A method for detecting the wet-slippery state of a tunnel pavement based on a mobile detection device

Abstract

The present application relates to a kind of detection methods of tunnel pavement wet slippery state based on mobile detection equipment, mobile detection equipment is set above tunnel pavement and can be controlled along the length direction of tunnel pavement reciprocating movement, different positions of pavement wet slippery coefficient value on tunnel pavement can be detected, the method comprises: tunnel pavement is divided into several unit grids;Control mobile detection equipment moves above tunnel pavement, and the pavement wet slippery state data corresponding to each unit grid is collected;The road surface wet slippery state data of each unit grid is carried out cluster analysis;According to the result of the cluster analysis, determine the pavement wet slippery coefficient of each unit grid.By the detection method of the present application, the pavement wet slippery state in tunnel can be detected and evaluated in time, reliable, and then for objectively judging the pavement state in tunnel and scientifically and effectively controlling traffic flow in tunnel provides decision-making basis, effectively guarantee the traffic safety and efficiency of vehicle.

Description

[0001] This application is a divisional application of the invention patent application with application number 202210754784.1, application date June 30, 2022, entitled "An evaluation method for the slippery state of tunnel pavement based on mobile detection equipment". Technical Field

[0002] This invention relates to a method for detecting the slippery condition of tunnel pavement, and more particularly to a method for detecting the slippery condition of tunnel pavement based on a mobile detection device. Background Technology

[0003] Compared to open roads, highway tunnels differ significantly in road surface conditions, lighting environment, and infrastructure. While highway tunnels are relatively enclosed environments, and the internal road surface is less affected by external weather changes, the tunnel entrances and exits are directly connected to the outside world and are susceptible to weather conditions. Especially in severe weather conditions such as freezing, rain, and snow, the road surface becomes highly slippery, resulting in a significant difference in slipperiness between the tunnel entrances / exits and the interior. Due to the change in lighting conditions, drivers are prone to experiencing "black hole effects" and "white hole effects" when entering and exiting tunnels, requiring time to adapt to the tunnel lighting environment. Consequently, drivers at tunnel entrances and exits have difficulty observing the road ahead. Furthermore, the dim lighting inside tunnels reduces visibility, further limiting the driver's ability to anticipate road conditions. In addition, as highway tunnels age, factors such as the surrounding geographical environment, weather conditions, structural design, and maintenance frequency contribute to water leakage, becoming one of the main problems affecting operating tunnels. Water leakage leading to road surface overflow and slipperiness is a major factor affecting vehicle safety and causing traffic accidents.

[0004] Extensive historical accident data and research have revealed that due to the harsh driving environment and relatively enclosed space of highway tunnels, drivers are not only forced to shorten their visibility distance under the influence of road alignment and lighting, but their psychological state is also prone to significant fluctuations. Especially when the road surface is slippery and the coefficient of friction is reduced, accidents such as rear-end collisions and collisions with tunnel walls are highly likely to occur. Moreover, once a traffic accident occurs in a tunnel, not only is rescue difficult and traffic management challenging, but the reduced visibility also makes secondary accidents highly likely, causing more severe congestion and greater losses.

[0005] Considering the aforementioned characteristics of the tunnel environment, limiting vehicle speed may help reduce the risk of accidents to some extent, but this will obviously lead to a decrease in the overall traffic capacity of the tunnel and sacrifice its traffic efficiency.

[0006] In order to balance the safety and efficiency of highway tunnels, it is necessary to conduct timely and reliable detection and evaluation of the slippery condition of the road surface inside the tunnel, so as to provide a basis for decision-making on objectively judging the road surface condition inside the tunnel and scientifically and effectively controlling traffic flow inside the tunnel.

[0007] There are generally two types of devices used to detect the slipperiness of road surfaces: one is a contact road surface condition detector, which is usually a vehicle-mounted device that measures the friction coefficient or slipperiness of the road surface while the vehicle is moving; the other is a non-contact road surface condition detector, which usually uses methods such as optical polarization, image recognition, and infrared spectroscopy, and is usually fixed above the road to detect specific locations on the road surface.

[0008] Fixed roadside inspection equipment can only detect a specific point or a very small area on the road surface. In practical applications, the detection results of a single point or area are usually used to represent the road surface slipperiness of a larger section of road. Obviously, this point-to-area detection result cannot accurately reflect the actual slipperiness of the tunnel road surface, thus leading to problems such as untimely detection of tunnel road surface or infrastructure defects, and over- or under-management of traffic flow. Therefore, manual inspections are still necessary to determine the precise location of road surface problems.

