Pavement wet skid state analysis method and system based on unit grid division

Abstract

The application provides a pavement wet slip state analysis method and system based on unit grid division. First, a controlled road section is divided into multiple unit grids with unique grid identification information and spatial position boundaries. Then, pavement detection data corresponding to each unit grid is obtained, which contains a set of detection parameters reflecting the pavement surface state. Based on the detection data and the spatial position boundaries, an association model of the unit grid and the pavement wet slip state is established. The model is used to calculate the pavement wet slip state parameters of each unit grid. Finally, the pavement wet slip state analysis result of the entire controlled road section is generated based on the parameters and spatial position boundaries of each unit grid, thereby realizing fine and dynamic monitoring and analysis of the pavement wet slip state.

Description

Technical Field

[0001] The present invention relates to the technical field of road traffic safety monitoring, and in particular to a method and system for analyzing road slippery conditions based on unit grid division. Background Art

[0002] In the field of road traffic safety, slippery roads are one of the important factors that cause traffic accidents. Accurately and timely grasping the slippery state of the road is of great significance for ensuring driving safety and rationally dispatching traffic resources. The traditional method of monitoring the slippery state of the road usually sets up a small number of fixed monitoring points on the road, and obtains road surface information in a limited area through these monitoring points. However, the road is a continuous and complex system, and the road conditions at different locations may vary significantly due to factors such as geographical location, drainage conditions, and surrounding environment. The fixed monitoring point method is difficult to fully and meticulously reflect the slippery state of the road surface in the entire controlled section, and is prone to monitoring blind spots, resulting in inaccurate judgment of the slippery state of the road surface. It is impossible to provide accurate decision-making basis for traffic management departments, and it is impossible to issue safety warnings to drivers in a timely and effective manner, thereby increasing the risk of traffic accidents. Summary of the Invention

[0003] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a method for analyzing a slippery road surface based on unit grid division, the method comprising:

[0004] Divide the controlled road section into multiple unit grids, each unit grid has unique grid identification information and spatial location boundaries;

[0005] Acquire road surface detection data corresponding to each unit grid, wherein the road surface detection data includes a set of detection parameters reflecting the surface state of the road surface;

[0006] Based on the road surface detection data and spatial position boundaries of each unit grid, a correlation model between the unit grid and the road surface slippery state is established;

[0007] Calculating the road surface slippery state parameter of each unit grid using the association relationship model;

[0008] Based on the road surface slippery state parameters and spatial position boundaries of each unit grid, the overall road surface slippery state analysis results of the controlled road section are generated.

[0009] On the other hand, an embodiment of the present invention also provides a road slippery state analysis system based on unit grid division, including a processor and a machine-readable storage medium, the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0010] Based on the above aspects, the embodiment of the present invention realizes the refined segmentation of road space by dividing the controlled road section into multiple unit grids with unique identifiers and spatial position boundaries, and obtains the road surface detection data corresponding to each unit grid, which can accurately capture the road surface state information of each small area, avoiding the problem of overly general monitoring data in traditional methods, and establishes a correlation model between unit grids and road surface slippery conditions, and uses the model to calculate the road surface slippery state parameters of each unit grid, so that the evaluation of road surface slippery conditions is more scientific and accurate. The overall road surface slippery state analysis results of the controlled road section are generated according to the parameters and spatial position boundaries of each unit grid, which can intuitively and comprehensively present the road surface slippery conditions of the entire road section, providing a reliable basis for traffic management departments to formulate targeted traffic control measures and reasonably allocate resources, while also providing drivers with more accurate and timely road condition information, effectively improving the safety and operation efficiency of road traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a schematic diagram of the execution flow of the road slippery state analysis method based on unit grid division provided by an embodiment of the present invention.

[0012] Figure 2 Schematic diagram of exemplary hardware and software components of a road slippery state analysis system based on unit grid division provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 This is a flow chart of a method for analyzing a slippery road surface condition based on unit grid division provided by an embodiment of the present invention. The method for analyzing a slippery road surface condition based on unit grid division is introduced in detail below.

[0014] Step S110: Divide the controlled road section into a plurality of unit grids, each unit grid having unique grid identification information and spatial position boundaries.

[0015] In this example, the controlled road section can be considered a key transportation hub in a city, connecting multiple commercial areas, residential areas, and public facilities. Due to the length and high traffic volume of this section, its road surface conditions are affected by various factors. To more accurately analyze the slippery road conditions, it is necessary to divide it into multiple unit grids with clear boundaries and unique identifiers.

[0016] By dividing the controlled road sections into unit grids, complex road conditions can be refined, allowing subsequent analysis to be carried out on each specific unit grid, thereby improving the accuracy and pertinence of the analysis. The unique grid identification information of each unit grid facilitates its effective management and tracking, while the spatial location boundary clarifies the specific location of the unit grid within the road section.

[0017] Step S111: Analyze the characteristics of the controlled road section to extract the linear direction characteristics, pavement structure type characteristics and environmental characteristics along the controlled road section.

[0018] In this embodiment, linear orientation features can be extracted by, on the one hand, utilizing high-precision Geographic Information System (GIS) technology. This system can acquire detailed geographic data for the road segment, including the coordinates of its starting and ending points, as well as key points along the way. Simultaneously, combined with satellite remote sensing imagery, macroscopic observation and analysis of the overall orientation of the road segment can be performed to determine whether the road segment is linear, curved, or a complex combination of multiple orientations. Furthermore, ground-based measurement equipment, such as total stations and GPS receivers, can be used to conduct field measurements of key points on the road segment, obtaining precise angle and distance information to further accurately determine the linear orientation features of the road segment.

[0019] Extracting pavement structural characteristics requires a combination of field surveys and data review. A professional road survey team should be organized to conduct on-site visits to road sections, carefully observing the pavement's material, texture, color, and other external features to preliminarily determine whether it is cement, asphalt, or another special material. Furthermore, specialized testing equipment, such as pavement radar and hardness testers, should be used to examine the pavement's internal structure and physical properties to accurately determine its structural type. Furthermore, relevant archival materials on road construction should be reviewed to understand the pavement's design standards, construction techniques, and materials used, further verifying and refining the assessment of pavement structural characteristics.

[0020] Extracting environmental features along the route involves multiple aspects. In terms of vegetation coverage, a combination of aerial photogrammetry and field research is used. Aerial photography is used to obtain large-scale vegetation images around the road section, and image recognition technology is used to analyze the type, coverage, and density of vegetation. At the same time, field surveys are conducted on the growth and distribution characteristics of vegetation to obtain more accurate vegetation coverage information. Regarding the distribution type of drainage facilities, on-site inspections are conducted to determine the laying of drainage pipes, the location and spacing of drainage outlets, and other information. The design drawings and maintenance records of the drainage system are also reviewed to fully understand the distribution of drainage facilities. Determining the type of terrain slope relies on professional topographic surveying equipment, such as levels and GPS measuring instruments. By measuring the elevation of multiple points on the road section and calculating the height difference and horizontal distance between different points, the magnitude and direction of the terrain slope can be accurately determined.

[0021] Step S112: Determine the main direction reference axis of the grid division based on the linear trend characteristics, so that the length direction of the unit grid is consistent with the linear trend characteristics of the controlled road section.

[0022] According to the linear trend features extracted above, it is necessary to determine the main direction reference axis of the grid division. If the road section shows an obvious straight line direction, then the main direction reference axis can be set directly along the straight line direction. In the above case, the direction vector of the straight line can be calculated by the coordinate information of the starting point and end point of the road section, which is used as the direction of the main direction reference axis. If there is a curve part in the road section, the main direction reference axis needs to be dynamically adjusted according to the tangent direction of the curve. The specific method is to divide the curve into multiple small segments, calculate the tangent direction of each small segment, and use the tangent direction as the main direction reference axis direction of the small segment. For example, at a bend, multiple points on the bend can be measured by a total station, and the tangent direction of each point can be calculated to determine the main direction reference axis at the bend.

