Vehicle-cloud integrated wet and slippery road vehicle driving risk early warning method and system

Abstract

The invention relates to a vehicle-cloud integrated wet and slippery road vehicle driving risk early warning method and system, and the method comprises the following steps: obtaining the forward video data of a vehicle, recognizing the type of a current driving road according to the forward video data through a road segmentation and road classification multi-task model, and carrying out the early warning of the driving risk of the vehicle through a pre-established road adhesion coefficient reference table. A road adhesion coefficient is obtained and uploaded to the cloud; continuously counting global road adhesion coefficients, establishing a road adhesion coefficient fusion model, and fusing the global road adhesion coefficients through road section matching and introduction of time difference factors to generate a wet and slippery road map; determining an early warning area corresponding to any wet and slippery road section on the wet and slippery road surface map through a dynamic calculation method; and for a wet and slippery road section, phase plane stability boundaries of different types of vehicles are calculated, typical driving control behaviors are simulated, instability risks are evaluated, and early warning information is sent out. Compared with the prior art, the method achieves the observation of the wet and slippery road surface, and has the advantages of high early warning precision, high pertinence and the like.

Description

Technical Field

[0001] The present invention relates to the field of driving risk warning, and in particular to a vehicle-cloud integrated driving risk warning method and system for vehicles on slippery roads. Background Art

[0002] Highway traffic accidents related to slippery road surfaces such as accumulated water, snow, slush and ice account for a large proportion of traffic accidents, and slippery road surfaces seriously affect driving safety.

[0003] Currently, fixed roadside meteorological monitoring equipment is mainly used to measure meteorological values ​​such as short-term rainfall, snowfall, and temperature, and then the driving risk level is divided according to experience, and finally standardized warning information is pushed to drivers through variable information boards. However, the existing technology lacks direct observation methods for slippery roads, cannot continuously observe the road surface conditions, and has limited application scenarios. In addition, the risk level divided according to experience is too coarse, and the warning information is not targeted, resulting in low driver compliance and poor risk prevention and control effects.

[0004] With the continuous improvement of the penetration rate of advanced driver assistance systems, intelligent connected vehicles have acquired rich and diverse perception and data collection and transmission capabilities, which has made it possible to directly observe road conditions and conduct research on accurate early warning of high-risk vehicles. Therefore, it is very important to develop a risk early warning system with highly targeted early warning information. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art, such as lack of means for observing slippery roads and poor pertinence of early warning information, and to provide a vehicle-cloud integrated slippery road vehicle driving risk early warning method and system.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A vehicle-cloud integrated slippery road vehicle driving risk warning method comprises the following steps: S1: acquiring forward video data of the vehicle, identifying the type of the current driving road according to the forward video data through a road segmentation and road classification multi-task model, obtaining the road adhesion coefficient through a pre-established road adhesion coefficient reference table and uploading it to the cloud;

[0008] S2: Repeat S1, continuously count the road adhesion coefficients of the entire area, establish a road adhesion coefficient fusion model, match the road sections and introduce the time difference factor to fuse the road adhesion coefficients of the entire area and generate a slippery road map;

[0009] S3: Construct a dynamic calculation method for the warning area of ​​the cloud warning system to determine the warning area corresponding to any slippery road section on the slippery road map;

[0010] S4: For specific slippery road sections, calculate the phase plane stability boundaries of different types of vehicles, simulate typical driving control behaviors, evaluate the instability risk of each type of vehicle, determine the warning targets, and issue warning information based on vehicle type.

[0011] Furthermore, step S1 includes the following steps:

[0012] S11: Obtain the forward video data of the vehicle, use the bilinear difference method to unify the image into a fixed size, and then use Z-score standardization to scale the data to obtain a preprocessed image;

[0013] S12: Based on a multi-task learning architecture that shares the underlying feature extraction network, a multi-task model for road segmentation and road classification is constructed to identify the road type corresponding to the preprocessed image;

[0014] S13: Using a pre-established road surface type adhesion coefficient reference table, find the road surface adhesion coefficient corresponding to the road type and upload it to the cloud.

