A highway traffic meteorological risk early warning method and system

Abstract

The application provides a highway traffic meteorological risk early warning method and system, relates to the field of highway meteorology, and comprises the following steps: acquiring and preprocessing multi-modal data, including satellite remote sensing data, radar-based data and real-time lightning positioning data; dividing the highway into road sections, performing grid division on the region near the highway according to the road sections, acquiring the relevant multi-modal data of the grids in the region near the highway, and analyzing the acquired data to obtain the meteorological conditions of each road section; performing risk type identification and grade division according to the meteorological conditions of each road section, and publishing a risk early warning report. The application adopts the above-mentioned highway traffic meteorological risk early warning method and system, realizes the precise perception-real-time analysis-high-efficiency decision-flexible publishing closed loop of highway meteorological risk, significantly improves the scientificity and response efficiency of traffic management under adverse weather, and has wide application prospects in the fields of intelligent traffic and emergency management.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of highway meteorology, in particular to a highway traffic meteorological risk early warning method and system. BACKGROUND

[0002] Patent 202411725123.1 discloses a highway traffic meteorological disaster risk prediction model testing method and device, which comprises obtaining traffic meteorological disaster event information of a selected highway section within a preset time range; based on the space-time deviation data, the first time range and the first road section range are corrected respectively, and the second time range and the second road section range of each traffic meteorological disaster event are obtained; based on the second time range and the second road section range of each traffic meteorological disaster event, the test samples of each traffic meteorological disaster event are determined from the risk prediction data output by the to-be-tested highway traffic meteorological disaster risk prediction model; based on the test samples of each traffic meteorological disaster event, the to-be-tested highway traffic meteorological disaster risk prediction model is evaluated.

[0003] However, the above-mentioned scheme cannot be adjusted according to real-time meteorological data, so there is a difference in the accuracy of real-time perception of meteorological risk. SUMMARY

[0004] The purpose of the present application is to provide a highway traffic meteorological risk early warning method and system, which realizes the closed loop of accurate perception-real-time analysis-efficient decision-making-flexible release of highway meteorological risk, significantly improves the scientificity and response efficiency of traffic management under adverse weather, and has wide application prospect in the fields of intelligent transportation and emergency management.

[0005] To achieve the above-mentioned purpose, the present application provides a highway traffic meteorological risk early warning method, comprising the following steps:

[0006] Obtain and preprocess multi-modal data, including satellite remote sensing data, radar-based data and real-time lightning positioning data;

[0007] Divide the highway into road sections, divide the area near the highway into grids according to the road sections, obtain the relevant multi-modal data of the grids in the nearby area, and analyze the obtained data to obtain the meteorological conditions of each road section;

[0008] According to the meteorological conditions of each road section, the risk type is identified and the grade is divided, and the risk early warning report is issued.

[0009] Preferably, the multi-modal data is preprocessed, comprising the following steps:

[0010] Introduce Kalman filtering to align satellite remote sensing data, radar-based data and real-time lightning positioning data of different resolutions in time and space;

[0011] Synthetic data for missing regions is generated using GANs.

[0012] Increasing ground weather station data as calibration benchmark, reducing remote sensing inversion error.

[0013] Preferably, the highway is divided into road sections, and the area near the highway is divided into grids according to the road sections, and the relevant multi-modal data of the grids in the area near the highway are obtained, and the weather conditions of each road section are obtained by analyzing the obtained data, including the following steps:

[0014] The area near the highway is divided into grids, which are divided into m layers of grids according to the distance from the highway;

[0015] Obtain radar-based data and real-time lightning positioning data in the grid, and analyze the radar-based data and real-time lightning positioning data to obtain first weather data in each grid;

[0016] The first weather data of each grid is taken as a node, and a topological structure is established according to the relationship between the first weather data of each grid, wherein the weight between nodes is adaptively adjusted according to the first weather data;

[0017] Construct a spatial index of the node, and the spatial index is the level of the grid where the node is located, and the spatial index of the node is represented as , wherein ;

[0018] Using the spatial index and topological structure of the node, analyze the weather conditions of each road section .

