A method, device and medium for monitoring road surface traffic conditions based on intelligent guardrails

Abstract

The embodiment of the application discloses a kind of based on intelligent guardrail road traffic condition monitoring method, equipment and medium, belong to wisdom traffic technical field, solve the problem that guardrail on current road is difficult to respond quickly when encountering bad weather.It includes, based on the multiple data acquisition devices on the guardrail of the road to be measured two sides, obtain the corresponding meteorological data and road image data of the road to be measured, and upload meteorological data and road image data to traffic management center;Through pre-set combination data extraction model, hazard element data is extracted from meteorological data;Based on the road image data and hazard element data corresponding to different target road sections respectively, the road traffic condition risk prediction value is determined;Through edge detection algorithm, target area is identified to road image data, and the actual risk value of road surface is determined based on target area;Based on road traffic condition risk prediction value and actual risk value of road surface, control two sides guardrail variation, to carry out road traffic regulation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent transportation, and in particular to a road surface traffic condition monitoring method based on an intelligent guardrail, a device and a medium. BACKGROUND

[0002] With the acceleration of urbanization and the continuous increase of traffic flow, road traffic safety problems are increasingly prominent. In order to effectively monitor and warn road traffic conditions, improve road traffic efficiency and safety, modern traffic management systems are gradually developing towards intelligent and fine.

[0003] When encountering severe weather environments such as heavy rain and snow, severe cold and freezing, roads often face the serious challenges of icing and water accumulation. In the prior art, although the method of installing monitoring cameras has been used to monitor road traffic conditions, this method only stays at the level of surface observation and highly depends on real-time monitoring of monitoring pictures by manual labor, which increases labor costs and is prone to omissions and delays in actual operation.

[0004] Guardrails are the most basic traffic protection facilities on highways and urban roads and play a very important role in traffic protection. The current design of guardrails on roads mostly still stays at the traditional level and only blocks vehicle collisions and the like, with a low degree of intelligence, which makes it difficult to respond quickly when encountering severe weather environments, thereby making it difficult to reduce the incidence of traffic accidents. SUMMARY

[0005] The embodiments of the present application provide a road surface traffic condition monitoring method based on an intelligent guardrail, a device and a medium, which are used to solve the technical problem that the current design of guardrails on roads only blocks vehicle collisions and the like, with a low degree of intelligence, which makes it difficult to respond quickly when encountering severe weather environments, thereby making it difficult to reduce the incidence of traffic accidents.

[0006] The embodiments of the present application adopt the following technical solutions:

[0007] The embodiment of the present application provides a kind of based on intelligent guardrail road traffic condition monitoring method.The method includes, based on the multiple data acquisition devices on the guardrail of the road to be measured, obtain the meteorological data and road image data corresponding to the road to be measured, and meteorological data and road image data are uploaded to traffic management center;Wherein, multiple data acquisition devices are respectively set in different target sections of the road to be measured;In traffic management center, hidden danger element data extraction is carried out on meteorological data by preset combination data extraction model;Based on the road image data and hidden danger element data corresponding to different target sections, the road traffic condition risk prediction value corresponding to the road to be measured is determined;The target area is identified in road image data by edge detection algorithm, to determine the actual risk value of road based on target area;Based on road traffic condition risk prediction value and actual risk value of road, the road traffic condition is graded early warning, and based on the graded early warning control two sides guardrail variation, to carry out road traffic regulation.

[0008] The embodiment of the present application realizes the automatic collection of road meteorological data and image data by installing multiple data acquisition devices on the guardrail, reduces the dependence of manual monitoring, improves the efficiency and accuracy of data acquisition. Traffic management center uses preset combination data extraction model and edge detection algorithm, which can automatically analyze data and predict road traffic condition risk, realizes the intelligentization of traffic management. Secondly, data acquisition devices are respectively set in different target sections, which can obtain more detailed and comprehensive road information, so that risk prediction is more accurate. Through the grading early warning mechanism, corresponding warning measures can be taken according to different risk levels, so as to timely remind the driver and traffic manager to pay attention to road condition and reduce the risk of traffic accident. According to the warning result, the guardrail is controlled to change, and the road traffic regulation is carried out, such as adjusting lane width and limiting speed, to further improve the safety and efficiency of road traffic.

[0009] In one implementation of the present application, before hidden danger element data extraction is carried out on meteorological data by preset combination data extraction model, the method further comprises: inputting historical meteorological data samples as training set into preset deep learning model to train preset deep learning model;Based on preset feature interaction table, the interaction influence relationship between different initial element data output by preset deep learning model is determined;Wherein, preset feature interaction table includes multiple different initial element data, and also includes the interaction influence relationship between multiple initial element data;Based on the interaction influence relationship, different initial element data is weighted and averaged to generate fusion training set;The fusion training set is input into preset gradient boosting tree model to train preset gradient boosting tree model;Based on trained preset deep learning model and trained preset gradient boosting tree model, preset combination data extraction model is obtained.