[0015] Further, step S2 includes the following steps:

[0016] S21: repeat S1 to obtain the global road adhesion coefficient uploaded by multiple locations, multiple times, and multiple vehicles;

[0017] S22: Establishing a time series neural network, fusing the road adhesion coefficient of the entire region by calculating the time difference factor, and outputting the fusion value of the road adhesion coefficient of each road section;

[0018] S23: Generate a slippery road map according to the fusion value of the road adhesion coefficient of each road section.

[0019] Furthermore, the calculation expression of the time difference factor is:

[0020]

[0021] In the formula, k Δt is the time difference factor, Δt is the time difference between two different road adhesion coefficients, in hours.

[0022] Furthermore, the computational expression of the temporal neural network is:

[0023] μ t,f =f x (μ t ,μ t-1,f ,k Δt c t-1,f )+b x ,

[0024] c t,f =f c (μ t,μ t-1,f ,c t-1,f )+b c

[0025] In the formula, μ t ,μ t-1,f ,c t-1,f are the inputs, which are the estimated road adhesion coefficient μ of a certain road section at time t. t , historical road adhesion coefficient μ at time t-1 t-1,f , historical network memory value c t-1,f , μ t,f ,c t,f are the outputs, respectively, the fusion road adhesion coefficient μ at time t t,f , the updated network memory value c t,f , f x (·) is the model parameter for predicting output, b x is the model deviation of the predicted output, f c (·) is the model parameter for updating the memory value, b c The model bias used to update the memory value.

[0026] Further, step S3 includes the following steps:

[0027] S31: determining the average speed and speed limit of the real-time traffic flow based on the slippery road map, and calculating the speed reduction range;

[0028] S32: Calculating the braking distance required by the vehicle driver after receiving the warning information;

[0029] S33: Construct a dynamic calculation method for the warning area of ​​the cloud-based warning system, and calculate the length of the warning area based on the speed reduction range, braking distance and traffic density.

[0030] Furthermore, the calculation expression of the speed reduction range is:

[0031]

[0032] In the formula, v t is the average speed of traffic flow at time t, in kilometers per hour, v l is the fusion road adhesion coefficient μ at time t t,f The corresponding speed limit value is in kilometers per hour, v a is the speed reduction amplitude;

[0033] The calculation expression of braking distance is:

[0034]

[0035] Where, d t is the braking distance in meters, vt is the average speed of traffic flow at time t, in kilometers per hour, v l is the speed limit in kilometers per hour, τ1 is the driver’s reaction time after receiving the warning information, τ2 is the brake response time, and a is the driver’s comfortable braking deceleration;

[0036] The calculation expression of the dynamic calculation method of the early warning area is:

[0037] d ROI =d0+d t ×(1+v a ×α)

[0038] Where, d ROI is the length of the warning area in meters; d0 is a fixed spacing, representing the minimum warning area; α is a coefficient set based on traffic density.

[0039] Further, step S4 includes the following steps:

[0040] S41: Establish a multi-type vehicle dynamics parameter library based on the average vehicle speed v of the traffic flow at time t t The fusion road adhesion coefficient at time t is used to calculate the average speed v of each type of vehicle in the warning area. t The center of mass sideslip angle-yaw rate phase plane is obtained, and the stable boundary is drawn using the envelope method to obtain the driving stability domain;

[0041] S42: Use numerical simulation software to calculate the sideslip angle and yaw rate of the center of mass of each type of vehicle under lane-changing maneuvers, and plot them on the phase plane. If they are within the stable domain, they are safe. Otherwise, if they are beyond the stable domain, there is a risk of instability and they are identified as warning objects.

[0042] S43: The cloud issues warning information to the warning objects in the warning area.

[0043] Furthermore, the envelope method includes the center of mass sideslip angle constraint and the yaw rate constraint, and the calculation expression of the envelope method is:

[0044] β lim =α r,sat +l r r / v t ,

[0045]

[0046] In the formula, β lim is the center of mass sideslip angle constraint, r lim is the yaw rate constraint; α r,sat is the rear wheel saturated slip angle, l r is the distance from the center of mass to the rear axle, r is the yaw angle, vt is the average vehicle speed, μ t,f is the road adhesion coefficient; g is the acceleration due to gravity.