[0023] Keeping the length of the unit grid consistent with the linear orientation of the controlled road section allows the divided unit grid to better fit the actual shape of the road section, avoiding significant deviations between the unit grid and the road section orientation. This allows for a more accurate reflection of the actual road surface during subsequent road surface inspection and analysis, improving the reliability and effectiveness of the analysis results.

[0024] Step S113: Determine the size division basis of the unit grid according to the pavement structure type characteristics and the environmental characteristics along the road, where the size division basis includes the pavement structure change frequency parameter and the environmental influencing factor distribution density parameter.

[0025] For example, step S1131: performing structure change point identification processing on the pavement structure type characteristics, marking the location points where the pavement structure type changes in the controlled road section, and obtaining a set of structure change points.

[0026] In order to accurately obtain the information of the change of the road surface structure, the structure change point identification processing of the road surface structure type characteristics is needed. A professional road survey team is organized, and advanced detection equipment such as road radar and geological detector is equipped to conduct comprehensive and detailed survey on the controlled road section. During the survey, the detection personnel detect the road surface according to a certain interval distance, and record various parameters of the road surface structure. When the parameters of the road surface structure change obviously, the position point is marked as a structure change point. At the same time, the high-precision GPS measuring instrument is used to record the accurate coordinate information of each structure change point. After the survey of the entire road section, all the marked structure change point information is sorted and summarized to form a structure change point set.

[0027] Step S1132: Calculate the length of the road section between adjacent structure change points in the structure change point set, calculate the average value and standard deviation of the length of the road section between all adjacent structure change points, and calculate the road surface structure change frequency parameter based on the average value and the standard deviation. The more dense the structure change points are, the larger the value of the road surface structure change frequency parameter is.

[0028] After obtaining the structure change point set, the length of the road section between adjacent structure change points needs to be calculated. A measuring tool such as a laser range finder or a total station is used to accurately measure the distance between adjacent structure change points. All the measured length data of the road section between adjacent structure change points are input into the statistical analysis software to calculate the average value and the standard deviation of these data. The average value reflects the average interval length of the road section structure change, and the standard deviation reflects the dispersion degree of the road section length data. According to the average value and the standard deviation, the road surface structure change frequency parameter is determined. The specific calculation method is as follows: first, define a function related to the average value and the standard deviation, the input of which is the average value and the standard deviation, and the output of which is the road surface structure change frequency parameter. When the structure change points are more dense, that is, the distance between adjacent structure change points is shorter, the average value will be smaller, and the standard deviation may also be smaller. According to the calculation rule of the function, the value of the road surface structure change frequency parameter will be larger. This indicates that the change of the road surface structure is more frequent, and more detailed division needs to be considered in the unit grid division.

[0029] Step S1133: Perform environmental impact factor identification processing on the environmental characteristics along the line, and extract the environmental impact factor types that affect the road surface wet and slippery state. The environmental impact factor types include vegetation coverage type, drainage facility distribution type and terrain slope type.

[0030] Identifying and extracting environmental factors from the environmental characteristics along the route is a systematic task. Identifying vegetation cover types requires a combination of field observation and specialized image analysis techniques. Field observations provide a direct understanding of vegetation appearance, such as tree species and grass growth. High-resolution satellite or aerial imagery is also used to classify and analyze vegetation using image recognition algorithms to determine vegetation cover types. Different types of vegetation, such as forests, grasslands, and shrubs, have different mechanisms of influence on road slipperiness. Forests may block sunlight, reducing evaporation and increasing road humidity. Grasslands, on the other hand, may have drainage and water storage functions, regulating road humidity.

[0031] Determining the distribution pattern of drainage facilities requires a combination of on-site inspection and documentation. The layout of drainage pipes, the location and spacing of drain outlets, and other information should be inspected on-site. Professional testing equipment, such as a pipe detector, should be used to inspect the internal condition of the pipes. Furthermore, the design drawings and maintenance records of the drainage system should be reviewed to understand the design standards, drainage capacity, and maintenance status of the drainage facilities. The distribution and operation of drainage facilities directly affect the amount of water accumulated on the road surface. Proper drainage facilities can effectively remove accumulated water from the road surface, reducing the risk of slippery roads.

[0032] Determining the type of terrain slope relies on topographic survey data. Using equipment such as a level and GPS, the elevation of multiple points along a road section is measured. By calculating the difference in elevation and horizontal distance between these points, the magnitude and direction of the terrain slope are determined. Terrain slope affects the speed and direction of rainwater flow. On steeper sections, rainwater flows more quickly, making accumulation less likely. On shallower or flatter sections, however, rainwater tends to accumulate, increasing the risk of slippery roads.

[0033] Step S1134: Count the distribution number of each type of environmental influencing factor in the controlled road section, and calculate the distribution number of environmental influencing factors within the unit section length as the environmental influencing factor distribution density parameter. The greater the distribution number, the greater the environmental influencing factor distribution density parameter value.

[0034] After determining the types of various environmental impact factors, it is necessary to count the distribution of each type of environmental impact factor in the controlled road section. For vegetation cover types, the number of vegetation can be counted by manual counting or image analysis software. For example, when using image analysis software, the software can automatically identify and count different types of vegetation based on the color, texture and other characteristics of the vegetation. For drainage facility distribution types, count the number of drainage outlets, the length of drainage pipes, etc. Through field investigations and reference to materials, record the location of each drainage outlet and the laying length of the drainage pipe, and then summarize and count them. For terrain slope types, count the length of the road section in each interval based on the division of different slope intervals. Divide the road section into several intervals according to the slope size, such as 0-5°, 5-10°, etc., and then use the measurement data to count the length of the road section in each interval.

[0035] The number of environmental factors distributed per unit section length is calculated as the environmental factor distribution density parameter. Divide the number of each environmental factor by the total length of the controlled section to obtain the number of factors distributed per unit section length. For example, if the total number of drainage outlets is N and the total length of the controlled section is L, then the number of drainage outlets distributed per unit section length is N / L. A greater number of drainage outlets indicates a denser distribution of the environmental factor within the section, potentially more significant impact on the slippery condition of the road surface, and a higher value for the environmental factor distribution density parameter.

[0036] Step S1135: Establish a comprehensive weight calculation model for the pavement structure change frequency parameter and the environmental influencing factor distribution density parameter, and calculate the comprehensive impact weight value according to a preset weight distribution ratio.

[0037] To comprehensively consider the impact of pavement structure changes and environmental factors on unit grid size, it is necessary to establish a comprehensive weighting calculation model for the pavement structure change frequency parameter and the environmental factor distribution density parameter. First, based on previous research experience and actual conditions, the weight distribution ratio of the pavement structure change frequency parameter and the environmental factor distribution density parameter is preset. For example, assuming that the impact of pavement structure changes on unit grid size is more important, the weight of the pavement structure change frequency parameter is set to 0.6, and the weight of the environmental factor distribution density parameter is set to 0.4.

[0038] Next, a comprehensive weight calculation model is established. This model can be a linear combination model. This model multiplies the pavement structure change frequency parameter by its weight, and the environmental impact factor distribution density parameter by its weight. The two are then added together to obtain a comprehensive impact weight. The specific calculation formula is: Comprehensive Impact Weight = Pavement Structure Change Frequency Parameter × Pavement Structure Change Frequency Parameter Weight + Environmental Impact Factor Distribution Density Parameter × Environmental Impact Factor Distribution Density Parameter Weight. By comprehensively considering these two parameters, a weight value is obtained that reflects the combined impact of pavement structure and environmental factors.

[0039] Step S1136: Determine the reference size of the unit grid based on the comprehensive influence weight value. The larger the comprehensive influence weight value, the smaller the reference size.

[0040] The comprehensive impact weight reflects the combined influence of pavement structural changes and environmental factors on the unit grid size. A high comprehensive impact weight indicates frequent pavement structural changes and complex environmental factors, requiring a more detailed unit grid division. Therefore, the base size should be smaller. Conversely, a low comprehensive impact weight indicates relatively stable road conditions, and the base size of the unit grid can be appropriately increased.