[0010] In an implementation form of the present application, different initial element data are weighted and averaged based on the interaction influence relationship to generate a fusion training set, specifically comprising:

[0011] Based on the preset data interaction function:

[0012]

[0013] The different initial element data are weighted and averaged to generate a fusion training set.

[0014] wherein X fused is the fused data; w ij represents the interaction weight between the i-th initial element data and the j-th initial element data, x i represents the feature value of the i-th initial element data; and x j represents the feature value of the j-th initial element data.

[0015] In an implementation form of the present application, based on the road image data and the hazard element data corresponding to different target road sections, the road surface passing condition risk prediction value corresponding to the to-be-tested road surface is determined, specifically comprising: based on the road image data, the road surface condition type corresponding to different target road sections is determined; wherein the road surface condition type includes road surface icing and road surface water; in the case of road surface icing, the hazard pressure data and the hazard temperature data of the road are determined through the hazard element data, the road freezing and thawing time and the road freezing and thawing speed are generated based on the hazard pressure data and the hazard temperature data, and the road surface warning information is generated based on the road freezing and thawing time and the road freezing and thawing speed; in the case of road surface water, the hazard precipitation, hazard humidity and hazard temperature of the road are determined through the hazard element data, the water accumulation dissipation time and the water accumulation dissipation speed are determined based on the hazard precipitation, hazard humidity, hazard temperature and the drainage system data corresponding to the to-be-tested road surface, and the road surface warning information is generated based on the water accumulation dissipation time and the water accumulation dissipation speed; based on the road surface warning information corresponding to different target road sections, the road surface passing condition risk prediction value corresponding to the to-be-tested road surface is determined.

[0016] In an implementation manner of the present application, the road surface passing state risk prediction value corresponding to the to-be-tested road surface is determined based on the road surface warning information corresponding to different target road sections, specifically including: determining the position information corresponding to different target road sections based on the installation positions of the guardrails; determining the historical traffic accident data corresponding to each target road section in the historical traffic database based on the position information, to perform first weight assignment on each target road section based on the historical traffic accident data; determining the road structure data corresponding to each target road section in the traffic network map corresponding to the to-be-tested road surface based on the position information, to perform second weight assignment on each target road section based on the road structure data; and performing weighted processing on the first weight assignment, the second weight assignment, and the road surface warning information corresponding to different target road sections, to obtain the road surface passing state risk prediction value corresponding to the to-be-tested road surface.

[0017] In an implementation manner of the present application, the target region recognition is performed on the road image data by using the edge detection algorithm, to determine the actual risk value of the road surface based on the target region, specifically including: performing target region detection on the road image data; wherein the target region at least includes the icing region and the water accumulation region; determining the boundary between the target region and the road surface based on the edge detection algorithm, to determine the area information of the target region based on the boundary; determining the water level line according to the image gray value difference between the target region and the road surface image, to obtain the water accumulation depth based on the water level line; obtaining the reference risk value corresponding to the road surface based on the area information and the water accumulation depth; and obtaining the road vehicle passing rate based on the preset time length and the corresponding vehicle passing number within the preset time length; adjusting the reference risk value based on the road vehicle passing rate, to obtain the actual risk value of the road surface.

[0018] In an implementation manner of the present application, the road surface passing state is graded and warned based on the road surface passing state risk prediction value and the actual risk value of the road surface, and the road passing is adjusted by controlling the change of the guardrails on both sides of the road based on the graded warning, specifically including: obtaining the risk value in the future period corresponding to the to-be-tested road surface based on the road surface passing state risk prediction value and the actual risk value of the road surface; determining the risk threshold value corresponding to different preset risk levels, dividing the risk value in the future period based on the risk threshold value, to obtain a plurality of risk time periods; matching the corresponding guardrail adjustment strategy based on different risk levels, and associating the guardrail adjustment strategy with the plurality of risk time periods; and controlling the change of the guardrails on both sides of the road in the corresponding risk time period based on the association relationship, to adjust the road passing.

[0019] In an implementation form of the present application, the risk value of the to-be-tested road in the future period is obtained based on the road traffic condition risk prediction value and the actual road risk value, and specifically includes: inputting the road traffic condition risk prediction value and the actual road risk value into a time series prediction model to obtain a reference risk value corresponding to the to-be-tested road; and constructing a risk value change line graph based on the reference risk value to determine the risk value of the to-be-tested road in the future period based on the risk value change line graph.