[0009] Currently, while vehicle-mounted mobile inspection equipment can continuously measure multiple points, the spacing between these points is relatively large, typically measuring 4-10 points within a 100-meter interval. This fails to accurately and comprehensively reflect the slipperiness of the tunnel surface. Furthermore, the vehicles carrying the sensors require manual driving, inevitably affecting other vehicles. Moreover, vehicle-mounted sensors are limited by road conditions and traffic flow, preventing continuous, cyclical measurements. Therefore, not only are the uses of vehicle-mounted sensors subject to numerous limitations, but the detected results still cannot accurately reflect the slipperiness of the tunnel surface and its changes, nor can they provide an accurate and real-time assessment of the slipperiness of the tunnel surface.

[0010] In summary, the existing methods for detecting and evaluating the slippery condition of road surfaces are not very precise and cannot reflect the slippery condition of tunnel surfaces in a timely and accurate manner. They also cannot reliably detect the slippery condition of tunnel surfaces simultaneously in terms of time and space.

[0011] Therefore, it is necessary to design an effective system and method for detecting the slippery condition of road surfaces inside tunnels, so as to reflect the slippery condition of the road surface in a timely and accurate manner, thereby providing a basis for decision-making on the scientific management of traffic flow inside tunnels, ensuring tunnel traffic efficiency, and effectively protecting vehicle traffic safety. Summary of the Invention

[0012] The purpose of this invention is to provide a method for detecting the slippery condition of tunnel pavement based on a mobile detection device, which can reflect the slippery condition of the pavement in a timely and accurate manner and make a reliable evaluation of the slippery condition of the pavement.

[0013] To achieve the above objectives, the present invention provides a method for detecting the slippery condition of tunnel pavement based on a mobile detection device. The mobile detection device is positioned above the tunnel pavement and can move back and forth in a controlled manner along the length of the tunnel pavement. The mobile detection device can detect the slippery coefficient value of the pavement at different locations on the tunnel pavement. The detection method includes the following steps:

[0014] Step S21: Divide the tunnel surface into several unit grids;

[0015] Step S22: Control the mobile detection device to move above the tunnel surface and collect the road surface wet and slippery state data corresponding to each unit grid. The road surface wet and slippery state data includes a detection dataset consisting of at least the road surface wet and slippery coefficient values ​​of different detection points within the unit grid.

[0016] Step S23: Perform cluster analysis on the road surface wet slip state data for each cell grid;

[0017] Step S24: Based on the results of the cluster analysis, determine the road surface slip coefficient for each cell grid.

[0018] Preferably, the length of the unit grid is 80 to 120 meters, more preferably 100 meters.

[0019] Preferably, the road surface slippery condition data includes the road surface slippery coefficient values ​​of 200 to 800 points within a unit grid, and more preferably, it includes the road surface slippery coefficient values ​​of 400 points within a unit grid.

[0020] Preferably, the detection dataset is X = {X1, X2, ... X}. i ,…X n}, where 1≤i≤n, step S23 further includes the following steps:

[0021] The number of cluster centroids K is randomly selected, where K is an integer, 1 ≤ K ≤ 10, and the randomized cluster centroids are C = {C1, C2, ... C}. j ,…C K}, where 1≤j≤K, C j Let C be the centroid of the j-th cluster, 0 ≤ C j ≤1;

[0022] Step A: Calculate the value of each data sample X in the detection dataset X using formula (1). i To the cluster centroid C of each cluster jEuropean distance,

[0023]

[0024] Compare each data sample X in turn i To each cluster centroid C j The Euclidean distance between each data sample X i Assign to the nearest cluster centroid C j From the clusters, we obtain K clusters {S1, S2, ... S}. j ,…S K}, where S j Let j represent the j-th cluster, which is the data set of the j-th cluster after re-clustering;

[0025] Step B: Recalculate the S values ​​for each cluster using formula (2). j The center of mass,

[0026]

[0027] Among them, C m Let S represent the centroid of the m-th cluster, 1 ≤ m ≤ K; m | represents the number of data samples in the m-th cluster; X d This represents the d-th object in the m-th cluster, where 1 ≤ d ≤ |S|. m |;

[0028] Step C: Repeat steps A and B until the centroids of the K clusters no longer change, and obtain the results of cluster analysis of the road surface slipperiness state data of the cell grid with the current K value.