[0041] Establish a functional relationship between the comprehensive impact weight and the reference size. Experimental or empirical data can be used to fit a function whose input is the comprehensive impact weight and whose output is the reference size of the unit grid. For example, assuming a linear function, reference size = ab × comprehensive impact weight, where a and b are constants. Their specific values ​​are determined by fitting actual data. Substituting the calculated comprehensive impact weight into the function will determine the corresponding reference size.

[0042] Step S1137: Integrate the pavement structure change frequency parameter, the environmental impact factor distribution density parameter, and the reference size as a basis for size division of the unit grid.

[0043] Finally, the previously calculated pavement structure change frequency parameters, environmental influencing factor distribution density parameters, and the determined benchmark size are integrated to form a unit grid size division basis. This size division basis comprehensively considers the impact of pavement structure and environmental factors on unit grid size.

[0044] Step S114: performing spatial grid cutting processing on the controlled road section according to the main direction reference axis and the size division basis to generate a preliminary unit grid set, each preliminary unit grid having an initial spatial position boundary.

[0045] Based on the previously determined principal reference axes and the grid cell size, geographic information system (GIS) software is used to create a spatial grid for the controlled road section. First, the geographic information for the road section is entered into the GIS software, including the coordinates of the starting and ending points, intermediate key points, and the boundary information. Next, the grid cell size parameters are set, and the length and width of the grid cells are determined based on the reference size. The software then divides the road section into sections along the principal reference axes, using the reference size as the interval, to generate a preliminary set of grid cells.

[0046] Each preliminary unit grid has its initial spatial position boundaries, which can be accurately represented by geographic coordinates. When generating a unit grid, the software automatically calculates the coordinates of the four vertices of each unit grid to determine its spatial position boundaries. For example, by setting the coordinates of the starting point of the grid and the side length of the grid, the software can calculate the vertex coordinates of each unit grid in turn according to the direction of the main direction reference axis. At the same time, in order to ensure that the division of the unit grid conforms to the actual situation of the road section, it may be necessary to manually adjust some special areas, such as bends, intersections, etc.

[0047] Step S115: performing boundary smoothing processing on the preliminary unit grid set to eliminate overlapping areas and gap areas between the boundaries of adjacent unit grids to obtain the final plurality of unit grids.

[0048] The initially generated set of cell grids may have overlapping boundaries or gaps, which can affect the accuracy of subsequent analysis. Therefore, boundary smoothing is required. In GIS software, specialized boundary processing algorithms are used to adjust the boundaries of the cell grids. For overlapping areas, the algorithm rationally distributes the overlapping portions so that the boundaries of adjacent cell grids connect with each other. Specifically, the area and shape of the overlapping portions are calculated and then allocated to adjacent cell grids according to predefined rules. For example, allocation can be based on the area ratio of adjacent cell grids, with cell grids with larger areas receiving more overlapping portions.

[0049] For gaps, the boundaries of adjacent cell meshes are expanded to fill the gaps, ensuring that there are no gaps between them. When expanding the boundaries, the cell mesh shape and the surrounding environment must be considered to avoid conflicts between the expanded cell mesh and other areas. After multiple iterations and adjustments, overlapping areas and gaps between adjacent cell mesh boundaries are eliminated, resulting in multiple cell meshes with clear and reasonable boundaries. During the process, visualization tools can be used to observe the effects of boundary adjustments in real time to ensure that the adjusted cell mesh meets the requirements.

[0050] Step S116: allocating unique grid identification information to each final unit grid, where the grid identification information includes a road segment number, a grid sequence number, and a coordinate range descriptor of a spatial position boundary.

[0051] To facilitate the management and identification of each unit grid, it is necessary to assign it unique grid identification information. Road section numbers are used to distinguish different controlled road sections. Typically, a set coding rule is used, such as coding based on the road section's geographic location or function. For example, a city's main roads can be coded using a set combination of numbers and letters for easier identification and management.

[0052] Grid sequence numbering is based on the order in which the grid cells are arranged within the road section, starting from the beginning of the section and increasing in order. During the numbering process, the grid cells can be sorted according to their primary reference axis to ensure continuity and logic. For example, the grid cells can be numbered from left to right and from top to bottom.

[0053] The coordinate range descriptor for the spatial location boundary describes the unit grid's boundaries using precise geographic coordinates, including the latitude and longitude coordinates of the unit grid's four vertices. These three pieces of information are combined to form a unique grid identifier for each unit grid. For example, the grid identifier can be expressed as "segment number - grid sequence number (coordinate range descriptor)", effectively identifying the specific location and road segment of each unit grid. In practical applications, this grid identifier information can be stored in a database to facilitate subsequent data query and analysis.

[0054] Step S120: Obtaining road surface detection data corresponding to each unit grid, where the road surface detection data includes a set of detection parameters reflecting the road surface state.

[0055] In order to accurately analyze the slippery state of the road surface, it is necessary to obtain the road surface detection data corresponding to each unit grid. Various types of detection equipment are installed in each unit grid, such as humidity sensors, friction coefficient sensors, temperature sensors, etc. The humidity sensor is used to measure the humidity of the road surface in real time. It can reflect the humidity of the road surface by measuring the water vapor content in the air or the moisture content of the road surface. The friction coefficient sensor is used to measure the friction coefficient between the road surface and the tire. The friction coefficient is an important indicator for measuring the anti-skid performance of the road surface. A lower friction coefficient means that the road surface is more prone to slippery. The temperature sensor is used to measure the temperature of the road surface. The temperature affects both the humidity and friction coefficient of the road surface. For example, a higher temperature may accelerate the evaporation of moisture from the road surface and reduce the humidity of the road surface.

[0056] These detection devices transmit measured data to a data collection center via wireless transmission modules. The data collection center organizes and stores the received data, generating a set of detection parameters reflecting the road surface condition. During data transmission, encryption and verification technologies are employed to ensure data accuracy and reliability, preventing data loss or tampering during transmission. The data collection center also monitors the data in real time and issues alerts when abnormal data is detected, allowing for further inspection and action.

[0057] Step S130: Based on the road surface detection data and spatial position boundaries of each unit grid, a correlation model between the unit grid and the road surface slippery state is established.

[0058] The purpose of establishing a correlation model between unit grids and road slippery conditions is to further explore the inherent connection between road slippery conditions and unit grids, thereby more accurately analyzing road slippery conditions. This requires in-depth mining of road surface detection data for each unit grid and combining its spatial location and boundary information.

[0059] Step S131: performing feature extraction processing on the road surface detection data of each unit grid to obtain a detection feature set of each unit grid. The detection feature set includes numerical features and categorical features in the road surface detection data.

[0060] Step S1311: performing data type identification processing on the road surface detection data of the unit grid to distinguish between numerical detection data and categorical detection data.

[0061] When extracting features from road surface inspection data, the first step is to identify the data type. A specialized data processing program is written to scan and analyze the collected road surface inspection data. Based on the format and content of the data, the program determines whether each data item is numerical or categorical. For example, humidity, friction coefficient, and temperature values ​​are numerical data that can be represented by specific numerical values ​​and can be mathematically calculated. However, information such as the presence of water or oil on the road surface is categorical data and is typically represented by discrete categories, such as "water present" and "no water present."

[0062] Step S1312: extract features from the numerical detection data and calculate statistical feature parameters of the numerical detection data. The statistical feature parameters include the central tendency parameter, dispersion parameter and distribution morphology parameter of the data sequence.

[0063] For numerical detection data, further feature extraction is performed. Statistical feature parameters are calculated, including central tendency parameters, dispersion degree parameters, and distribution shape parameters. The central tendency parameter is usually represented by the average value, which is obtained by adding the numerical data in a period of time and dividing by the number of data. The average value reflects the average level of the data. The dispersion degree parameter can be measured by the standard deviation, which is the square root of the average value of the square sum of the difference between each data and the average value. The standard deviation reflects the fluctuation of the data. The distribution shape parameter can be obtained by drawing a histogram of the data, observing the shape of the data distribution, and determining whether it is a normal distribution, a skewed distribution, or other distribution types. For example, if the data is normally distributed, the data distribution is relatively uniform; if it is skewed, the data has a certain bias.