[0020] The embodiment of the present application provides a road traffic condition monitoring device based on an intelligent guardrail, which comprises at least one processor and a memory in communication connection with the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: acquire meteorological data and road image data corresponding to a to-be-tested road based on a plurality of data acquisition devices on guardrails on both sides of the to-be-tested road, and upload the meteorological data and the road image data to a traffic management center; the plurality of data acquisition devices are arranged at different target road sections of the to-be-tested road; hidden danger element data of the meteorological data is extracted by a preset combined data extraction model at the traffic management center; a road traffic condition risk prediction value corresponding to the to-be-tested road is determined based on road image data and hidden danger element data corresponding to different target road sections; a road actual risk value is determined based on target region identification of the road image data by an edge detection algorithm; and a graded early warning of the road traffic condition is performed based on the road traffic condition risk prediction value and the road actual risk value, and the guardrails on both sides are controlled based on the graded early warning to adjust road traffic.

[0021] The embodiment of the present application provides a non-volatile computer storage medium, which stores computer executable instructions, and the computer executable instructions are configured to: acquire meteorological data and road image data corresponding to a to-be-tested road based on a plurality of data acquisition devices on guardrails on both sides of the to-be-tested road, and upload the meteorological data and the road image data to a traffic management center; the plurality of data acquisition devices are arranged at different target road sections of the to-be-tested road; hidden danger element data of the meteorological data is extracted by a preset combined data extraction model at the traffic management center; a road traffic condition risk prediction value corresponding to the to-be-tested road is determined based on road image data and hidden danger element data corresponding to different target road sections; a road actual risk value is determined based on target region identification of the road image data by an edge detection algorithm; and a graded early warning of the road traffic condition is performed based on the road traffic condition risk prediction value and the road actual risk value, and the guardrails on both sides are controlled based on the graded early warning to adjust road traffic.

[0022] The above at least one technical solution adopted by the embodiments of the present application can achieve the following beneficial effects: The embodiments of the present application realize automatic collection of road surface meteorological data and image data by installing multiple data acquisition devices on the guardrails, reduce the dependence on manual monitoring, and improve the efficiency and accuracy of data acquisition. The traffic management center can automatically analyze data and predict road traffic condition risks by using the pre-set combined data extraction model and edge detection algorithm, realizing intelligent traffic management. Secondly, the data acquisition devices are respectively arranged on different target road sections, which can obtain more detailed and comprehensive road surface information, making the risk prediction more accurate. Through the hierarchical early warning mechanism, corresponding warning measures can be taken according to different risk levels, timely reminding the drivers and traffic managers to pay attention to the road surface conditions, and reducing the risk of traffic accidents. According to the warning result, the guardrails are controlled to change, and the road traffic is adjusted, such as adjusting the lane width and limiting the speed, further improving the safety and efficiency of road traffic. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art without creative labor. In the drawings:

[0024] Figure 1 A flow chart of a road traffic condition monitoring method based on an intelligent guardrail is provided for the embodiments of the present application.

[0025] Figure 2 A structural schematic diagram of a road traffic condition monitoring device based on an intelligent guardrail is provided for the embodiments of the present application.

[0026] Reference signs:

[0027] 200: a road traffic condition monitoring device based on an intelligent guardrail, 201: a processor, 202: a memory. DETAILED DESCRIPTION

[0028] The embodiments of the present application provide a road traffic condition monitoring method, device and medium based on an intelligent guardrail.

[0029] In order to enable the person skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by the person skilled in the art without creative labor should belong to the protection scope of the present application.

[0030] The technical solutions proposed by the embodiments of the present application will be described in detail below with reference to the drawings.

[0031] The intelligent guardrail utilizes precise perception technology and Internet of Things technology to realize real-time perception of multiple parameters such as inclination, displacement and vibration of the traffic guardrail, and to transmit the data to the traffic management center in real time through the NB-IOT network, so as to realize real-time intelligent monitoring and automatic early warning of the guardrail. The traffic management center supports one-stop alarm, receiving and operation and maintenance management, can automatically decide the fault processing flow, automatically assign orders and workers, and automatically confirm maintenance, and establishes and improves an integrated, standardized and automated rapid response mechanism, so as to realize real-time and reliable intelligent monitoring of the guardrail, predict dangers in advance, prevent secondary traffic accidents caused by abnormal guardrails, reduce safety accidents, and improve operation and management capabilities.

[0032] The intelligent guardrail parameters in the embodiments of the present application are as follows:

[0033] Wireless communication: NB-IoT / 4G Internet of Things technology is adopted, and a SIM card is built-in.

[0034] Inclination monitoring: a low-power six-axis acceleration sensor is adopted, inclination 0-180 degree alarm is supported, remote threshold setting is supported, and a built-in anti-shake processing algorithm is adopted.

[0035] Vibration monitoring: a high-precision vibration sensor is adopted, remote acceleration setting is supported, and real-time vibration monitoring is supported.