[0029] Preferably, the detection method further includes a step of optimizing the K value:

[0030] Step 1: For data sample X i Calculate X i The average distance to all other elements in the same cluster is denoted as a. i ;

[0031] Step 2: Select another cluster b and calculate X. i Find the average distance to all elements in b, and iterate through all other clusters to find X. i The average distance to the nearest cluster from all other clusters is denoted as b. i ;

[0032] Step 3: Calculate element X using formula (3) i The profile coefficient,

[0033]

[0034] Step 4: Calculate the silhouette coefficient of all data samples X, and calculate the overall silhouette coefficient corresponding to the current K value through formula (4),

[0035]

[0036] Step 5: Adjust the value of K to other integers between 2 and 10, repeat steps A to C, and calculate the overall silhouette coefficient corresponding to each K value through steps 1 to 4, wherein the maximum overall silhouette coefficient E k' corresponding K=k' is the optimal K value;

[0037] Step 6: Use the clustering result corresponding to the optimal K value as the result of the clustering analysis.

[0038] Preferably, said step S24 further comprises: determining the centroid of the cluster with the largest number of samples as the road surface slip coefficient Sp of the unit grid.

[0039] Preferably, the detection method further comprises: setting three thresholds m1, m2, and m3 for the road surface slip coefficient of the unit grid, wherein 0<m1<m2<m3<1, so that the road surface slip state is divided into four evaluation levels, namely:

[0040] Level I: Sp∈(m3,1);

[0041] Level II: Sp∈(m2,m3);

[0042] Level III: Sp∈(m1,m2);

[0043] Level IV: Sp∈(0,m1).

[0044] Through the detection method for slip state of tunnel road surface based on mobile detection equipment of the present invention, timely and reliable detection and evaluation of the slip state of road surface in the tunnel can be realized, which further provides decision-making basis for objectively judging the road surface state in the tunnel and scientifically and effectively managing and controlling the traffic flow in the tunnel, which effectively ensures the traffic safety of vehicles and also advantageously ensures the tunnel traffic efficiency. Description of Drawings

[0045] Figure 1 is a schematic diagram of the arrangement mode and data acquisition mode of the mobile detection device according to the preferred embodiment of the present invention;

[0046] Figure 2 shows a flow chart of the detection method for slip state of tunnel road surface based on mobile detection equipment according to the preferred embodiment of the present invention. Detailed Description

[0047] In this invention, the term "slippery road surface" refers to a road surface that is wet, waterlogged, icy, or covered in snow due to weather or geographical conditions such as ice, rain, snow, or fog. Under these conditions, the road surface friction coefficient decreases, leading to difficulty in braking, increased braking distance, and reduced vehicle stability, which can easily cause serious traffic accidents. The road surface slippage coefficient is typically used to characterize the slippery road surface condition; it is a value between 0 and 1, with a smaller value indicating a more slippery surface.

[0048] The road surface wet skid coefficient can be measured using non-contact methods such as optical polarization and infrared spectroscopy. Based on the differences in optical properties of different areas of the road surface, the state of the road surface can be distinguished, such as dry, water accumulation, snow accumulation, and ice formation. This leads to an evaluation index representing the degree of road surface wet skid, which is used to characterize the anti-skid performance of the road surface and has the same significance as the coefficient of friction.

[0049] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Those skilled in the art should recognize that the present invention is not limited to the specific embodiments described herein, but can be modified and varied according to the concept of the present invention.

[0050] Figure 1 A schematic diagram illustrating the deployment and data acquisition method of a tunnel movement detection device according to a preferred embodiment of the present invention is shown. Figure 1 In the embodiment shown, a track 11 is installed on the top of the tunnel. The length of the track 11 is approximately the same as the length of the tunnel. A mobile detection device 12 (including a non-contact road condition detector) for detecting the slippery condition of the road surface is installed on the track 11. It can move back and forth along the track 11 at a certain speed and detect the slippery coefficient value of the road surface at different locations at certain intervals while moving.