[0047] According to a second aspect of the present invention, a vehicle-cloud integrated slippery road vehicle driving risk warning system comprises a cloud server and a client. The cloud server comprises a processor. When the processor runs, it executes any of the above vehicle-cloud integrated slippery road vehicle driving risk warning methods. The client comprises a video acquisition module and a warning notification module. The video acquisition module records and uploads the vehicle's forward video data and road adhesion coefficient to the cloud server. The warning notification module receives and notifies the warning information.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1) Based on the forward video data of the vehicle, the present invention continuously observes the road surface, constructs a road adhesion coefficient fusion model, and generates a slippery road map, which overcomes the limitations of traditional observation methods and provides support for vehicle risk warning on slippery roads. The warning area is determined by a dynamic calculation method of the warning area, and the phase plane stability boundary of different vehicle models on slippery roads is calculated to evaluate the instability risk of different vehicle types and determine the warning object, which is conducive to improving the pertinence of risk warning information push.

[0050] 2) The present invention determines the phase plane stability boundary by calculating the center of mass sideslip angle constraint and the yaw angular velocity constraint, thereby providing a more accurate calculation method for the lateral instability risk assessment of the vehicle on a slippery road surface, improving the risk prevention and control effect, and preventing the vehicle from entering an unstable state.

[0051] 3) The present invention uses a moving car to collect the road adhesion coefficient, reflects the current road conditions in real time, integrates data through the cloud to achieve full-area observation, reduces the practical difficulty of large-scale installation of roadside sensing equipment, and accurately delivers early warning information to each type of vehicle, refines risk assessment results, and achieves the purpose of balancing safety and traffic capacity. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a structural schematic diagram of the present invention;

[0053] Figure 2 A schematic diagram of a multi-task model architecture for road segmentation and road classification using a shared feature extraction network established in an embodiment;

[0054] Figure 3 Schematic diagram of the dimensions of a double lane change test lane in an embodiment;

[0055] Figure 4 Schematic diagram of the stability domain in the embodiment;

[0056] Figure 5Schematic diagram of the architecture of a vehicle slippery road driving risk warning system based on vehicle-cloud integration in an embodiment; DETAILED DESCRIPTION

[0057] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0058] Example 1

[0059] The present invention is a vehicle-cloud integrated early warning method for vehicle driving risk on slippery roads. Figure 1 As shown, the following steps are included:

[0060] S1: Obtain the forward video data of the vehicle, identify the type of the current road based on the forward video data through the road segmentation and road classification multi-task model, obtain the road adhesion coefficient through the pre-established road adhesion coefficient reference table and upload it to the cloud;

[0061] S11: Obtain the forward video data of the vehicle, use the bilinear difference method to unify the image into a fixed size, and then use Z-score standardization to scale the data to obtain a preprocessed image;

[0062] In this embodiment, the front road image is acquired by the vehicle-mounted forward camera Its size is (720×1080), and bilinear interpolation is used to unify the image to a fixed size (224×224), and then Z-score normalization is used to scale the data to obtain the preprocessed input image I(x) norm ;

[0063] The specific calculation formula of bilinear interpolation is as follows:

[0064] I(x) resize =Bilinear(I(x),target_size)

[0065] Where Bilinear(·) is the bilinear interpolation function, target_size is the target size of the image, is the output.

[0066] In this embodiment, Z-score normalization is used to adjust the pixel values ​​on the three channels of RGB of the image to a standard normal distribution to accelerate the convergence of the model. The specific calculation formula is as follows:

[0067]

[0068] In the formula, x i is the image I(x)resize Input on channel i, is the standardized output. In this embodiment, the mean value of the brightness value on the RGB channel is set to μ i ={0.471,0.448,0.408}, the average value of the standard deviation of brightness value σ i ={0.234,0.239,0.242}.