[0019] Preferably, the weight between nodes is adaptively adjusted according to the first weather data, including the following steps:

[0020] Calculate the absolute difference value of the first weather data of two nodes, and normalize the absolute difference value;

[0021] Combine multiple indexes in the normalized absolute difference value into a single index through principal component analysis, which is a comprehensive difference index;

[0022] Use the function of the node and weather similarity to position the weight between nodes, which is represented as:

[0023] ;

[0024] Wherein, k is an adjustment parameter, is the comprehensive difference index between nodes, the smaller the difference, the closer the weight to 1; the larger the difference, the weight tends to 0;

[0025] Set a difference threshold θ, if If θ, consider the weak association between nodes, give low weight; otherwise, according to the node and The function of the meteorological similarity calculates the weight, and the threshold θ is dynamically adjusted according to the historical data distribution;

[0026] In order to reduce the interference of historical data on weight, a forgetting factor λ is introduced, 0<λ<1, so that the contribution of new data is higher, and when updating the weight:

[0027] ;

[0028] Where, The difference function at the current time, λ controls the weight ratio of new and old data;

[0029] When the first meteorological data mutation is monitored, the weight recalculation is triggered, and the update frequency is increased;

[0030] According to the weight size, the effective connection edge is screened, and the edge with >μ is retained, and the weakly associated edge is removed; When the weight changes due to node data changes, dynamically add or delete edges; Wherein, μ is the global weight average;

[0031] According to the weight interval, the node level is divided, the high weight node constitutes the core layer, and the low weight node is used as the peripheral layer; In extreme weather, the weight threshold of the core layer is dynamically lowered to include more nodes.

[0032] Preferably, according to the meteorological conditions of each road section, the risk type is identified and classified, and the risk warning report is issued, including the following steps:

[0033] The meteorological conditions of each road section are identified to obtain the risk type, and the risk type includes precipitation, wind speed, snowfall, hail, visibility and ground icing;

[0034] According to the risk type, set the first risk threshold, dynamically classify the level according to the first risk threshold, duration and influence range, issue risk warning report of different levels, and take corresponding measures;

[0035] Set the second risk threshold, when a risk type exceeds the second risk threshold, a pop-up window is directly generated to prompt and lock the road section.

[0036] Preferably, the risk warning report includes the following forms:

[0037] Each road section is marked as p, and the attribute information P of the midpoint of each road section is edited, which contains the starting point and terminal information of the road, the coding of each road section, and the province, city and county information of each road section;

[0038] According to the route release, the route information and attribute information are screened and spliced to obtain a complete route and a corresponding meteorological risk type;

[0039] According to the marked area release, the key area is marked, the information of the key area is compared and screened with the attribute information, the corresponding road section of the key area is obtained, and the meteorological risk type of the key area is obtained by splicing the screened road section.

[0040] A highway traffic meteorological risk early warning system comprises

[0041] A data processing module is configured to acquire and preprocess multi-modal data.

[0042] A meteorological analysis module is configured to divide a highway into road sections, divide a region near the highway into grids according to the road sections, acquire relevant multi-modal data of the grids in the region, and analyze the acquired data to obtain meteorological conditions of each road section.

[0043] A risk report generation module is configured to identify a risk type and divide a level according to the meteorological conditions of each road section, and release a risk early warning report.

[0044] Therefore, the highway traffic meteorological risk early warning method and system have the following technical effects:

[0045] Similarity-driven node association: based on a comprehensive difference index of principal component analysis (PCA), the meteorological similarity between grids is quantified, a dynamic topology structure is constructed, and redundant calculation is reduced.

[0046] Forgotten factor (lambda): weaken the interference of historical data, strengthen the weight of new data, and improve the response speed of the model to sudden weather changes (such as thunderstorms and sudden rain).

[0047] Difference threshold (theta) and dynamic adjustment: through theta, strong correlation nodes are screened, combined with mu (global weight average), weak connections are dynamically removed, the network structure is optimized, and noise interference is avoided.