[0051] Since both track 11 and mobile detection equipment 12 are located in the space above the tunnel, they will not affect the passage of vehicles in the tunnel. Moreover, since mobile detection equipment 12 can move back and forth cyclically throughout the entire length of the tunnel, it can perform multiple tests on the wet and slippery condition of the road surface throughout the entire tunnel at very short time intervals, thereby achieving all-weather, full-coverage, and uninterrupted detection of the road surface condition of the highway tunnel.

[0052] Those skilled in the art will understand that the specific structure and arrangement of the track, the coordination between the track and the mobile detection equipment, and the driving method of the mobile detection equipment can all be achieved through existing technical means, so they will not be elaborated here.

[0053] Furthermore, those skilled in the art will understand that the mobile detection device is not limited to including a specific type of sensor or detection device. In a preferred embodiment of the invention, the tunnel is divided into several segments (i.e., cell grids) with a length L = 100 meters along its length. The mobile detection device continuously detects the road surface within each cell grid at 0.25-meter intervals along the length of the tunnel, thereby measuring 400 points within one cell grid and obtaining 400 road surface slip coefficient values ​​characterizing the slippery state of the road surface at each point. Since the detection device measures based on the optical properties of the road surface, and light travels very quickly, existing non-contact road surface condition detection technologies are already mature and widely used, thus the detection of one cell grid can be completed in a very short time. However, depending on the type of mobile detection device and the subsequent analysis requirements, the number, location, and detection order of detection points within each cell grid can vary. For example, the mobile detection device can scan and detect various points on the road surface perpendicular to its direction of movement while moving along the track, or scan and detect various points on the road surface parallel to its direction of movement, or detect various points on the road surface in a random order. Furthermore, if necessary, multiple tracks and multiple mobile detection devices can be installed above the tunnel; or multiple mobile detection devices can be installed on one track; or multiple detection sensors can be installed in one mobile detection device. In short, the installation method and detection method of the mobile detection devices are not limited to a specific form, as long as they can provide the data information required to satisfy the analysis method of the road surface slipperiness of this invention.

[0054] Mobile detection equipment positioned above tunnels can detect the road surface's slipperiness at short intervals, obtaining slipper coefficient values ​​at many locations. However, in reality, this detailed detection at numerous points cannot directly help drivers, road management units, or personnel determine the impact of slippery road conditions on driving safety. The inventors of this invention have discovered that: on the one hand, a high slipper coefficient value at a single detection point does not necessarily pose a serious threat to driving safety; on the other hand, while averaging slipper coefficient values ​​at multiple locations on the road can avoid biases caused by accidental factors in single-point measurements and reflect the overall slipperiness to some extent, such an average cannot reflect local differences in slipperiness and cannot accurately reflect the impact of slipperiness at different locations on passing vehicles.

[0055] In order to accurately and objectively reflect the slipperiness of the road, facilitate drivers' safe driving, and enable road management personnel to scientifically control traffic flow while taking into account both traffic safety and efficiency, the inventors divided the detection section into multiple unit grids, and then performed comprehensive calculations based on the road surface slip coefficient detection data of multiple points within the unit grid to obtain the road surface slip coefficient of that unit grid.

[0056] Figure 2 A flowchart for detecting the slippery state of a road surface according to a preferred embodiment of the present invention is shown.

[0057] The first step (S21) is to divide the tunnel pavement into unit grids for the wet and slippery condition of the pavement to be tested.

[0058] The purpose of dividing the road surface into unit grids is to discretize the continuous road surface into many small segments, i.e., unit grids. Generally speaking, the size of the unit grid can be determined based on data such as the tunnel's design speed, the actual average speed of vehicles on the road, and the safe stopping sight distance of vehicles, so as to facilitate accurate perception of the road surface's slippery condition and scientific management of traffic flow.

[0059] For example, in a long tunnel exceeding 1000 meters in length, the average vehicle speed is approximately 80 km / h. Under slippery road conditions, within the timeframe for driver reaction and braking, the vehicle will continue to travel approximately 80-120 meters. This demonstrates that slippery road conditions directly impact vehicle performance during braking. Therefore, dividing the road into grid cells at such intervals facilitates a more scientific and effective detection of road slipperiness, meeting both driver reaction and braking needs, as well as the requirements for refined road maintenance and management. Those skilled in the art will understand that while dividing the grid into cells with excessively small lengths may improve the accuracy of slippery road condition detection to some extent, it has no practical significance for traffic flow management and increases the system's detection and computational load. Conversely, dividing the grid into cells with excessively large lengths may result in insufficient accuracy in slippery road condition detection, leading to untimely traffic flow management and inadequate maintenance management.