[0064] Step S1313: Feature extraction is performed on the categorical detection data, and a category encoding method is used to convert the categorical detection data into a numerical category feature vector. The category encoding method is determined according to the number of categories and the relationship between categories of the categorical detection data.

[0065] For categorical detection data, it needs to be converted into a numerical category feature vector. According to the number of categories and the relationship between categories of the categorical data, a suitable encoding method is selected. If the categorical data has only two categories, such as whether the road surface has water (1 for water and 0 for no water), binary encoding can be used. The above encoding method is simple and intuitive, and can effectively convert categorical data into numerical data. If the categorical data has multiple categories, such as the pollution types of the road surface, including oil, dirt, and others, the One-Hot Encoding method can be used to represent each category with a binary vector, with only the corresponding category position being 1 and the other positions being 0. For example, for the oil category, it can be represented as [1, 0, 0]; for the dirt category, it can be represented as [0, 1, 0]; and for the other category, it can be represented as [0, 0, 1]. Through the above encoding method, the categorical data can be converted into a numerical feature vector suitable for machine learning model processing.

[0066] Step S1314: Perform feature dimension alignment processing on the statistical feature parameters and the category feature vector, so that the numerical features and the categorical features have the same number of samples and timestamp labels.

[0067] To ensure effective merging and analysis of numerical and categorical features, the statistical feature parameters and categorical feature vectors must be aligned. By matching the data's timestamps, we filter out numerical and categorical feature data with the same timestamp, ensuring they have the same sample size. For example, if a time period contains only numerical data and no categorical data, or vice versa, the data for that time period is removed to ensure data consistency. Furthermore, the data is sorted and arranged in timestamp order to facilitate subsequent analysis and processing.

[0068] Step S1315: perform feature normalization on the dimensionally aligned statistical feature parameters and category feature vectors, and perform feature merging on the normalized statistical feature parameters and category feature vectors to generate a detection feature set containing numerical features and categorical features. Each feature in the detection feature set has a unique feature name and feature value range description.

[0069] After dimension alignment, perform feature normalization on the statistical feature parameters and categorical feature vectors to scale the data to a set range, such as the interval [0, 1]. You can use a normalization algorithm, such as min-max normalization, to subtract the minimum value from each data item and then divide it by the difference between the maximum and minimum values ​​to obtain the normalized data. Normalization eliminates dimensional differences between different features, allowing the model to treat each feature more fairly.

[0070] Merge the normalized statistical feature parameters and categorical feature vectors to generate a detection feature set containing both numerical and categorical features. During the merging process, the statistical feature parameters and categorical feature vectors are concatenated in a specified order to form a new feature vector. Assign each feature a unique name and clearly describe its value range for subsequent analysis and use. For example, the humidity average feature might be named "Humidity Average" and its value range described as [0, 1].

[0071] Step S132: extracting spatial topological relationship features from the spatial position boundary of each unit grid, where the spatial topological relationship features include spatial distance parameters and direction orientation parameters of adjacent unit grids.

[0072] Step S1321: performing boundary coordinate extraction processing on the spatial position boundary of each unit grid to obtain a vertex coordinate set of the spatial position boundary.

[0073] The spatial position boundary of each unit grid is processed using GIS software. The software extracts the coordinates of the four vertices of the unit grid based on the stored spatial information of the unit grid, forms a set of vertex coordinates of the spatial position boundary, and thus accurately describes the spatial position and shape of the unit grid. When extracting the coordinates, the accuracy and accuracy of the coordinates are ensured, and a high-precision coordinate system such as the WGS84 coordinate system is used. At the same time, the coordinate data is checked and corrected to prevent errors or missing coordinates.

[0074] Step S1322: Calculate the geometric center coordinates of each unit grid based on the set of vertex coordinates. The geometric center coordinates are calculated by the arithmetic mean of the vertex coordinates.

[0075] According to the extracted set of vertex coordinates, the geometric center coordinates of each unit grid are calculated. The horizontal coordinates of the four vertices are added and divided by 4 to obtain the horizontal coordinates of the geometric center; the vertical coordinates of the four vertices are added and divided by 4 to obtain the vertical coordinates of the geometric center. The geometric center coordinates can represent the center position of the unit grid.

[0076] Step S1323: Traverse all unit grids of the controlled road section to determine the set of adjacent unit grids of each unit grid, which includes other unit grids that have common edges or common vertices with the spatial position boundary of the current unit grid.

[0077] All unit grids of the controlled road section are checked one by one by writing a traversal algorithm. For each unit grid, it is determined whether other unit grids have common edges or common vertices with its spatial position boundary. If so, the unit grid is included in the set of adjacent unit grids of the current unit grid. In the judgment process, a spatial analysis algorithm is used to compare the boundary coordinates of the unit grids to determine whether there are common edges or common vertices. For example, it can be determined whether two unit grids are adjacent by checking whether their vertex coordinates are the same or their edge coordinates are coincident.

[0078] Step S1324: Calculate the straight-line distance between the geometric center coordinates of the current unit grid and the geometric center coordinates of each adjacent unit grid in the set of adjacent unit grids, and take the straight-line distance as the spatial distance parameter of the adjacent unit grid.

[0079] According to the calculated coordinates of the geometric center of the unit grid, the straight-line distance between the geometric center of the current unit grid and the geometric center of each adjacent unit grid in the set of adjacent unit grids is calculated. The straight-line distance is calculated by the square root of the sum of the squares of the coordinate differences using the distance calculation formula. These straight-line distances are taken as the spatial distance parameters of the adjacent unit grids, which reflect the spatial proximity between the unit grids. In the calculation process, considering the curvature of the earth and the conversion of the coordinate system, a distance formula suitable for geographic coordinate calculation, such as the Haversine formula, is used to ensure the accuracy of the calculation results.

[0080] Step S1325: According to the coordinates of the geometric center of the current unit grid and the coordinates of the geometric center of the adjacent unit grid, the direction azimuth of the adjacent unit grid relative to the current unit grid is calculated, with the geometric center of the current unit grid as the origin and the linear direction of the control section as the reference axis.

[0081] The direction azimuth of the adjacent unit grid relative to the current unit grid is calculated with the geometric center of the current unit grid as the origin and the linear direction of the control section as the reference axis. The direction azimuth is calculated by the method of trigonometric functions according to the difference between the coordinates of the geometric centers of the two unit grids. The specific calculation process is as follows: First, the difference between the coordinates of the geometric center of the adjacent unit grid and the coordinates of the geometric center of the current unit grid on the x-axis and y-axis is calculated to obtain Δx and Δy. Then, the azimuth θ is calculated using the arctangent function: θ = arctan(Δy / Δx). Finally, according to the positive and negative of Δx and Δy, the azimuth is adjusted to be within the range of 0-360°. The direction azimuth can describe the direction relationship of the adjacent unit grid relative to the current unit grid.

[0082] Step S1326: The spatial distance parameters and the direction azimuth are dimensionally expanded according to the number of adjacent unit grids to generate a spatial topological relationship feature vector with fixed dimensions, and each element in the spatial topological relationship feature vector corresponds to a combination of the spatial distance parameter and the direction azimuth parameter of an adjacent unit grid.

[0083] The calculated spatial distance parameters and direction azimuths are dimensionally expanded according to the number of adjacent unit grids. For each adjacent unit grid, its spatial distance parameter and direction azimuth are combined into an element to form a vector. In this way, a spatial topological relationship feature vector with fixed dimensions is generated, which facilitates subsequent splicing and analysis with other features. In the expansion process, if the number of adjacent unit grids is less than the preset dimension, a set value (such as 0) is used for padding; if the number of adjacent unit grids exceeds the preset dimension, the unit grids with closer distances are selected for retention.