[0036] Displacement monitoring: GPS+Beidou positioning is adopted, and high positioning accuracy is achieved.

[0037] Standby current: micro-power technology is adopted, an ARM architecture processor is used, and the sleep current is small.

[0038] Power supply mode: solar power supply is adopted.

[0039] Data upload frequency: fast response, when the guardrail inclination, vibration and displacement events are detected, the data is uploaded within 15 seconds.

[0040] Operating temperature: -40-85℃.

[0041] Waterproof level: IP68.

[0042] Open protocol docking third-party platform, convenient integration, unified management.

[0043] Figure 1 A flow chart of a road surface traffic condition monitoring method based on an intelligent guardrail is provided for the embodiments of the present application, as shown in Figure 1 The road surface traffic condition monitoring method based on the intelligent guardrail includes the following steps:

[0044] S101, based on a plurality of data acquisition devices on the guardrails on both sides of the road to be tested, acquiring corresponding meteorological data and road image data of the road to be tested, and uploading the meteorological data and road image data to the traffic management center.

[0045] In an embodiment of the present application, a plurality of data acquisition devices are installed on the guardrails on both sides of the road to be tested. Among them, the plurality of data acquisition devices are respectively arranged on different target road sections of the road to be tested. These devices include meteorological sensors such as temperature sensors, humidity sensors, anemometers, etc., and image acquisition equipment such as high-definition cameras, and ensure that they can cover different areas of the road.

[0046] Further, the meteorological sensor is responsible for real-time monitoring and recording of meteorological data of the road and its surroundings, such as temperature, humidity, wind speed, wind direction, precipitation, etc., and the image acquisition equipment is responsible for capturing road image data, including vehicle flow, pedestrian activity, road conditions (such as water accumulation, icing, cracks, etc.). The collected meteorological data and road image data are uploaded in real time to the traffic management center through wired or wireless means, such as 4G / 5G network, Wi-Fi, dedicated communication network, etc.

[0047] S102, in the traffic management center, hidden danger element data extraction is performed on the meteorological data through a preset combined data extraction model.

[0048] In an embodiment of the present application, historical meteorological data samples are input into a preset deep learning model as a training set to train the preset deep learning model. Based on a preset feature interaction table, the interaction influence relationship between different initial element data output by the preset deep learning model is determined; wherein the preset feature interaction table includes a plurality of different initial element data, and also includes the interaction influence relationship between the plurality of initial element data. Based on the interaction influence relationship, the different initial element data are weighted and averaged to generate a fusion training set. The fusion training set is input into a preset gradient boosting tree model to train the preset gradient boosting tree model. Based on the trained preset deep learning model and the trained preset gradient boosting tree model, a preset combined data extraction model is obtained.

[0049] Specifically, historical meteorological data samples are collected and organized, which include data of meteorological elements such as temperature, humidity, wind speed, precipitation, etc. These historical meteorological data samples are input into the preset deep learning model as a training set to train the preset deep learning model. The deep learning model in the embodiments of the present application can be a convolutional neural network.

[0050] Further, the interaction influence relationship between different initial element data output by the deep learning model is determined through the preset feature interaction table. The feature interaction table in the embodiments of the present application is a priori knowledge base, which lists a plurality of initial element data (such as temperature, humidity, etc.) and their interaction influence relationship (such as temperature rise may cause humidity to decrease). Based on the determined interaction influence relationship, the different initial element data are weighted and averaged. Through weighted averaging, a fusion training set is generated, which contains the initial element data adjusted by the interaction influence relationship.

[0051] Further, the generated fusion training set is input into the preset gradient boosting tree model. It should be noted that gradient boosting tree is a powerful ensemble learning method that improves prediction performance by constructing multiple weak learners and combining their results. Through training, the gradient boosting tree model learns the relationship and pattern between the data in the fusion training set. The trained deep learning model and the trained gradient boosting tree model are combined to construct a preset combined data extraction model. The preset combined data extraction model can take advantage of the deep learning model in capturing complex relationships, as well as the ability of the gradient boosting tree model in handling nonlinear relationships and feature interactions. By combining the two models, more accurate and comprehensive data extraction results can be obtained.

[0052] In an embodiment of the present application, based on the preset data interaction function:

[0053]

[0054] The different initial element data are weighted and averaged to generate a fusion training set;

[0055] wherein X fused is the fused data; w ij represents the interaction weight between the i th initial element data and the j th initial element data, x i represents the feature value of the i th initial element data; and x j represents the feature value of the j th initial element data.

[0056] S103, based on the road image data and the hazard element data corresponding to different target road sections respectively, determine the road surface traffic condition risk prediction value corresponding to the to-be-measured road surface.