[0069] In this embodiment, the specific calculation formula of the road adhesion coefficient μ is as follows:

[0070] μ=Tab(MulNet(I(x) norm ))

[0071] Wherein, Tab is a reference table of adhesion coefficients of road types, and in this embodiment, the median of the interval of the corresponding adhesion coefficient is selected as μ.

[0072] S12: Based on a multi-task learning architecture that shares the underlying feature extraction network, a multi-task model for road segmentation and road classification is constructed to identify the road type corresponding to the preprocessed image;

[0073] In this embodiment, a lightweight convolutional neural network MobileNetV2 is used as a shared underlying feature extraction network, and then the road segmentation and road classification task heads are built respectively, and the road segmentation and road classification multi-task model MulNet is constructed to quickly identify the road type as snow and lightly compacted snow;

[0074] S13: Using a pre-established road surface type adhesion coefficient reference table, find the road surface adhesion coefficient corresponding to the road type and upload it to the cloud.

[0075] The road surface type adhesion coefficient reference table in this embodiment is shown in Table 1. The corresponding road surface adhesion coefficient interval is [0.25, 035], and the median of the interval is selected as the road surface adhesion coefficient μ=0.30;

[0076] Table 1 Road surface type-adhesion coefficient reference table

[0077]

[0078] S2: Repeat S1, continuously count the road adhesion coefficients of the entire area, establish a road adhesion coefficient fusion model, match the road sections and introduce the time difference factor to fuse the road adhesion coefficients of the entire area and generate a slippery road map;

[0079] S21: repeat S1 to obtain the global road adhesion coefficient uploaded by multiple locations, multiple times, and multiple vehicles;

[0080] The cloud matches different road sections according to the vehicle position, and then collects the road adhesion coefficients uploaded by different vehicles. Multiple vehicles within the continuous observation range are used to obtain the road adhesion coefficients of the entire area.

[0081] S22: Establishing a time series neural network, fusing the road adhesion coefficient of the entire region by calculating the time difference factor, and outputting the fusion value of the road adhesion coefficient of each road section;

[0082] In this embodiment, the following is established: Figure 2 The time series neural network based on LSTM (Long Short-Term Memory) is shown in Figure 1. The time difference factor is proposed to eliminate the large error that may occur in the road adhesion coefficient of a certain section of road over a long period of time. The calculation expression of the time difference factor is:

[0083]

[0084] In the formula, k Δt is the time difference factor, Δt is the time difference between two different road adhesion coefficients, in hours.

[0085] The calculation expression of the temporal neural network is:

[0086] μ t,f =f x (μ t ,μ t-1,f ,k Δt c t-1,f )+b x ,

[0087] c t,f =f c (μ t ,μ t-1,f ,c t-1,f )+b c

[0088] In the formula, μ t ,μ t-1,f ,c t-1,f are the inputs, which are the estimated road adhesion coefficient μ of a certain road section at time t. t , historical road adhesion coefficient μ at time t-1 t-1,f =0.40, historical network memory value c t-1,f =θ t-1 , μ t,f ,c t,f are the outputs, respectively, the fusion road adhesion coefficient μ at time t t,f , the updated network memory value c t,f , f x (·) is the model parameter for predicting output, b xis the model deviation of the predicted output, f c (·) is the model parameter for updating the memory value, b c The model bias used to update the memory value.

[0089] S23: Generate a slippery road map according to the fusion value of the road adhesion coefficient of each road section.

[0090] According to the fusion value of the road adhesion coefficient of different road sections output by the time series neural network, a slippery road map with low adhesion coefficient is generated initially, and then the map is updated according to the fusion value of the road adhesion coefficient at the next moment.