[0084] Step S133: performing feature concatenation processing on the detection feature set of each unit grid and the spatial topological relationship feature to generate a joint feature vector of the unit grid. The dimension of the joint feature vector is equal to the sum of the dimensions of the detection feature set and the spatial topological relationship feature.

[0085] After obtaining the detection feature set and spatial topological relationship features for each unit grid, they are then spliced ​​together. Since the detection feature set and spatial topological relationship features each have their own dimensions, when they are spliced ​​together, the dimension of the joint feature vector for the generated unit grid is equal to the sum of the dimensions of the detection feature set and the spatial topological relationship features. For example, if the dimension of the detection feature set is m and the dimension of the spatial topological relationship features is n, then the dimension of the joint feature vector is m + n. Through this splicing method, the road surface detection data and spatial location information are integrated, providing more comprehensive feature information for establishing the association relationship model.

[0086] Step S134: performing feature selection processing on the joint feature vector, retaining a key feature subset associated with the slippery road condition, the key feature subset being determined by a feature importance evaluation algorithm.

[0087] To improve the accuracy and efficiency of the association model, feature selection is required for the joint feature vector. Each feature in the joint feature vector is evaluated using a feature importance assessment algorithm, such as the one used in the random forest algorithm. The algorithm assigns an importance score to each feature based on its contribution to predicting the slippery road condition. The specific evaluation process is as follows: First, a prediction model is constructed using the random forest algorithm, taking the joint feature vector as input and the slippery road condition as output. Then, during model training, the splitting of each feature in the decision tree and its impact on model performance are recorded. Finally, based on this recorded information, an importance score is calculated for each feature.

[0088] Based on the scores, we select features with higher scores and retain a subset of key features associated with slippery road conditions. This reduces the model's input dimensions, reduces computational complexity, and improves model performance. When selecting a subset of key features, we can set a threshold to select only features with scores above the threshold. We can also select a set number of features based on their ranking, selecting the top-ranked features.

[0089] Step S135: Based on the key feature subset and the corresponding road slippery state reference sample data, construct the input layer, hidden layer and output layer structure of the association relationship model. The number of neurons in the input layer is consistent with the dimension of the key feature subset, and the number of neurons in the output layer corresponds to the dimension of the road slippery state parameter.

[0090] Based on the selected key feature subset and the corresponding reference sample data for slippery road conditions, the structure of the association model is constructed. This association model uses an artificial neural network architecture, consisting of an input layer, a hidden layer, and an output layer. The number of neurons in the input layer matches the dimensionality of the key feature subset, ensuring accurate input of the key feature subset information into the model. For example, if the dimension of the key feature subset is k, then the input layer will have k neurons.

[0091] Hidden layers can contain multiple neurons, which perform nonlinear transformations and feature extraction on the input information. The number of neurons in the hidden layer can be adjusted based on the specific problem and experimental results. Generally speaking, increasing the number of neurons in the hidden layer improves the model's expressiveness, but also increases model complexity and training time. In this model, we can experiment with different numbers of hidden layer neurons to select the setting that achieves the best performance.

[0092] The number of neurons in the output layer corresponds to the dimensionality of the road slipperiness parameter. This parameter can be represented by a vector, whose dimensionality depends on the aspect of the slippery road condition that needs to be described. For example, if the only requirement is to predict whether the road is slippery (yes or no), the output layer can have only one neuron. If the requirement is to predict multiple parameters, such as the road surface moisture level and friction coefficient, the number of neurons in the output layer equals the number of parameters.

[0093] Step S136: Parameter training is performed on the constructed association model using the key feature subset and the reference sample data of the slippery road condition, and the connection weight parameters between the layers of the model are adjusted until the output error of the model meets the preset convergence condition.

[0094] The constructed association model is trained using a subset of key features and reference sample data for slippery road conditions. Using the key feature subset as input and the reference sample data for slippery road conditions as the target output, the backpropagation algorithm continuously adjusts the connection weight parameters between each layer of the model. During training, the input data is first fed into the model's input layer. After undergoing nonlinear transformations in the hidden layer, the model's predicted output is ultimately obtained at the output layer. The error between the model's output and the target output is then calculated. A commonly used error calculation method is the mean squared error (MSE), which is the average of the sum of the squares of the differences between the predicted and target outputs.

[0095] Based on the error, the backpropagation algorithm is used to update the weight parameters. Using the chain rule, the backpropagation algorithm propagates the error from the output layer back to the input layer, calculating the gradient of each weight parameter with respect to the error. The weight parameters are then adjusted based on the magnitude and direction of the gradient to gradually reduce the error. This process is repeated until the model's output error meets the preset convergence criteria—that is, the error reaches a small threshold—at which point the model is considered trained. During training, a maximum number of training iterations can be set to prevent the model from entering an infinite loop. Furthermore, a validation set is used to monitor model performance to prevent overfitting.

[0096] Step S140: Calculate the road surface slippery state parameter of each unit grid using the association relationship model.

[0097] Step S141: Obtain the current road surface detection data and the current spatial position boundary of the unit grid to be calculated.

[0098] To accurately calculate the road surface slipperiness parameters for a specific cell, the current road surface measurement data and its current spatial boundaries must be obtained. Monitoring equipment installed within the cell collects real-time data on road surface humidity, friction coefficient, temperature, and other parameters. These devices collect data at set intervals and transmit it to a data collection center. Simultaneously, the current spatial boundary information for the cell is obtained from a geographic information system (GIS), including the coordinates of the cell's four vertices.

[0099] Step S142: performing the same feature extraction process on the current road surface detection data as that in establishing the association relationship model, and obtaining the current detection feature set of the unit grid to be calculated.

[0100] The current road surface detection data for the unit grid to be calculated is subjected to feature extraction processing, using the same process as when establishing the association model. Following the method described in step S131, data type identification is first performed to distinguish between numerical detection data and categorical detection data. Statistical feature parameters are then calculated for the numerical detection data, and category coding conversion is performed on the categorical detection data. Feature dimension alignment and normalization are then performed, and finally, the current detection feature set for the unit grid to be calculated is obtained by merging.

[0101] Step S143: extracting the current spatial topology relationship features in the current spatial position boundary. The current spatial topology relationship features are extracted in the same manner as the spatial topology relationship features extracted when establishing the association relationship model.

[0102] Extract the current spatial topological relationship features from the current spatial position boundary of the unit grid to be calculated. According to the method described in step S132, first extract the vertex coordinate set of the spatial position boundary, calculate the geometric center coordinates, determine the set of adjacent unit grids, calculate the spatial distance parameters and direction orientation parameters, and finally generate the current spatial topological relationship feature vector.

[0103] Step S144: performing feature splicing processing on the current detection feature set and the current spatial topological relationship feature to generate a current joint feature vector of the unit grid to be calculated.

[0104] Perform feature concatenation on the current detection feature set of the unit grid to be calculated and the current spatial topological relationship feature. Similar to step S133, the two feature sets are concatenated to generate a current joint feature vector for the unit grid to be calculated, whose dimension is equal to the sum of the dimensions of the current detection feature set and the current spatial topological relationship feature.

[0105] Step S145: performing the same feature selection process on the current joint feature vector as that in establishing the association relationship model, and retaining the current key feature subset corresponding to the key feature subset.

[0106] The current joint feature vector of the computational unit grid is subjected to feature selection in the same manner as when establishing the association model. Using the same feature importance evaluation algorithm, the features in the current joint feature vector are evaluated, and the current key feature subset corresponding to the key feature subset is retained. This ensures that the features input to the association model are consistent with the features used during training, improving the accuracy of the model calculation.

[0107] Step S146: Input the current key feature subset into the association relationship model, and obtain the initial road surface slippery state parameters of the unit grid to be calculated through the forward propagation calculation of the model.

[0108] The current subset of key features for the cell grid to be calculated is input into the trained association model. The model follows a forward propagation process, starting from the input layer and undergoing nonlinear transformations in the hidden layers, ultimately obtaining the initial road slippery state parameters for the cell grid to be calculated at the output layer. During this forward propagation process, the input features are calculated and transformed based on the trained connection weight parameters, and the predicted results are output.