[0057] In an embodiment of the present application, based on the road image data, the road surface condition types corresponding to different target road sections are determined; wherein the road surface condition types include road icing and road water accumulation; in the case of road icing, the hazard pressure data and the hazard temperature data of the road are determined through the hazard element data, the road freezing and thawing time and the road freezing and thawing speed are generated based on the hazard pressure data and the hazard temperature data, and the road surface warning information is generated based on the road freezing and thawing time and the road freezing and thawing speed. In the case of road water accumulation, the hazard precipitation, the hazard humidity and the hazard temperature of the road are determined through the hazard element data, the water accumulation dissipation time and the water accumulation dissipation speed are determined based on the hazard precipitation, the hazard humidity, the hazard temperature and the drainage system data corresponding to the road surface to be measured, and the road surface warning information is generated based on the water accumulation dissipation time and the water accumulation dissipation speed. Based on the road surface warning information corresponding to different target road sections, the road surface traffic condition risk prediction value corresponding to the road surface to be measured is determined.

[0058] Specifically, by using the road image data, the road surface condition types corresponding to different target road sections are determined through image recognition technology, wherein the road surface condition types in the embodiments of the present application mainly include road icing and road water accumulation. When the road surface condition type is road icing, the hazard pressure data and the hazard temperature data of the road are determined through the hazard element data such as pressure and temperature in meteorological data, and the road freezing and thawing time and the road freezing and thawing speed are obtained based on the hazard pressure data and the hazard temperature data by using a preset calculation model. The training process of the preset calculation model is as follows: taking the hazard element data sample as input, taking the road freezing and thawing time and the road freezing and thawing speed sample data corresponding to the input sample as output, training the preset neural network model to obtain the preset calculation model. These parameters reflect the melting rate and the required time of the ice layer under specific meteorological conditions. According to the road freezing and thawing time and the road freezing and thawing speed, the road surface warning information is generated. The warning information can include the icing degree, the expected recovery time, etc.

[0059] Further, when the road surface condition type is road water accumulation, the water accumulation dissipation time and the water accumulation dissipation speed are obtained through the preset calculation model by using the hazard element data such as precipitation, humidity, temperature, etc., and the drainage system data corresponding to the road surface to be measured. The training process of the preset calculation model is as follows: taking the hazard element data sample as input, taking the water accumulation dissipation time and the water accumulation dissipation speed sample data corresponding to the input sample as output, training the preset neural network model to obtain the preset calculation model. Wherein, the drainage system data can include the capacity of the drainage pipeline, the drainage efficiency and other information. According to the water accumulation dissipation time and the water accumulation dissipation speed, the road surface warning information is generated. The warning information can include the water accumulation degree, the expected recovery time, etc.

[0060] Further, based on the pavement warning information corresponding to different target road sections respectively, the pavement traffic condition risk prediction value corresponding to the to-be-tested pavement is determined.

[0061] In an embodiment of the present application, based on the installation position of the guardrail, the position information corresponding to different target road sections respectively is determined. Based on the position information, the historical traffic accident data corresponding to each target road section respectively is determined in the historical traffic database, so as to perform first weight assignment on each target road section based on the historical traffic accident data. Based on the position information, the road structure data corresponding to each target road section respectively is determined in the traffic network map corresponding to the to-be-tested pavement, and second weight assignment is performed on each target road section respectively based on the road structure data. The first weight assignment, the second weight assignment and the pavement warning information corresponding to different target road sections respectively are weighted to obtain the pavement traffic condition risk prediction value corresponding to the to-be-tested pavement.

[0062] Specifically, the guardrail is usually installed at the key positions of the road, such as bridges, tunnels, sharp bends, steep slopes, etc., and these positions are more likely to cause traffic accidents. By measuring the installation position of the guardrail, the specific position information of different target road sections in the traffic network map can be determined, such as the starting point, the ending point, the length, the direction, etc. Using the established historical traffic database, according to the position information of the target road section, the traffic accident data of each road section occurring in the past period of time can be retrieved, including the accident type, the accident cause, the number of casualties, the vehicle damage condition, etc. According to the severity and frequency of the historical traffic accident data, weight assignment is performed on each target road section. In the traffic network map of the to-be-tested pavement, according to the position information of the target road section, the road structure data of each road section can be obtained, including the road grade, the number of lanes, the pavement material, the design speed, the traffic signs and markings, etc. According to the advantages and disadvantages and the complexity of the road structure data, weight assignment is performed on each target road section.

[0063] Further, the first weight assignment, the second weight assignment and the pavement warning information corresponding to different target road sections respectively are weighted, and the result of the weighted processing is the pavement traffic condition risk prediction value corresponding to the to-be-tested pavement. This value reflects the traffic risk that each road section may face in the current and future period of time.

[0064] S104, target region recognition is performed on the road image data by an edge detection algorithm, so as to determine the pavement actual risk value based on the target region.