[0060] In a preferred embodiment of the present invention, a 100-meter-long unit grid is used as the evaluation object for the slippery state of the road surface.

[0061] Considering that the environments of all lanes within the tunnel are basically the same, and the road surface slippage conditions of each lane are also basically the same, arranging the detection points sequentially along the length of a single lane can accurately detect and analyze the road surface slippage conditions. However, those skilled in the art will understand that if it is necessary to obtain road surface slippage condition information for each lane, mobile detection equipment can also be used to detect different lanes separately. That is, different lanes can be divided into their own cell grids, and the road surface slippage coefficient values ​​of the corresponding points in the cell grids can be detected separately.

[0062] The second step (S22) involves collecting data on the wet and slippery condition of the road surface using mobile detection equipment.

[0063] Different locations within each grid cell exhibit varying road surface wetness and slipperiness. By moving the detection device, the road surface wetness and slipperiness coefficient values ​​at these different points can be obtained. Based on the road surface wetness and slipperiness dataset obtained from the detection within each grid cell, the overall road surface wetness and slipperiness coefficient of each grid cell can be comprehensively analyzed.

[0064] In a preferred embodiment of the invention, the mobile detection device continuously detects the tunnel surface along the length of the tunnel at 0.25-meter intervals within each 100-meter-long cell grid, thereby measuring 400 points within one cell grid and obtaining 400 road surface slip coefficient values ​​accordingly. These slip coefficient values ​​can fully reflect the various possible slip conditions of the road surface within the cell grid.

[0065] Those skilled in the art should understand that the number of detection points within each cell grid is not limited to 400. The number of detection points within a cell grid typically depends on the needs of the clustering algorithm, which will be described in detail below. Generally speaking, if the number of points is too small, there will be less data suitable for the clustering algorithm, making it impossible to achieve the ideal clustering effect and potentially affecting the accuracy of detecting slippery road surfaces. On the other hand, if the detection points are too dense, data redundancy may occur, affecting the operating efficiency of the clustering algorithm and thus potentially impacting the overall detection efficiency. Typically, setting 200-800 detection points within each cell grid can balance detection accuracy and efficiency.

[0066] The inventors discovered that the difference in road surface slipperiness within a tunnel is not significant along the width direction. Therefore, sequentially arranging detection points along the length of a single lane can accurately reflect the road surface slipperiness. However, those skilled in the art should understand that the detection points within each grid cell are not limited to the tunnel's length direction but can be arranged along both the tunnel's length and width directions. In some preferred embodiments of the invention, the detection points are evenly distributed within the grid cells, which facilitates a more accurate reflection of the overall road surface slipperiness within the grid cell.

[0067] The road surface slipperiness data obtained through mobile detection equipment may include detection time, detection location, and the road surface slipperiness coefficient value at the detection point. In some preferred embodiments, the detection location can be simplified to the cell grid described at the detection point, such as the cell grid number. Further processing of the road surface slipperiness data belonging to the same cell grid can yield the road surface slipperiness coefficient for that cell grid.

[0068] During the process of mobile detection equipment detecting the slippery condition of the road surface, vehicles passing by at corresponding points will inevitably cause some interference to the detection, which may lead to inaccurate detection results for that point. However, since such detection results only account for a very small proportion of all detection results in the entire cell grid, they will not affect the judgment of the slippery condition of the entire cell grid. Moreover, on the one hand, the clustering algorithm, which will be described in detail below, can easily identify and eliminate the influence of these inaccurate data; on the other hand, since the mobile detection equipment repeatedly detects within the tunnel, the data affected by passing vehicles will also be corrected during the repeated detection process.

[0069] The third step (S23) involves cluster analysis of the collected road surface slippery state data for each unit grid.

[0070] In the road surface wet slip condition data corresponding to each unit grid, the data from different detection points have both similarities and differences. By using cluster analysis, data with high sample similarity and small differences in the detection data can be grouped into one category.