[0048] Extreme weather adaptability: in the case of heavy rain, snowstorm and other scenes, the core layer threshold is dynamically adjusted downward, the range of associated nodes is expanded, and the system robustness under complex weather is enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 The flowchart of the highway traffic meteorological risk early warning method of the present application. DETAILED DESCRIPTION

[0050] The technical solutions of the present application are further described below by means of the accompanying drawings and examples.

[0051] Unless otherwise defined, technical terms or scientific terms used in the present application shall have the meanings that are commonly understood by a person of ordinary skill in the art to which the present application belongs.

[0052] Embodiment one

[0053] As Figure 1 shown, a highway traffic weather risk early warning method, comprising the following steps:

[0054] Obtain and preprocess multi-modal data, including satellite remote sensing data, radar-based data and real-time lightning location data, preprocess the multi-modal data, including the following steps:

[0055] Preprocessing of satellite remote sensing data: radiation correction, geometric registration, cloud detection and removal, NDVI / NDWI index calculation, to obtain the surface temperature, albedo, snow / permafrost coverage.

[0056] Preprocessing of radar-based data: noise filtering, velocity de-blinking, polar coordinate conversion to plane coordinate, QPE (quantitative precipitation estimation), to obtain minute-level precipitation intensity, wind field vector, storm tracking results.

[0057] Preprocessing of real-time lightning location data: spatiotemporal clustering (eliminate isolated points), lightning classification (cloud flash / ground flash), density gridding, to obtain lightning occurrence frequency, positive / negative flash ratio, three-dimensional lightning location heat map.

[0058] Introduce Kalman filter to align satellite remote sensing data, radar-based data and real-time lightning location data in time and space with different resolutions;

[0059] Time alignment: take radar data timestamp as reference, satellite data downsample to minute level, lightning data aggregate to 5-minute window.

[0060] Spatial alignment: unify projection to WGS84 coordinate system, resample to 100m x 100m grid.

[0061] Use GANs to generate synthetic data for missing areas.

[0062] Add ground weather station data as calibration reference to reduce remote sensing inversion error.

[0063] Divide the highway into sections, divide the area near the highway into grids according to the section, obtain the relevant multi-modal data of the grid in the nearby area, and analyze the obtained data to obtain the weather conditions of each section, including the following steps:

[0064] Divide the area near the highway into grids, divide it into m layers of grids according to the distance from the highway; grid level attributes: (0-1km, core influence area), (1-2km),..., (> (m-1) km).

[0065] Obtaining radar-based data and real-time lightning positioning data within the grid, and analyzing the radar-based data and the real-time lightning positioning data to obtain first meteorological data within each grid;

[0066] Taking the first meteorological data of each grid as a node, and establishing a topological structure according to the relationship between the first meteorological data of each grid, wherein the weight between the nodes is adaptively adjusted according to the first meteorological data;

[0067] Constructing a spatial index of the node, the spatial index being a level of a grid where the node is located, and the spatial index of the node being represented as , wherein, ;

[0068] Using the spatial index of the node and the topological structure to analyze a meteorological condition of each road section .

[0069] The weight between the nodes is adaptively adjusted according to the first meteorological data, including the following steps:

[0070] Calculating an absolute difference value of the first meteorological data of two nodes, and performing normalization processing on the absolute difference value;

[0071] Combining multiple indexes in the normalized absolute difference value into a single index by principal component analysis, that is, a comprehensive difference index;

[0072] Positioning the weight between the nodes by using a function of meteorological similarity between the nodes, and the weight being represented as:

[0073] ;

[0074] wherein k is an adjustment parameter, is the comprehensive difference index between the nodes, the smaller the difference is, the closer the weight is to 1; the larger the difference is, the weight tends to 0;

[0075] Setting a difference threshold θ, if > θ, it is considered that the correlation between the nodes is weak, and a low weight is given; otherwise, the weight is calculated according to the function of meteorological similarity between the nodes The threshold value θ is dynamically adjusted according to historical data distribution;

[0076] In order to reduce the interference of historical data on the weight, a forgetting factor λ is introduced, 0 < λ < 1, so that the contribution of new data is higher, and when updating the weight:

[0077] ;​​​

[0078] in, This represents the difference function at the current moment, where λ controls the weight ratio between the new and old data.