[0065] In an embodiment of the present application, target region detection is performed on the road image data; wherein the target region at least includes an icy region and a waterlogged region. Based on an edge detection algorithm, the boundary between the target region and the road surface is determined to determine the area information of the target region based on the boundary. According to the difference in image gray value between the target region and the road surface image, the water level line is determined to obtain the waterlogged depth based on the water level line. Based on the area information and the waterlogged depth, the reference risk value corresponding to the road surface is obtained. And based on the preset time length and the corresponding number of vehicle passing within the preset time length, the road surface vehicle passing rate is obtained. The reference risk value is adjusted based on the road surface vehicle flow passing rate to obtain the actual risk value of the road surface.

[0066] Specifically, using image processing technology, the road image data is processed to identify and mark the target region. These target regions at least include icy regions and waterlogged regions. The edge detection algorithm (such as Canny edge detection, Sobel operator, etc.) is applied to identify the boundary between the target region and the road surface. By calculating the area of the region surrounded by the boundary, the area information of the target region is obtained.

[0067] Further, the gray value difference between the target region, especially the waterlogged region and the road surface image is analyzed. According to the trend of the gray value, the water level line position is determined, and the waterlogged depth is calculated through the relationship between the water level line position and the road surface height. The waterlogged depth is an important indicator for evaluating the risk of waterlogging. Combined with the area information of the target region and the waterlogged depth, the reference risk value corresponding to the road surface is determined. This value reflects the risk level that the road passing may face under the current conditions, for example, the larger the target region area and the deeper the waterlogged depth, the larger the reference risk value.

[0068] Further, a preset time length, such as the past ten minutes, is set to count the number of vehicles passing through the road within the time length. By calculating the ratio of the number of vehicles to the preset time length, the road surface vehicle passing rate is obtained. This value reflects the traffic flow of the road. Considering the influence of traffic flow on road risk, the reference risk value is adjusted using the road surface vehicle passing rate. If the traffic flow is large, the road risk may increase accordingly, so the risk value needs to be increased; on the contrary, if the traffic flow is small, the road risk may decrease, so the risk value can be appropriately reduced. The adjusted risk value is the actual risk value of the road surface, which more accurately reflects the risk level of road passing under the current traffic flow conditions.

[0069] S105, based on the road passing condition risk prediction value and the road actual risk value, the road passing condition is graded and warned, and the two side guardrails are controlled to change based on the graded warning to adjust the road passing.

[0070] In one embodiment of this application, the risk value of the road surface in the future period is obtained based on the predicted risk value of road traffic conditions and the actual risk value of the road surface. Risk thresholds corresponding to different preset risk levels are determined, and the risk value in the future period is divided based on these risk thresholds to obtain multiple risk time periods. Based on different risk levels, corresponding guardrail adjustment strategies are matched, and these strategies are associated with multiple risk time periods. Based on these associations, the guardrails on both sides of the road are adjusted during the corresponding risk time periods to regulate road traffic.

[0071] Specifically, the calculated predicted road traffic condition risk value is combined with the actual road risk value to obtain a comprehensive risk assessment value. This value represents the risk level that the road surface under test may face in a future period. In one embodiment of this application, the predicted road traffic condition risk value and the actual road risk value are input into a time series prediction model to obtain a reference risk value corresponding to the road surface under test; a risk value change line graph is constructed based on the reference risk value, and the risk value of the road surface under test in the future period is determined based on the risk value change line graph.

[0072] Specifically, firstly, the predicted risk value of road traffic conditions is input into a time series prediction model. Through calculation by the time series prediction model, a reference risk value for the road surface under test in a future period can be obtained. This value is a prediction result that integrates historical and real-time data, making it more accurate and reliable than using either data alone. A risk value change line graph is then constructed using the reference risk value. This line graph shows the trend of risk value changes over time, providing a visual understanding of the risk level changes of the road surface under test in future periods. The risk value change line graph can include risk values ​​at multiple time points, which can be continuous (e.g., hourly, daily) or specific time points (e.g., peak traffic hours, periods of severe weather). Based on the risk value change line graph, the corresponding risk value for the road surface under test in the future period is determined.

[0073] Furthermore, different preset risk levels are established, and corresponding risk thresholds are determined for each level. These thresholds are used to divide the risk assessment values ​​within future time periods, resulting in multiple time periods with different risk levels. For example, the risk levels can be divided into three categories: low, medium, and high, with corresponding thresholds set for each. Based on different risk levels, corresponding guardrail adjustment strategies are developed. These strategies may include increasing the height or density of guardrails, or changing the type of guardrails, to adapt to road traffic needs under different risk levels. These guardrail adjustment strategies are then correlated with the previously defined risk time periods to ensure that more stringent guardrail adjustment measures are implemented during higher-risk time periods.