[0071] If most of the road surface within a cell grid is in a slippery state, then that cell grid is necessarily slippery; that is, the slipperiness coefficient of the road surface at most points within the cell grid is closer to 0. By using a clustering algorithm to group these similar slippery road surface data together, the centroid of the cluster with the largest number of clustered samples represents the slippery state of most areas within the cell grid and can be used as a parameter characterizing the overall slippery state of the road surface within that cell grid.

[0072] In a preferred embodiment of the present invention, the K-means clustering algorithm is used to perform clustering operations on the road surface slipperiness state data of each cell grid. This algorithm is an iterative clustering algorithm that uses Euclidean distance as a similarity index to cluster discrete data into K clusters (classes), with the centroid of each cluster being the center point. By calculating the distance between samples and clusters and iterating repeatedly, the algorithm can cluster samples with high similarity and low difference into one cluster, resulting in high similarity among samples within a cluster and high difference between different clusters, ultimately obtaining the optimal centroid value.

[0073] After moving the detection device to detect n points in a cell grid, the detection dataset X of the road surface wet skid coefficient of that cell grid can be obtained, that is, X = {X1, X2, ... X}. i ,…X n}, where 1≤i≤n. Assume the initial value of K is k0, and the initial cluster centroids are... Where 1≤j≤k0, C j Let K be the initial centroid of the j-th cluster, physically representing the coefficient of slippage. The initial K value can generally be randomly selected from an integer between 1 and 10, and the initial centroid value can generally be randomly selected from a value between 0 and 1. The choice of initial values ​​does not affect the final clustering result.

[0074] The following formula (1) can be used to calculate the value of each data sample X in the detection dataset X. i To the cluster centroid C of each cluster j European distance,

[0075]

[0076] Step A: Compare each data sample X sequentially. i To each cluster centroid C j The Euclidean distance between each data sample X i Assign to the nearest cluster centroid C j From the clusters, we obtain k0 clusters. Among them, S j This represents the j-th cluster, which physically means the data set of the j-th cluster after re-clustering.

[0077] Step B: Recalculate the S values ​​for each cluster using the following formula (2). j The center of mass,

[0078]

[0079] Among them, C m Denotes the centroid of the m-th cluster, 1≤m≤k0; |S m | represents the number of data samples in the m-th cluster; X d This represents the d-th object in the m-th cluster, where 1 ≤ d ≤ |S|. m |; that is, for those belonging to S m Calculate the average value of all objects in a cluster.

[0080] Step C: Repeat Step A and Step B until the centroids of the k0 clusters no longer change, that is, the clustering results of the road surface slippery state data of the cell grid when K=k0 are obtained.

[0081] Considering the uncertainty of the road surface's slippery state within the cell mesh, the K value can be further optimized to obtain more accurate results:

[0082] Optimizing the K-value can be achieved by comparing the overall silhouette coefficients corresponding to different numbers of clusters (different K-values) and selecting the number of clusters (K-value) corresponding to the maximum overall silhouette coefficient. That is, the optimal solution for the K-value is determined with the goal of maximizing the overall silhouette coefficient. The silhouette coefficient is a key indicator describing the differences between clusters, combining the cluster's cohesion and separation, and is used to evaluate the clustering effect. Its value is between -1 and 1, with a larger value indicating a better clustering effect.

[0083] In a preferred embodiment of the present invention, the optimization of the K value is performed through the following steps:

[0084] Step 1: For the i-th data sample X i Calculate X i The average distance to all other elements in the same cluster is denoted as a. i .

[0085] Step 2: Select another cluster b and calculate X. i Find X by traversing all other clusters and finding the average distance from all elements in b. i The average distance to the nearest cluster from all other clusters is denoted as b. i .

[0086] Step 3: Calculate element X using the following formula (3). i The profile coefficient,

[0087]

[0088] Step 4: Calculate the silhouette coefficients of all data samples X, and then calculate the overall silhouette coefficient of the current cluster (K=k0) using the following formula (4):

[0089]

[0090] Step 5: Adjust the value of K, for example, by taking an integer between 2 and 10. Repeat steps A to C to obtain the clustering results, and calculate the overall silhouette coefficient corresponding to each K value using steps 1 to 4, where the maximum overall silhouette coefficient E is... k' The corresponding K = k' is the optimal K value.

[0091] Finally, based on the results of the above cluster analysis, the road surface slip coefficient Sp of this cell grid was determined.

