is determined to be 0.5 ≤ P < 0.75; When the focus rating is "focused", X p When it is 0, If 0≤N z When the value is less than 1, the probability of the folding danger occurring is determined to be 0 ≤ P < 0.

5.

5. A folding prevention and warning control system for semi-trailers as described in claim 1, characterized in that... The driver safety module estimates the stable value N of the folding hazard based on the folding hazard prediction submodule. z Control the driver safety execution submodule; 1) When the probability of a folding hazard occurring is 0 ≤ P < 0.5, the hazard level is relatively safe. Yellow lights flash inside the cockpit until 0≤Z x <1, at this time the driver's attention level is assessed as focused, the yellow light stops flashing, and the alarm is off; 2) When the probability of a folding hazard occurring is 0.5 ≤ P < 0.75, the hazard level is classified as general hazard. A red light flashes in the cockpit, and a voice prompt instructs the driver to apply the brakes until 0 ≤ Z. x When u ≤ 30 km / h, the red light stops flashing and the alarm is off; 3) When the probability of a folding hazard occurring is 0.75≤P≤1, the hazard level is emergency hazard; Red lights flash simultaneously inside and outside the cockpit, while a voice prompt reminds the driver to change lanes, brake, and pull over to the side of the road until the operation is completed and the alarm is deactivated.

6. A folding prevention and warning control system for semi-trailers as described in claim 1, characterized in that... The vehicle-side safety module receives information from the folding hazard prediction submodule of the driver safety module, and further determines the degree of danger of the vehicles by collecting information on the number of vehicles, speed of vehicles, lateral distance between vehicles and vehicles, and longitudinal distance between vehicles. The specific calculation steps are as follows: 1) Where D p w6 is the hazard level factor for the adjacent vehicle, w6 is the hazard weight coefficient for the adjacent vehicle, n is the number of adjacent vehicles, u1……u n u is the speed of the vehicle next to it. max P is the maximum speed of the vehicle next to it. h P is the lateral distance between adjacent vehicles. z S is the longitudinal distance between the two vehicles, and S is the area of ​​the region with a radius of 100m centered on the center of gravity of the semi-trailer. 2) The method for determining the hazard weight coefficient w6 of the adjacent vehicle adopts fuzzy control, and the specific implementation method is as follows: The input of this fuzzy control is the number of adjacent vehicles, the average speed of the adjacent vehicles, the lateral distance between adjacent vehicles, and the longitudinal distance between adjacent vehicles. Its fuzzy subset is defined as n={ZE,PO,PS,PM,PB}, that is, {zero, positive, positive small, positive middle, positive large}. That is, {zero, positive, positive small, positive medium, positive large}, P h ={ZE,PO,PS,PM,PB}, which means {zero, positive, positive small, positive medium, positive large}, P z ={ZE,PO,PS,PM,PB}, i.e., {zero, positive, positive small, positive medium, positive large}, S = {ZE,PO,PS,PM,PB}, i.e., {zero, positive, positive small, positive medium, positive large}, the output is the side vehicle hazard weight coefficient w6 = {ZE,PO,PS,PM,PB}, i.e., {zero, positive, positive small, positive medium, positive large}, where n is the number of side vehicles. P represents the average speed of the vehicles alongside. h P is the lateral distance between adjacent vehicles. z S is the longitudinal distance between the two vehicles, and S is the area of ​​the region with a radius of 100m centered on the center of gravity of the semi-trailer. 3) When N z When ≥1, D p ≥2, at this point, emergency control is required; When N z When ≥1, 0≤D p <2, this is simple control; When 0≤N z When <1, 0≤D p <2, in this case, no control.

7. The anti-folding warning control system for semi-trailers as described in claim 1, characterized in that... The vehicle safety module includes a vehicle safety execution submodule; 1) When N z When ≥1, D p ≥2 indicates emergency control mode. The sensor warning submodule flashes a red light, and a voice prompt instructs the driver of the adjacent vehicle to brake early and change lanes, until 0≤N. z <1, 0≤D p When the time is less than 2, the red light stops flashing, and the voice warning system stops broadcasting. 2) When N z When ≥1, 0≤D p <2, this is the simple control mode. The induction warning light flashes yellow, and a voice prompt tells the driver of the adjacent vehicle to brake until 0≤N z When the yellow light stops flashing after 1 hour, the voice warning system will also stop broadcasting. 3) When 0 ≤ N z When <1, 0≤D p <2, This is the no-control mode, and the driver of the adjacent vehicle is driving normally.

8. A folding prevention and warning control system for semi-trailers as described in claim 1, characterized in that... The vehicle safety module includes a vehicle safety area prediction submodule; the vehicle safety area prediction submodule predicts the safe area for the vehicle to travel in using a dual unscented particle filter algorithm. The safe point state prediction equation is as follows: x a =f(x a-1 ,m a-1 )+n a-1 Where, x a Let x be the coordinates of the vehicle's safe point position at time a. a = [x0, y0] is the estimated safe zone coordinates, f(x a-1 ,m a-1 ) represents the state parameters from the previous time step, n a-1 This represents the vibration noise from the previous moment.

9. A folding prevention and warning control system for semi-trailers as described in claim 1, characterized in that... The big data acquisition module utilizes a neural network algorithm to train the parameters collected by the driver safety module and the adjacent vehicle safety module, thereby more accurately estimating the probability of a folding accident. This neural network includes two sub-neural networks: a driver sub-neural network and an adjacent vehicle sub-neural network. Each sub-neural network has three layers: an input layer, a hidden layer, and an output layer. The driver sub-neural network is trained using information collected by the driver safety module, including driver behavior information and vehicle road information. Driver behavior information includes whether the driver is using a mobile phone for entertainment, whether they are holding the steering wheel with both hands, the number of eyelid twitches, and historical violations. Vehicle road information includes current vehicle speed, yaw rate, lateral acceleration, distance between vehicles, road surface adhesion coefficient, road slope, road curvature, and hinge angle change rate. The adjacent vehicle neural network is trained using information collected by the adjacent vehicle safety module, including the number of adjacent vehicles, their speed, lateral distance, and longitudinal distance. The output layers of both the driver and adjacent vehicle neural networks output probability values ​​between 0% and 100% to determine the probability of a folding accident.

Images