[0109] Step S147: performing neighborhood consistency calibration on the initial road surface slippery state parameter, referring to the road surface slippery state parameters of adjacent unit grids of the unit grid to be calculated, adjusting the value of the initial road surface slippery state parameter, and obtaining the final road surface slippery state parameter.

[0110] Step S1471: Determine a set of adjacent unit grids of the unit grid to be calculated, where the set of adjacent unit grids is consistent with the set of adjacent unit grids determined when extracting spatial topological relationship features.

[0111] To perform neighborhood consistency calibration, we first need to determine the set of neighboring cell grids of the cell grid to be calculated. This set of neighboring cell grids is consistent with the set of neighboring cell grids determined when extracting spatial topological relationship features, ensuring data consistency and coherence.

[0112] Step S1472: Obtain the calculated road surface slippery state parameter of each adjacent unit grid in the adjacent unit grid set.

[0113] From the calculated results, the road slippery state parameters of each adjacent unit grid in the adjacent unit grid set are obtained. These parameters are calculated using the same calculation process and the association model.

[0114] Step S1473: Calculate the parameter difference between the initial road surface slippery state parameter of the unit grid to be calculated and the calculated road surface slippery state parameter of each adjacent unit grid.

[0115] The initial road slippery state parameters of the cell to be calculated are compared with the calculated road slippery state parameters of each adjacent cell, and the parameter difference between them is calculated. This difference reflects the difference in road slippery state between the cell to be calculated and the adjacent cell.

[0116] Step S1474: weighting the parameter difference according to the spatial distance parameter of the adjacent unit grids. The smaller the spatial distance parameter, the greater the parameter difference weight corresponding to the adjacent unit grid.

[0117] The parameter differences are weighted based on the spatial distance parameter of adjacent cell grids. The smaller the spatial distance parameter, the closer the spatial position of the adjacent cell grid is to the cell grid to be calculated, and the stronger the correlation between the slippery road conditions, so the corresponding parameter difference weight is larger. A weight function can be used to calculate the weight of each adjacent cell grid, for example, weight = 1 / (spatial distance parameter + ε), where ε is a small positive number to prevent the denominator from being zero. Then, the parameter difference is multiplied by the corresponding weight to obtain the weighted parameter difference.

[0118] Step S1475: Calculate the average value of the weighted parameter differences as the calibration correction value.

[0119] The weighted parameter differences are summed and then divided by the number of adjacent grid cells to obtain the average of the weighted parameter differences, which is used as the calibration correction. This calibration correction reflects the overall difference between the grid cell to be calculated and its adjacent grid cells.

[0120] Step S1476: superimpose the calibration correction amount and the initial road wet slippery state parameter of the unit grid to be calculated to obtain the calibrated road wet slippery state parameter.

[0121] The calibration correction amount is superimposed with the initial road wet slippery state parameter of the unit grid to be calculated. In the above manner, the initial road wet slippery state parameter is adjusted to be more consistent with the neighborhood.

[0122] Step S1477: perform rationality checking on the calibrated road wet slippery state parameter, and determine whether the calibrated parameter is within the preset reasonable value range. If it exceeds the reasonable value range, replace it with a preset boundary value to obtain the final road wet slippery state parameter.

[0123] The calibrated road wet slippery state parameter is checked for rationality. A reasonable value range is preset to determine whether the calibrated parameter is within the range. For example, the reasonable value range of road wetness may be [0, 100%], and the reasonable value range of friction coefficient may be [0, 1]. If the calibrated parameter exceeds the reasonable value range, the parameter may be abnormal, and a preset boundary value is used to replace it to ensure that the final road wet slippery state parameter is reasonable and reliable.

[0124] Step S150: generate the road wet slippery state analysis result of the whole control road section according to the road wet slippery state parameters and spatial position boundaries of each unit grid.

[0125] Step S151: collect the road wet slippery state parameters and corresponding spatial position boundaries of all unit grids in the control road section.

[0126] Through the data management system, the road wet slippery state parameters and corresponding spatial position boundary information of all unit grids in the control road section are collected. The above information is stored in the database for subsequent processing and analysis. During the collection process, the integrity and accuracy of the data are ensured, and the data are checked and cleaned to remove invalid or incorrect data. At the same time, the data are sorted and classified according to the identification information of the unit grid, facilitating subsequent query and use.

[0127] Step S152: perform parameter classification processing on the road wet slippery state parameters of each unit grid, and divide the road wet slippery state parameters with similar numerical ranges into the same state category.

[0128] The collected road surface wetness state parameters are classified. A number of numerical range intervals are set, and road surface wetness state parameters with similar numerical ranges are classified into the same state category. For example, road surface wetness state parameters can be classified into low wetness, medium wetness, high wetness, and other different state categories. The specific classification criteria can be determined according to actual conditions and experience.

[0129] Step S153: Based on the spatial position boundaries of each unit grid, a spatial grid map of the managed road section is constructed, which contains the spatial position distribution information of all unit grids.

[0130] A geographic information system (GIS) software is used to construct a spatial grid map of the managed road section based on the spatial position boundary information of each unit grid. The software will draw the graph of each unit grid according to the vertex coordinates of the unit grid, and combine them together to form a spatial grid map containing the spatial position distribution information of all unit grids. When constructing the map, ensure the accuracy and readability of the map, set appropriate scale and symbols, so that the map can effectively show the distribution of unit grids.

[0131] Step S154: Label the state category of each unit grid on the spatial grid map to generate a road surface wetness state spatial distribution map layer.

[0132] On the constructed spatial grid map, label according to the state category of each unit grid. Different colors or symbols are used to represent different state categories to generate a road surface wetness state spatial distribution map layer. For example, low wetness state can be represented by green, medium wetness state can be represented by yellow, and high wetness state can be represented by red. Through the road surface wetness state spatial distribution map layer, the spatial distribution of road surface wetness state in the managed road section can be observed intuitively.

[0133] Step S155: Analyze the spatial aggregation characteristics of unit grids with the same state category in the road surface wetness state spatial distribution map layer, identify the continuously distributed state category regions, and the spatial aggregation characteristics include the geometric shape parameters and area proportion parameters of the regions.

[0134] Step S1551: Traverse all unit grids in the road surface wetness state spatial distribution map layer, group the unit grids according to the state category to obtain a subset of unit grids for each state category.

[0135] A traversal algorithm is written to traverse all unit grids in the road surface wetness state spatial distribution map layer. According to the state category of the unit grid, they are grouped and processed to obtain a subset of unit grids for each state category. In the traversal process, the state category is used as the grouping basis, and unit grids with the same state category are grouped together, which facilitates subsequent analysis.

[0136] Step S1552: Spatial connectivity analysis is performed on each unit grid subset of the state category to determine whether the unit grids in the unit grid subset have spatial adjacency relationship, and the unit grids with continuous adjacency relationship are grouped into a connected unit grid cluster.

[0137] Spatial connectivity analysis is performed on each unit grid subset of the state category. Using a spatial analysis algorithm, it is determined whether the unit grids in the unit grid subset have spatial adjacency relationship. If the boundaries of two unit grids are adjacent, they are considered to have adjacency relationship. Through a depth-first search or breadth-first search algorithm, the unit grids with continuous adjacency relationship are grouped into a connected unit grid cluster.

[0138] Step S1553: Boundary extraction processing is performed on each connected unit grid cluster to determine the peripheral boundary vertex coordinate set of the connected unit grid cluster.

[0139] Boundary extraction processing is performed on each connected unit grid cluster. Using a boundary extraction algorithm, the peripheral boundary vertex coordinate set of the connected unit grid cluster is determined, which can accurately describe the shape and position of the connected unit grid cluster.

[0140] Step S1554: Based on the peripheral boundary vertex coordinate set, the geometric shape parameters of the connected unit grid cluster are calculated, including the length and width of the minimum circumscribed rectangle, the boundary perimeter, and the shape complexity index.