[0074] Further, based on the above correlation, the guardrails on both sides of the road are automatically controlled to change in the corresponding risk period. These changes aim to improve the safety of the road and reduce the risk of traffic accidents. For example, in a period with high risk, the height or density of the guardrails can be increased to better protect pedestrians and vehicles; while in a period with low risk, the density or height of the guardrails can be appropriately reduced to improve the efficiency of road traffic.

[0075] Figure 2 A structural schematic diagram of a road traffic condition monitoring device based on intelligent guardrails is provided for the embodiments of the present application. As shown in Figure 2 The road traffic condition monitoring device based on intelligent guardrails 200 includes at least one processor 201 and a memory 202 communicatively connected to the at least one processor 201. The memory 202 stores instructions executable by the at least one processor 201. The instructions are executed by the at least one processor 201 to enable the at least one processor 201 to: based on a plurality of data acquisition devices on the guardrails on both sides of the road to be tested, acquire corresponding weather data and road image data of the road to be tested, and upload the weather data and road image data to a traffic management center; wherein the plurality of data acquisition devices are respectively arranged at different target road sections of the road to be tested; in the traffic management center, hidden danger element data is extracted from the weather data by a pre-set combined data extraction model; based on the road image data and hidden danger element data corresponding to different target road sections, a road traffic condition risk prediction value corresponding to the road to be tested is determined; a target area is identified from the road image data by an edge detection algorithm to determine an actual risk value of the road based on the target area; based on the road traffic condition risk prediction value and the actual risk value of the road, a graded warning of the road traffic condition is performed, and the guardrails on both sides are controlled to change based on the graded warning to regulate road traffic.

[0076] A non-volatile computer storage medium is provided by the embodiments of the present application, which stores computer executable instructions. The computer executable instructions are configured to: based on a plurality of data acquisition devices on the guardrails on both sides of the road to be tested, acquire corresponding weather data and road image data of the road to be tested, and upload the weather data and road image data to a traffic management center; wherein the plurality of data acquisition devices are respectively arranged at different target road sections of the road to be tested; in the traffic management center, hidden danger element data is extracted from the weather data by a pre-set combined data extraction model; based on the road image data and hidden danger element data corresponding to different target road sections, a road traffic condition risk prediction value corresponding to the road to be tested is determined; a target area is identified from the road image data by an edge detection algorithm to determine an actual risk value of the road based on the target area; based on the road traffic condition risk prediction value and the actual risk value of the road, a graded warning of the road traffic condition is performed, and the guardrails on both sides are controlled to change based on the graded warning to regulate road traffic.

[0077] The various embodiments in this application are described in a progressive manner, and the same or similar parts among the various embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, the device, apparatus, and non-transitory computer storage medium embodiments are described simply because they are substantially similar to the method embodiments, and the relevant parts can be referred to the description of the method embodiments.