[0092] According to the value of k' (the optimal K value), determine the centroids of the k' clusters and the number of samples contained in each cluster, and determine the centroid of the cluster with the largest number of samples as the road surface slipperiness coefficient Sp of the unit grid. The cluster containing the largest number of aggregated samples can best reflect the overall slipperiness state of the unit grid, and the road surface slipperiness coefficient value Sp corresponding to the centroid of this cluster is also most suitable for representing the overall road surface slipperiness coefficient of the grid unit.

[0093] In the aforementioned K-means clustering algorithm, the selection of the K value is an important factor affecting the final calculation result. Using a fixed K value results in relatively low calculation load, but there may be certain deviations in the calculation of the road surface slipperiness coefficient of unit grids. Through the step of K value optimization, the selection of K value can be optimized according to the actual situation of detection data in unit grids, which helps improve the accuracy of calculating the road surface slipperiness coefficient of unit grids.

[0094] The advantages of adopting the K-means clustering algorithm are that the data converges fast, which can meet the requirement of real-time detection of tunnel road surface slipperiness state; moreover, the classified data is scalable, and can reflect the slipperiness state of unit grids in a scientific and realistic manner as much as possible.

[0095] Through clustering analysis of detection data at multiple points in unit grids, the distribution of detected road surface slipperiness coefficient values can be scientifically judged, and then the overall road surface slipperiness state of the unit grid can be accurately determined. The clustering method can not only effectively eliminate the influence of point data with inaccurate detection caused by interference from external factors, but also shield the influence of very small-sized slippery areas that do not endanger driving safety on the overall road surface slipperiness coefficient value of the unit grid.

[0096] Based on the road surface slipperiness coefficient of each unit grid in the tunnel, the road surface slipperiness state can be divided into several levels, which serves as a basis for road traffic flow control and tunnel maintenance.

[0097] For example, three thresholds m1, m2, m3 can be set for the road surface slipperiness coefficient of unit grids, where 0<m1<m2<m3<1, so that the road surface slipperiness state is divided into four levels, namely:

[0098] Level I (dry), Sp∈(m3,1);

[0099] Level II (slightly slippery), Sp∈(m2,m3);

[0100] Level III (relatively slippery), Sp∈(m1,m2);

[0101] Level IV (extremely slippery), Sp∈(0,m1).

[0102] Those skilled in the art should understand that, depending on the subsequent application and the actual needs of tunnel traffic flow management, the road surface slippery condition can be divided into more or fewer levels; and the intervals between the various thresholds can be the same or different.

[0103] Based on an accurate assessment of the current slippery condition of the road surface inside the tunnel, road managers can effectively control traffic flow on the road, including speed control, lane control, or by issuing information prompts through variable message signs.

[0104] Those skilled in the art will understand that, as the mobile detection device moves, it can detect n cell grids at different locations within the tunnel, for example, in the order of 1, 2, ..., n. By comparing the detection results of cell grids at different locations, the change in the wet and slippery state of the tunnel pavement with location can be obtained, thus revealing the spatial variation law of the wet and slippery state of the pavement.

[0105] Furthermore, as the mobile detection equipment moves back and forth, it can cyclically detect n grid cells within the tunnel, for example, continuously in the order of 1, 2, ..., n, n, ..., 2, 1. In this way, each grid cell is detected once at specific time intervals. By comparing the detection results at different times, the change in the road surface slipperiness of that grid cell over time can be obtained, thus providing a basis for predicting the changing trend of the road surface slipperiness.

[0106] Based on reliable assessments of the spatiotemporal variation trends of slippery road surfaces within tunnels, road managers can develop more scientific plans for the maintenance of tunnel surfaces and infrastructure.

[0107] As can be seen from the foregoing description in conjunction with the preferred embodiments of the present invention, the method for detecting the slippery state of tunnel pavement based on a mobile detection device of the present invention improves the spatial resolution of the detection of the slippery state of tunnel pavement, enabling accurate detection of the slippery pavement condition in the tunnel using a small unit grid as the basic detection unit; and improves the temporal resolution of the detection of the slippery state of tunnel pavement, enabling effective tracking of changes in the slippery state of the pavement in the tunnel.

[0108] It should be understood that the embodiments described above can be adjusted within a reasonable range according to the specific application, and are not limited to the specific situations in the above embodiments. That is to say, those skilled in the art can make various changes and / or modifications to specific embodiments without departing from the concept of the present invention.