[0141] Based on the peripheral boundary vertex coordinate set of the connected unit grid cluster, its geometric shape parameters are calculated. Using geometric calculation method, the length and width of the minimum circumscribed rectangle are calculated, which can reflect the approximate size of the connected unit grid cluster. The boundary perimeter is calculated to understand the length of its boundary. At the same time, the shape complexity index is calculated to describe the shape complexity of the connected unit grid cluster. The shape complexity index can be calculated by factors such as the tortuosity of the boundary, the ratio of area to perimeter, etc.

[0142] Step S1555: The number of unit grids contained in the connected unit grid cluster is calculated, and the total area parameter of the connected unit grid cluster is calculated in combination with the area parameter of each unit grid.

[0143] The number of unit grids contained in each connected unit grid cluster is counted, and the total area parameter of the connected unit grid cluster is calculated in combination with the known area parameter of each unit grid. In this way, the area size of each continuously distributed state category region can be accurately understood.

[0144] Step S1556: The total area parameter is compared with the total area parameter of the controlled road section to obtain the area proportion parameter of the connected unit grid cluster.

[0145] The total area parameter of the connected unit grid cluster is compared with the total area parameter of the control road section to obtain an area ratio parameter of the connected unit grid cluster. The area ratio parameter reflects the proportion of the continuously distributed state category region in the control road section.

[0146] Step S1557: The geometric shape parameter and the area ratio parameter are taken as spatial aggregation features, and the connected unit grid cluster containing the spatial aggregation features is marked as a continuously distributed state category region.

[0147] The calculated geometric shape parameter and area ratio parameter are taken as spatial aggregation features, and the connected unit grid cluster is marked. The connected unit grid cluster containing the spatial aggregation features is determined as a continuously distributed state category region for subsequent analysis and display.

[0148] Step S156: Calculate the spatial proportion and distribution density of each state category region in the control road section to generate state category statistical features.

[0149] According to the area ratio parameter of the connected unit grid cluster calculated in the foregoing, the spatial proportion of each state category region in the control road section is calculated. The area ratio parameters of each state category region are added to obtain the total spatial proportion of the state category region in the control road section. At the same time, the number of unit grids in each state category region is counted, and the distribution density is calculated in combination with the area of the region. The distribution density can be expressed as the number of unit grids per unit area. The spatial proportion and the distribution density are taken as state category statistical features, which can quantitatively describe the distribution of different state category regions in the control road section.

[0150] Step S157: The road surface wet slip state spatial distribution layer, the state category statistical features, and the road surface wet slip state parameters of each unit grid are integrated into the road surface wet slip state analysis result of the whole control road section. The road surface wet slip state analysis result contains spatial visualization information and statistical description information.

[0151] Finally, the road surface wet slip state spatial distribution layer, the state category statistical features, and the road surface wet slip state parameters of each unit grid are integrated to form the road surface wet slip state analysis result of the whole control road section. The road surface wet slip state analysis result contains spatial visualization information, such as the road surface wet slip state spatial distribution layer, which can intuitively show the spatial distribution of the road surface wet slip state; and statistical description information, such as the state category statistical features, which can accurately describe the distribution characteristics of the road surface wet slip state from the data level.

[0152] Figure 2A schematic diagram illustrates exemplary hardware and software components of a unit-grid-based road slippery condition analysis system 100, provided in some embodiments of the present application, that can implement the concepts of the present application. For example, a processor 120 can be used in the unit-grid-based road slippery condition analysis system 100 to perform the functions described in the present application.

[0153] The unit grid-based road slippery condition analysis system 100 can be a general-purpose server or a special-purpose server, both of which can be used to implement the unit grid-based road slippery condition analysis method of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0154] For example, the road slippery state analysis system 100 based on unit grid division can include a network port 110 connected to the network, one or more processors 120 for executing program instructions, a communication bus 130, and storage media 140 in different forms, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the road slippery state analysis system 100 based on unit grid division can also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The road slippery state analysis system 100 based on unit grid division also includes an I / O interface 150 between the computer and other input and output devices.

[0155] For ease of explanation, only one processor is described in the road surface slippery state analysis system 100 based on unit grid division. However, it should be noted that the road surface slippery state analysis system 100 based on unit grid division in this application can also include multiple processors, so the steps performed by one processor described in this application can also be performed jointly or individually by multiple processors. For example, if the processor of the road surface slippery state analysis system 100 based on unit grid division executes step A and step B, it should be understood that step A and step B can also be performed jointly by two different processors or individually in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor execute steps A and B together.

[0156] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the above-mentioned road slippery state analysis method based on unit grid division is implemented.

[0157] It should be noted that the foregoing description of embodiments of the application has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the application to the precise form described, and many modifications, variations, and alternatives are possible.

Claims

1. A method for analyzing road slippery conditions based on unit grid division, characterized in that: The method comprises: Divide the controlled road section into multiple unit grids, each unit grid has unique grid identification information and spatial location boundaries; Acquire road surface detection data corresponding to each unit grid, wherein the road surface detection data includes a set of detection parameters reflecting the surface state of the road surface; Based on the road surface detection data and spatial position boundaries of each unit grid, a correlation model between the unit grid and the road surface slippery state is established; Calculating the road surface slippery state parameter of each unit grid using the association relationship model; Generate the overall road slippery state analysis results of the controlled road section based on the road slippery state parameters and spatial position boundaries of each unit grid; The method of establishing a correlation model between the unit grid and the road surface slippery state based on the road surface detection data and spatial position boundaries of each unit grid includes: Performing feature extraction processing on the road surface detection data of each unit grid to obtain a detection feature set for each unit grid, wherein the detection feature set includes numerical features and categorical features in the road surface detection data; Extracting spatial topological relationship features in the spatial position boundary of each unit grid, wherein the spatial topological relationship features include spatial distance parameters and direction orientation parameters of adjacent unit grids; Perform feature splicing processing on the detection feature set of each unit grid and the spatial topological relationship feature to generate a joint feature vector of the unit grid, where the dimension of the joint feature vector is equal to the sum of the dimensions of the detection feature set and the spatial topological relationship feature; performing feature selection processing on the joint feature vector to retain a subset of key features associated with the slippery road condition, wherein the subset of key features is determined by a feature importance evaluation algorithm; Based on the key feature subset and the corresponding road slippery state reference sample data, constructing an input layer, hidden layer, and output layer structure of the association relationship model, wherein the number of neurons in the input layer is consistent with the dimension of the key feature subset, and the number of neurons in the output layer corresponds to the dimension of the road slippery state parameter; The constructed association model is trained with parameters using the key feature subset and the road slippery state reference sample data, and the connection weight parameters between the layers of the model are adjusted until the output error of the model meets the preset convergence condition.

2. The method for analyzing road slippery conditions based on unit grid division according to claim 1, characterized in that: The controlled road section is divided into a plurality of unit grids, including: Conducting section feature analysis on the controlled road sections to extract the linear direction features, pavement structure type features, and environmental features along the controlled road sections; Determine the main direction reference axis of the grid division based on the linear trend characteristics, so that the length direction of the unit grid is consistent with the linear trend characteristics of the controlled road section; Determine the size division basis of the unit grid according to the characteristics of the pavement structure type and the environmental characteristics along the road, wherein the size division basis includes a pavement structure change frequency parameter and an environmental influencing factor distribution density parameter; According to the main direction reference axis and the size division basis, the controlled road section is subjected to spatial grid cutting processing to generate a preliminary unit grid set, each preliminary unit grid having an initial spatial position boundary; Performing boundary smoothing processing on the preliminary unit grid set to eliminate overlapping areas and gap areas between adjacent unit grid boundaries to obtain a final plurality of unit grids; Unique grid identification information is allocated to each final unit grid, and the grid identification information includes a road segment number, a grid sequence number, and a coordinate range descriptor of a spatial position boundary.