[0078] The above only describes the embodiments of the present application and is not intended to limit the present application. The embodiments of the present application can be variously changed and modified by those skilled in the art. These modifications or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for monitoring road surface traffic conditions based on intelligent guardrails, characterized in that, The method comprises: Based on a plurality of data acquisition devices on the guardrails on both sides of the to-be-tested road, meteorological data and road image data corresponding to the to-be-tested road are acquired, and the meteorological data and the road image data are uploaded to a traffic management center; wherein the plurality of data acquisition devices are respectively arranged on different target road sections of the to-be-tested road; In the traffic management center, hidden danger element data is extracted from the meteorological data by a pre-set combined data extraction model; Based on the road image data and the hidden danger element data corresponding to the different target road sections, a road surface traffic condition risk prediction value corresponding to the to-be-tested road is determined; A target area of the road image data is identified by an edge detection algorithm, so as to determine a road surface actual risk value based on the target area; Based on the road surface traffic condition risk prediction value and the road surface actual risk value, a hierarchical early warning of the road surface traffic condition is performed, and the guardrails on both sides are controlled based on the hierarchical early warning to adjust the road traffic; The determination of the road surface traffic condition risk prediction value corresponding to the to-be-tested road based on the road image data corresponding to the different target road sections and the hidden danger element data comprises: Based on the road image data, a road surface condition type corresponding to the different target road sections is determined; wherein the road surface condition type comprises road surface icing and road surface water accumulation; In the case that the road surface condition type is the road surface icing, hidden danger air pressure data and hidden danger temperature data of the road are determined based on the hidden danger element data, road icing and thawing time and road icing and thawing speed are generated based on the hidden danger air pressure data and the hidden danger temperature data, and road surface early warning information is generated based on the road icing and thawing time and the road icing and thawing speed; In the case that the road surface condition type is the road surface water accumulation, hidden danger precipitation, hidden danger humidity and hidden danger temperature of the road are determined based on the hidden danger element data, and accumulation water dissipation time and accumulation water dissipation speed are determined based on the hidden danger precipitation, the hidden danger humidity, the hidden danger temperature and drainage system data corresponding to the to-be-tested road, so as to generate the road surface early warning information based on the accumulation water dissipation time and the accumulation water dissipation speed; Based on the road surface early warning information corresponding to the different target road sections, the road surface traffic condition risk prediction value corresponding to the to-be-tested road is determined; The determination of the road surface traffic condition risk prediction value corresponding to the to-be-tested road based on the road surface early warning information corresponding to the different target road sections comprises: Based on the installation position of the guardrails, position information corresponding to the different target road sections is determined; Based on the position information, historical traffic accident data corresponding to each of the target road sections is determined in a historical traffic database, so as to perform first weight assignment on each of the target road sections based on the historical traffic accident data; Based on the position information, road structure data corresponding to each of the target road sections is determined in a traffic network diagram corresponding to the to-be-tested road, and second weight assignment is performed on each of the target road sections based on the road structure data. The first weight assignment, the second weight assignment, and the road surface early warning information corresponding to the different target road sections are weighted to obtain a road surface traffic condition risk prediction value corresponding to the to-be-tested road surface. The target region identification on the road image data is performed through an edge detection algorithm, and a road surface actual risk value is determined based on the target region, specifically including: The target region detection is performed on the road image data; wherein the target region at least includes an icing region and a water accumulation region; Based on the edge detection algorithm, the boundary between the target region and the road surface is determined to determine the area information of the target region based on the boundary; Based on the image gray value difference between the target region and the road surface image, a water level line is determined to obtain the water accumulation depth based on the water level line; Based on the area information and the water accumulation depth, a reference risk value corresponding to the road surface is obtained; Based on the preset time length and the corresponding number of vehicle passages within the preset time length, a road surface vehicle passage rate is obtained; The reference risk value is adjusted based on the road surface vehicle passage rate to obtain a road surface actual risk value; Based on the road surface traffic condition risk prediction value and the road surface actual risk value, a hierarchical early warning of the road surface traffic condition is performed, and the road traffic regulation is performed by controlling the change of the guardrails on both sides of the road based on the hierarchical early warning, specifically including: Based on the road surface traffic condition risk prediction value and the road surface actual risk value, a risk value within a future time period corresponding to the to-be-tested road surface is obtained; Risk threshold values corresponding to different preset risk levels are determined, and the risk value within the future time period is divided based on the risk threshold values to obtain a plurality of risk time periods; Based on different risk levels, corresponding guardrail adjustment strategies are matched, and the guardrail adjustment strategies are associated with a plurality of risk time periods; Based on the association relationship, the change of the guardrails on both sides of the road is controlled in the corresponding risk time period to regulate the road traffic. 2.The road surface traffic condition monitoring method based on the intelligent guardrail according to claim 1, wherein, Before the hazard element data extraction on the meteorological data through the preset combination data extraction model, the method further includes: The historical meteorological data samples are input into the preset deep learning model as a training set to train the preset deep learning model; Based on a preset feature interaction table, an interaction influence relationship between different initial element data output by the preset deep learning model is determined; wherein the preset feature interaction table includes a plurality of different initial element data, and further includes an interaction influence relationship between a plurality of initial element data; Based on the interaction influence relationship, different initial element data are weighted and averaged to generate a fusion training set; The fusion training set is input into the preset gradient boosting tree model to train the preset gradient boosting tree model; Based on the trained preset deep learning model and the trained preset gradient boosting tree model, the preset combination data extraction model is obtained. 3.The method of claim 2, wherein, Based on the interaction influence relationship, different initial element data are weighted and averaged to generate a fusion training set, specifically including: Based on a preset data interaction function: The different initial element data are weighted and averaged to generate the fused training set; wherein X fused is the fused data; w ij represents the interaction weight between the i-th initial element data and the j-th initial element data, x i represents the feature value of the i-th initial element data; and x j represents the feature value of the j-th initial element data. 4.The method of claim 1, wherein, The process of obtaining the future risk value of the road surface under test based on the predicted road traffic condition risk value and the actual road surface risk value specifically includes: The predicted risk value of the road traffic condition and the actual risk value of the road are input into the time series prediction model to obtain the reference risk value corresponding to the road surface to be tested. A risk value change line graph is constructed based on the reference risk value, and the risk value of the road surface to be tested in the future time period is determined based on the risk value change line graph.

5. An intelligent guard rail based road surface passable condition monitoring apparatus, characterized by, The device includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to perform the method described in any one of claims 1-4.

6. A non-transitory computer storage medium storing computer-executable instructions that, when executed, cause a computer to perform: The computer-executable instructions are capable of performing the method described in any one of claims 1-4.

Images