[0079] When a sudden change in the first meteorological data is detected, the weights are recalculated and the update frequency is increased;

[0080] Valid connection edges are filtered based on their weights, and those that remain are retained. > μ Weakly related edges are removed, and edges are dynamically added or deleted when changes in node data cause changes in weights; where μ is the global weight mean.

[0081] The nodes are divided into layers based on weight ranges, with high-weight nodes forming the core layer and low-weight nodes forming the outer layer. In extreme weather conditions, the weight threshold of the core layer is dynamically lowered to include more nodes.

[0082] Based on the meteorological conditions of each road section, risk types are identified and levels are classified, and risk warning reports are issued, including the following steps:

[0083] Weather conditions for each section of the road The risk types are identified, including precipitation, wind speed, snowfall, hail, visibility, and ground icing.

[0084] A first risk threshold is set according to the risk type, and the risk level is dynamically classified according to the first risk threshold, duration and scope of impact. Risk warning reports of different levels are issued and corresponding measures are taken.

[0085] Set a second risk threshold. When a certain risk type exceeds the second risk threshold, a pop-up message will be generated and the road segment will be blocked.

[0086] Risk warning reports can take the following forms:

[0087] Each road segment is marked as p. The midpoint of each road segment is edited with its attribute information P. The attribute information includes the start and end points of the highway, the code of each road segment, and the province, city, district and county where each road segment is located.

[0088] Based on the published driving route, each road segment is filtered and pieced together according to the driving route information and attribute information to obtain the complete route and its corresponding meteorological risk type;

[0089] According to the marked areas, key areas are marked, and the information and attribute information of key areas are compared and filtered to obtain the road segments corresponding to the key areas. The filtered road segments are spliced ​​together to obtain the meteorological risk type corresponding to the key areas.

[0090] Table 1 Risk Types and Risk Levels

[0091] ;

[0092] A highway traffic meteorological risk early warning system comprises

[0093] A data processing module is configured to acquire and preprocess multi-modal data.

[0094] A meteorological analysis module is configured to divide a highway into road segments, divide a region near the highway into grids according to the road segments, acquire relevant multi-modal data of the grids in the region, and analyze the acquired data to obtain meteorological conditions of each road segment.

[0095] A risk report generation module is configured to identify a risk type and divide a risk level according to the meteorological conditions of each road segment, and publish a risk early warning report.

[0096] Therefore, the highway traffic meteorological risk early warning method and system can realize precise perception-real-time analysis-efficient decision-making-flexible publishing of a closed loop of highway meteorological risks, significantly improve the scientific nature and response efficiency of traffic management under adverse weather, and have wide application prospects in the fields of intelligent traffic and emergency management.