Claims

1. A detection method for a slippery state of a tunnel pavement based on non-contact mobile detection equipment, wherein the mobile detection equipment is arranged on a track in a space above the tunnel pavement and can reciprocate along the track in a length direction of the tunnel pavement, and the mobile detection equipment can detect pavement slippery coefficient values at a plurality of different positions on the tunnel pavement, the detection method comprises the following steps: Step S21: dividing the tunnel pavement into a plurality of unit grids along the length direction; Step S22: controlling the mobile detection equipment to move above the tunnel pavement, and collecting pavement slippery state data corresponding to each unit grid, wherein the pavement slippery state data comprises a detection data set consisting at least of detection time, detection positions and pavement slippery coefficient values corresponding to different detection points in the unit grid; Step S23: performing cluster analysis on the pavement slippery state data of each unit grid; Step S24: determining the pavement slippery coefficient of each unit grid according to a result of the cluster analysis.

2. The detection method according to claim 1, wherein, A length of the unit grid is 80 to 120 meters.

3. The detection method according to claim 2, wherein, The length of the unit grid is 100 meters.

4. The detection method according to claim 1, wherein, The pavement slippery state data comprises pavement slippery coefficient values of 200 to 800 detection points in the unit grid.

5. The detection method according to claim 4, wherein, The pavement slippery state data comprises pavement slippery coefficient values of 400 detection points in the unit grid.

6. The detection method according to claim 1, wherein, The detection dataset is X = {X1, X2, ..., X} i ,…X n }, where 1≤i≤n, step S23 further includes the following steps: The number of cluster centroids K is randomly selected, where K is an integer, 1 ≤ K ≤ 10, and the randomized cluster centroids are C = {C1, C2, ... C}. j ,…C K }, where 1≤j≤K, C j Let C be the centroid of the j-th cluster, 0 ≤ C j ≤1; Step A: Calculate the value of each data sample X in the detection dataset X using formula (1). i To the cluster centroid C of each cluster j European distance, Compare each data sample X in turn i To each cluster centroid C j The Euclidean distance between each data sample X i Assign to the nearest cluster centroid C j From the clusters, we obtain K clusters {S1, S2, ... S}. j ,…S K }, where S j Let j represent the j-th cluster, which is the data set of the j-th cluster after re-clustering; Step B: Recalculate the S values ​​for each cluster using formula (2). j The center of mass, Among them, C m Let S represent the centroid of the m-th cluster, 1 ≤ m ≤ K; m | represents the number of data samples in the m-th cluster; X d This represents the d-th object in the m-th cluster, where 1 ≤ d ≤ |S|. m |; Step C: repeatedly executing step A and step B until centroids of K clusters no longer change, and obtaining a cluster analysis result of the pavement slippery state data of the current unit grid corresponding to a current K value.

7. The detection method according to claim 6, further comprising a step of optimizing a K value: Step 1: For data sample X i Calculate X i The average distance to all other elements in the same cluster is denoted as a. i ; Step 2: Select another cluster b and calculate X. i Find the average distance to all elements in b, and iterate through all other clusters to find X. i The average distance to the nearest cluster from all other clusters is denoted as b. i ; Step 3: Calculate element X using formula (3) i The profile coefficient, Step 4: calculating a silhouette coefficient of all data samples X, and calculating an overall silhouette coefficient corresponding to a current K value by formula (4), Step 5: Adjust the value of K to any other integer between 2 and 10, repeat steps A to C, and calculate the overall profile coefficient corresponding to each K value using steps 1 to 4, where the maximum overall profile coefficient E is... k' The corresponding K = k' is the optimal K value; Step 6: taking a clustering result corresponding to an optimal K value as a result of the cluster analysis.

8. The detection method according to claim 6 or 7, wherein, Said step S24 further comprises: determining a centroid of a clustering cluster containing the largest number of samples as the pavement slippery coefficient Sp of the unit grid.

9. The detection method according to claim 1 or 2, further comprising: setting three thresholds m1, m2, and m3 for the pavement slippery coefficient of the unit grid, wherein 0<m1<m2<m3<1, so as to divide the pavement slippery state into four evaluation grades, that is: Grade I: Sp∈(m3,1); Grade II: Sp∈(m2,m3); Grade III: Sp∈(m1,m2); Grade IV: Sp∈(0,m1).

Images