[0091] S3: Construct a dynamic calculation method for the warning area of ​​the cloud warning system to determine the warning area corresponding to any slippery road section on the slippery road map;

[0092] S31: determining the average speed and speed limit of the real-time traffic flow based on the slippery road map, and calculating the speed reduction range;

[0093] In this embodiment, μ is determined based on a low adhesion coefficient slippery road map. t,f =0.36, and obtain the real-time average traffic flow speed v according to the road section speed measurement t = 60km / h, and the speed limit value v under the current adhesion condition is obtained by looking up the table l =30km / h, the speed reduction v is calculated a =50%;

[0094] The calculation expression of the speed reduction range is:

[0095]

[0096] In the formula, v t is the average speed of traffic flow at time t, in kilometers per hour, v l is the fusion road adhesion coefficient μ at time t t,f The corresponding speed limit value is in kilometers per hour, v a is the speed reduction amplitude;

[0097] S32: Calculating the braking distance required by the vehicle driver after receiving the warning information;

[0098] The calculation expression of braking distance is:

[0099]

[0100] Where, d t is the braking distance in meters, v t is the average speed of traffic flow at time t, in kilometers per hour, v lis the speed limit in kilometers per hour, τ1 = 1.5s is the driver's reaction time after receiving the warning information, τ2 = 0.2s is the brake response time, and a = 1.67m / s 2 Braking deceleration for driver comfort;

[0101] In this embodiment, the braking characteristics of the driver are taken into consideration and the braking distance d required by the driver after receiving the warning information is calculated. t =90.7m;

[0102] S33: Construct a dynamic calculation method for the warning area of ​​the cloud-based warning system, and calculate the length of the warning area based on the speed reduction range, braking distance and traffic density.

[0103] The calculation expression of the dynamic calculation method of the early warning area is:

[0104] d ROI =d0+d t ×(1+v a ×α)

[0105] Where, d ROI is the length of the warning area, in meters; d0 is a fixed spacing, representing the minimum warning area; α is a coefficient set based on vehicle density. In this embodiment, α is the traffic operation safety margin value under different traffic flow types. In the free flow state, the vehicle spacing is large, and the margin value is 1; in the synchronous flow state, the vehicle spacing is small, and the margin value is 1.5; in the congested flow state, the vehicle spacing is the smallest, and the margin value is set to 2. The calculation result is d ROI =168.7m.

[0106] S4: For specific slippery road sections, calculate the phase plane stability boundaries of different types of vehicles, simulate typical driving control behaviors, evaluate the instability risk of each type of vehicle, determine the warning targets, and issue warning information based on vehicle type.

[0107] S41: Establish a multi-type vehicle dynamics parameter library based on the average vehicle speed v of the traffic flow at time t t The fusion road adhesion coefficient at time t is used to calculate the average speed v of each type of vehicle in the warning area. t The center of mass sideslip angle-yaw rate phase plane is obtained, and the stable boundary is drawn using the envelope method to obtain the driving stability domain;

[0108] In this embodiment, a vehicle dynamics parameter library is established for five types of vehicles with relatively high market share, namely passenger cars, vans, buses, SUVs, and trucks. t and the adhesion coefficient μ on a slippery road t,f, calculate the sideslip angle-yaw rate phase plane of each type of vehicle under the condition of maintaining the current speed, and use the envelope method to draw the stability boundary to obtain the driving stability domain;

[0109] The envelope method includes the center of mass sideslip angle constraint and the yaw rate constraint. The calculation expression of the envelope method is:

[0110] β lim =α r,sat +l r r / v t ,

[0111]

[0112] In the formula, β lim is the center of mass sideslip angle constraint, r lim is the yaw rate constraint; α r,sat is the rear wheel saturated slip angle, l r is the distance from the center of mass to the rear axle, r is the yaw angle, v z is the average vehicle speed, μ t,f is the road adhesion coefficient; g is the acceleration due to gravity.

[0113] S42: Use numerical simulation software to calculate the sideslip angle and yaw rate of the center of mass of each type of vehicle under lane-changing maneuvers, and plot them on the phase plane. If they are within the stable domain, they are safe. Otherwise, if they are beyond the stable domain, there is a risk of instability and they are identified as warning objects.