3. The method for analyzing road slippery conditions based on unit grid division according to claim 1, characterized in that: The feature extraction process is performed on the road surface detection data of each unit grid to obtain a detection feature set of each unit grid, including: Perform data type recognition processing on the road surface detection data of the unit grid to distinguish between numerical detection data and categorical detection data; Performing feature extraction on the numerical detection data and calculating statistical feature parameters of the numerical detection data, wherein the statistical feature parameters include a central tendency parameter, a dispersion degree parameter, and a distribution morphology parameter of the data sequence; Performing feature extraction on the categorized detection data, and converting the categorized detection data into a numerical category feature vector using a category coding method, wherein the category coding method is determined based on the number of categories of the categorized detection data and the relationship between the categories; Performing feature dimension alignment processing on the statistical feature parameters and the category feature vectors so that the numerical features and the categorical features have the same number of samples and timestamp marks; The statistical feature parameters and category feature vectors after dimension alignment are subjected to feature normalization processing, and the normalized statistical feature parameters and category feature vectors are subjected to feature merging processing to generate a detection feature set including numerical features and categorical features, wherein each feature in the detection feature set has a unique feature name and feature value range description.

4. The method for analyzing road slippery conditions based on unit grid division according to claim 1, characterized in that: The extracting of spatial topological relationship features in the spatial position boundary of each unit grid includes: Perform boundary coordinate extraction processing on the spatial position boundary of each unit grid to obtain a vertex coordinate set of the spatial position boundary; Calculating the geometric center coordinates of each unit grid based on the vertex coordinate set, wherein the geometric center coordinates are obtained by calculating the arithmetic average of the vertex coordinates; Traversing all unit grids of the controlled road section, determining a set of adjacent unit grids for each unit grid, wherein the set of adjacent unit grids includes other unit grids that have common edges or common vertices with the spatial position boundary of the current unit grid; Calculating the straight-line distance between the geometric center coordinates of the current unit grid and the geometric center coordinates of each adjacent unit grid in the set of adjacent unit grids, and using the straight-line distance as a spatial distance parameter of the adjacent unit grids; Calculate the direction and azimuth of the adjacent unit grid relative to the current unit grid based on the geometric center coordinates of the current unit grid and the geometric center coordinates of the adjacent unit grid. The direction and azimuth are calculated with the geometric center of the current unit grid as the origin and the linear direction of the controlled section as the reference axis. The spatial distance parameter and the direction azimuth are dimensionally expanded according to the number of adjacent unit grids to generate a spatial topological relationship feature vector with a fixed dimension, wherein each element in the spatial topological relationship feature vector corresponds to a combination of the spatial distance parameter and the direction azimuth parameter of an adjacent unit grid.

5. The method for analyzing road slippery conditions based on unit grid division according to claim 1, characterized in that: The calculating of the road surface slippery state parameter of each unit grid by using the association relationship model includes: Obtain the current road surface detection data and current spatial position boundary of the unit grid to be calculated; Performing the same feature extraction process as that used in establishing the association model on the current road surface detection data to obtain a current detection feature set of the unit grid to be calculated; Extracting current spatial topological relationship features in the current spatial position boundary, wherein the current spatial topological relationship features are extracted in the same manner as the spatial topological relationship features extracted when establishing the association relationship model; Perform feature splicing processing on the current detection feature set and the current spatial topological relationship feature to generate a current joint feature vector of the unit grid to be calculated; Performing the same feature selection process on the current joint feature vector as that used in establishing the association relationship model, and retaining the current key feature subset corresponding to the key feature subset; Inputting the current key feature subset into the association relationship model, and obtaining the initial road surface slippery state parameters of the unit grid to be calculated through forward propagation calculation of the model; The initial road surface slippery state parameter is subjected to neighborhood consistency calibration processing, and the value of the initial road surface slippery state parameter is adjusted with reference to the road surface slippery state parameter of the adjacent unit grid of the unit grid to be calculated to obtain the final road surface slippery state parameter.

6. The method for analyzing road slippery conditions based on unit grid division according to claim 5, characterized in that: The neighborhood consistency calibration process is performed on the initial road surface slippery state parameter, and the road surface slippery state parameter of the adjacent unit grids of the unit grid to be calculated is referred to, and the value of the initial road surface slippery state parameter is adjusted to obtain the final road surface slippery state parameter, including: Determine a set of adjacent unit grids of the unit grid to be calculated, where the set of adjacent unit grids is consistent with the set of adjacent unit grids determined when extracting spatial topological relationship features; Obtaining a calculated road surface slippery state parameter of each adjacent unit grid in the set of adjacent unit grids; Calculating the parameter difference between the initial road surface slippery state parameter of the unit grid to be calculated and the calculated road surface slippery state parameter of each adjacent unit grid; The parameter difference is weighted according to the spatial distance parameter of the adjacent unit grids, and the adjacent unit grids with smaller spatial distance parameters have larger parameter difference weights; Calculate the average of the weighted parameter differences as the calibration correction; Superimposing the calibration correction amount with the initial road surface slippery state parameter of the unit grid to be calculated to obtain a calibrated road surface slippery state parameter; The calibrated road slippery state parameter is rationally checked to determine whether the calibrated parameter is within a preset reasonable value range. If it exceeds the reasonable value range, the preset boundary value is used to replace it to obtain the final road slippery state parameter.

7. The method for analyzing road slippery conditions based on unit grid division according to claim 1, characterized in that: The method of generating an analysis result of the overall road slippery state of the controlled road section based on the road slippery state parameters and spatial position boundaries of each unit grid includes: Collect the road surface slippery state parameters and corresponding spatial position boundaries of all unit grids in the controlled road section; The parameters of the road slippery state of each unit grid are classified into parameters with similar numerical ranges and classified into the same state category; Based on the spatial position boundaries of each unit grid, a spatial grid map of the controlled road section is constructed, wherein the spatial grid map includes the spatial position distribution information of all unit grids; Marking the state category of each unit grid on the spatial grid map to generate a road slippery state spatial distribution layer; Analyzing spatial aggregation characteristics of unit grids of the same state category in the slippery road state spatial distribution layer to identify continuously distributed state category areas, wherein the spatial aggregation characteristics include geometric shape parameters and area ratio parameters of the areas; Calculate the spatial proportion and distribution density of each status category area in the controlled road section and generate the statistical characteristics of the status category; The road slippery state spatial distribution layer, state category statistical characteristics and road slippery state parameters of each unit grid are integrated into the road slippery state analysis result of the entire controlled road section, and the road slippery state analysis result includes spatial visualization information and statistical description information.

8. The method for analyzing road slippery conditions based on unit grid division according to claim 7, characterized in that: The analyzing the spatial aggregation characteristics of the unit grids of the same state category in the road slippery state spatial distribution layer to identify the continuously distributed state category areas includes: Traverse all the unit grids in the road slippery state spatial distribution layer, group the unit grids according to the state category, and obtain the unit grid subset for each state category; Perform spatial connectivity analysis on the cell grid subset of each state category to determine whether the cell grids in the cell grid subset have spatial adjacent relationships, and group the cell grids with continuous adjacent relationships into connected cell grid clusters; Performing boundary extraction processing on each connected unit mesh cluster to determine the outer boundary vertex coordinate set of the connected unit mesh cluster; Calculating geometric shape parameters of the connected unit grid cluster based on the outer boundary vertex coordinate set, the geometric shape parameters including the length and width of the minimum circumscribed rectangle, the boundary perimeter and the shape complexity index; Calculate the number of unit grids contained in the connected unit grid cluster, and calculate the total area parameter of the connected unit grid cluster by combining the area parameter of each unit grid; Calculate the ratio of the total area parameter to the total area parameter of the controlled road section to obtain the area ratio parameter of the connected unit grid cluster; The geometric shape parameters and area ratio parameters are used as spatial aggregation features, and the connected unit grid clusters containing the spatial aggregation features are marked as continuously distributed state category areas.

9. A road slippery state analysis system based on unit grid division, characterized in that: It includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the road slippery state analysis method based on unit grid division as described in any one of claims 1 to 8.

Images