[0097] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application but not to limit it, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that: it can still modify or equivalently replace the technical solutions of the present application, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. A method for early warning of traffic meteorological risks on highways, characterized in that, Includes the following steps: Acquire and preprocess multimodal data, including satellite remote sensing data, radar-based data, and real-time lightning location data; The highway is divided into road segments, and the area near the highway is divided into grids according to the road segments. Relevant multimodal data of the grids in the vicinity are obtained, and the obtained data is analyzed to obtain the meteorological conditions of each road segment, including the following steps: The area near the highway is divided into grids, with each grid layer consisting of m layers based on its distance from the highway. Acquire radar base data and real-time lightning location data within the grid, and analyze the radar base data and real-time lightning location data to obtain the first meteorological data within each grid; The first meteorological data of each grid is taken as a node. A topology structure is established based on the relationship between the first meteorological data of each grid. The weights between nodes are adaptively adjusted according to the first meteorological data. Construct the spatial index of the node, where the spatial index is the level of the grid in which the node resides. The spatial index of the node is then represented as: ; By utilizing the spatial index and topology of nodes, the meteorological conditions of each road segment are analyzed. ; The weights between nodes are adaptively adjusted based on the first meteorological data, including the following steps: Calculate the absolute difference between the first meteorological data of the two nodes, and then normalize the absolute difference. Principal component analysis combines multiple indicators in the normalized absolute difference into a single indicator, which is the comprehensive difference indicator. Utilizing nodes and The function of meteorological similarity determines the weights between nodes, expressed as: ； Where k is the adjustment parameter, This is a comprehensive difference index between nodes. The smaller the difference, the closer the weight is to 1; the larger the difference, the closer the weight is to 0. Set a difference threshold θ, if If the value is greater than θ, the nodes are considered to have weak connections and are assigned low weights; otherwise, they are assigned weights based on the nodes. and The weights of the meteorological similarity function are calculated, and the threshold θ is dynamically adjusted based on the distribution of historical data. To reduce the interference of historical data on the weights, a forgetting factor is introduced. ,0< <1 makes new data contribute more, so when updating weights: ； in, The difference function represents the difference at the current time. Control the weighting ratio between new and old data; When a sudden change in the first meteorological data is detected, the weights are recalculated and the update frequency is increased; Filter valid connection edges based on their weights and retain them. Edges are processed by removing weakly related edges, and edges are dynamically added or deleted when changes in node data cause changes in weights; among them... The global weighted average; The nodes are divided into layers based on weight ranges, with high-weight nodes forming the core layer and low-weight nodes forming the outer layer. Under extreme weather conditions, the weight threshold of the core layer is dynamically lowered to include more nodes. Risk types and levels are identified and classified based on the weather conditions of each road section, and risk warning reports are issued.

2. The method for early warning of highway traffic meteorological risks according to claim 1, characterized in that, Preprocessing multimodal data includes the following steps: Kalman filtering is introduced to align satellite remote sensing data, radar base data, and real-time lightning location data of different resolutions in time and space; Use GANs to generate synthetic data for missing regions; Adding ground meteorological station data as a calibration benchmark reduces remote sensing inversion errors.

3. The method for early warning of highway traffic meteorological risks according to claim 1, characterized in that, Based on the meteorological conditions of each road section, risk types are identified and levels are classified, and risk warning reports are issued, including the following steps: Weather conditions for each section of the road The risk types are identified, including precipitation, wind speed, snowfall, hail, visibility, and ground icing. A first risk threshold is set according to the risk type, and the risk level is dynamically classified according to the first risk threshold, duration and scope of impact. Risk warning reports of different levels are issued and corresponding measures are taken. Set a second risk threshold. When a certain risk type exceeds the second risk threshold, a pop-up message will be generated and the road segment will be blocked.

4. The method for early warning of highway traffic meteorological risks according to claim 3, characterized in that, Issue a risk warning report Including the following forms: Each road segment is marked as p. The midpoint of each road segment is edited with its attribute information P. The attribute information includes the start and end points of the highway, the code of each road segment, and the province, city, district and county where each road segment is located. Based on the published driving route, each road segment is filtered and pieced together according to the driving route information and attribute information to obtain the complete route and its corresponding meteorological risk type; According to the marked areas, key areas are marked, and the information and attribute information of key areas are compared and filtered to obtain the road segments corresponding to the key areas. The filtered road segments are spliced ​​together to obtain the meteorological risk type corresponding to the key areas.

5. A highway traffic meteorological risk early warning system, characterized in that, include The data processing module is used to acquire and preprocess multimodal data; The meteorological analysis module is used to divide the highway into road segments, divide the area near the highway into grids according to the road segments, obtain relevant multimodal data of the grids in the nearby area, and analyze the obtained data to obtain the meteorological conditions of each road segment. The risk report generation module is used to identify the risk type and classify the risk level based on the weather conditions of each road segment, and to issue a risk warning report.

Images