[0114] like Figure 3 As shown in the figure, a standard double lane change test lane is used to simulate typical driving maneuvers, and then Matlab is used to calculate the real-time state of the vehicle's center of mass slip angle and yaw angular velocity when the vehicle is driving on the double lane change test lane, and the real-time state of the vehicle's center of mass slip angle and yaw angular velocity is plotted on the phase plane. Figure 4 As shown in part (a) of , if the real-time state of the vehicle is in the stable domain, then this type of vehicle can maintain the current speed and safely pass through the slippery road; otherwise, Figure 4 As shown in part (b) of , if it exceeds the stability domain, then this type of vehicle is at risk of instability when driving on a slippery road and is an object of warning.

[0115] S43: The cloud issues warning information to the warning objects in the warning area.

[0116] In this embodiment, the update cycle of the warning area and the warning object is 10s. Therefore, after the vehicle in the warning area receives the initial warning information, if in any subsequent cycle, when the real-time status of the vehicle in the double lane change test is within the stable domain, there is no need to issue a warning to this type of vehicle; otherwise, the warning will continue and iterative updates will be made to achieve accurate push of warning information, improve driver compliance, and improve risk prevention and control effects.

[0117] Example 2

[0118] Based on Example 1, the present invention also provides a vehicle-cloud integrated slippery road vehicle driving risk warning system, including a cloud server and a vehicle-mounted client. The cloud server includes a processor. When the processor runs, it executes the slippery road vehicle driving risk warning method based on Example 1. The client includes a video acquisition module and a warning notification module. The video acquisition module records and uploads the vehicle's forward video data and road adhesion coefficient to the cloud server, and the warning notification module receives and notifies the warning information.

[0119] In this embodiment, the architecture of the vehicle-cloud integrated vehicle slippery road driving risk warning system is as follows: Figure 5 As shown in the figure, the vehicle estimates the slippery road adhesion coefficient through video data and uploads it to the cloud. The cloud relies on powerful data collection and computing capabilities to generate a slippery road map, determine the warning area, determine the warning vehicle model, and finally push warning information to the vehicle group, forming a vehicle-cloud-vehicle group data flow. In addition, vehicles in the warning area will also upload the estimated slippery road adhesion coefficient at a certain time in the future, thereby realizing continuous observation of the road state and updating of the slippery road map.

[0120] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A vehicle-cloud integrated slippery road vehicle driving risk warning method, characterized in that: The following steps are involved: S1: Obtain the forward video data of the vehicle, identify the type of the current road based on the forward video data through the road segmentation and road classification multi-task model, obtain the road adhesion coefficient through the pre-established road adhesion coefficient reference table and upload it to the cloud; S2: Repeat S1, continuously count the road adhesion coefficients of the entire area, establish a road adhesion coefficient fusion model, match the road sections and introduce the time difference factor to fuse the road adhesion coefficients of the entire area and generate a slippery road map; S3: Construct a dynamic calculation method for the warning area of ​​the cloud warning system to determine the warning area corresponding to any slippery road section on the slippery road map; S4: For specific slippery road sections, calculate the phase plane stability boundaries of different types of vehicles, simulate typical driving control behaviors, evaluate the instability risk of each type of vehicle, determine the warning targets, and issue warning information based on vehicle type.

2. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 1 is characterized in that: The step S1 comprises the following steps: S11: Obtain the forward video data of the vehicle, use the bilinear difference method to unify the image into a fixed size, and then use Z-score standardization to scale the data to obtain a preprocessed image; S12: Based on a multi-task learning architecture that shares the underlying feature extraction network, a multi-task model for road segmentation and road classification is constructed to identify the road type corresponding to the preprocessed image; S13: Using a pre-established road surface type adhesion coefficient reference table, find the road surface adhesion coefficient corresponding to the road type and upload it to the cloud.

3. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 1 is characterized in that: The step S2 comprises the following steps: S21: repeat S1 to obtain the global road adhesion coefficient uploaded by multiple locations, multiple times, and multiple vehicles; S22: Establishing a time series neural network, fusing the road adhesion coefficient of the entire region by calculating the time difference factor, and outputting the fusion value of the road adhesion coefficient of each road section; S23: Generate a slippery road map according to the fusion value of the road adhesion coefficient of each road section.

4. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 3 is characterized in that: The calculation expression of the time difference factor is: In the formula, k Δt is the time difference factor, Δt is the time difference between two different road adhesion coefficients, in hours.

5. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 3 is characterized in that: The calculation expression of the temporal neural network is: m t,f =f x (m t ,m t-1,f ,k Δt c t-1,f )+b x , c t,f =f c (m t ,m t-1,f ,c t-1,f )+b c In the formula, μ t ,μ t-1,f ,c t-1,f are the inputs, which are the estimated road adhesion coefficient μ of a certain road section at time t. t , historical road adhesion coefficient μ at time t-1 t-1,f , historical network memory value c t-1,f , μ t,f ,c t,f are the outputs, respectively, the fusion road adhesion coefficient μ at time t t,f , the updated network memory value c t,f , f x (·) is the model parameter for predicting output, b x is the model deviation of the predicted output, f c (·) is the model parameter for updating the memory value, b c The model bias used to update the memory value.

6. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 1 is characterized in that: The step S3 comprises the following steps: S31: determining the average speed and speed limit of the real-time traffic flow based on the slippery road map, and calculating the speed reduction range; S32: Calculating the braking distance required by the vehicle driver after receiving the warning information; S33: Construct a dynamic calculation method for the warning area of ​​the cloud-based warning system, and calculate the length of the warning area based on the speed reduction range, braking distance and traffic density.

7. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 6 is characterized in that: The calculation expression of the speed reduction range is: In the formula, v t is the average speed of traffic flow at time t, in kilometers per hour, v l is the fusion road adhesion coefficient μ at time t t,f The corresponding speed limit value is in kilometers per hour, v a is the speed reduction amplitude; The calculation expression of the braking distance is: Where, d t is the braking distance in meters, v t is the average speed of traffic flow at time t, in kilometers per hour, v l is the speed limit in kilometers per hour, τ1 is the driver’s reaction time after receiving the warning information, τ2 is the brake response time, and a is the driver’s comfortable braking deceleration; The calculation expression of the early warning area dynamic calculation method is: d ROI =d0+d t ×(1+v a ×α) Where, d ROI is the length of the warning area in meters; d0 is a fixed spacing, representing the minimum warning area; α is a coefficient set based on traffic density.

8. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 1 is characterized in that: The step S4 comprises the following steps: S41: Establish a multi-type vehicle dynamics parameter library based on the average vehicle speed v of the traffic flow at time t t The fusion road adhesion coefficient at time t is used to calculate the average speed v of each type of vehicle in the warning area. t The center of mass sideslip angle-yaw rate phase plane is obtained, and the stable boundary is drawn using the envelope method to obtain the driving stability domain; S42: Use numerical simulation software to calculate the sideslip angle and yaw rate of the center of mass of each type of vehicle under lane-changing maneuvers, and plot them on the phase plane. If they are within the stable domain, they are safe. Otherwise, if they are beyond the stable domain, there is a risk of instability and they are identified as warning objects. S43: The cloud issues warning information to the warning objects in the warning area.

9. The vehicle-cloud integrated slippery road vehicle driving risk warning method according to claim 8, characterized in that: The envelope method includes center of mass sideslip angle constraint and yaw rate constraint, and the calculation expression of the envelope method is: β lim ＝α r,sat +l r r / v t , In the formula, β lim is the center of mass sideslip angle constraint, r lim is the yaw rate constraint; α r,sat is the rear wheel saturated slip angle, l r is the distance from the center of mass to the rear axle, r is the yaw angle, v t is the average vehicle speed, μ t,f is the road adhesion coefficient; g is the acceleration due to gravity.

10. A vehicle-cloud integrated slippery road vehicle driving risk warning system, characterized in that: It includes a cloud server and a client, the cloud server includes a processor, and when the processor is running, it executes a vehicle-cloud integrated slippery road vehicle driving risk warning method as described in any one of claims 1-9, the client includes a video acquisition module and a warning notification module, the video acquisition module records and uploads the vehicle's forward video data and road adhesion coefficient to the cloud server, and the warning notification module receives and notifies the warning information.

Images