A highway traffic data visual analysis system

Abstract

The application discloses a highway traffic data visual analysis system, which builds a special space-time grid model, discards the whole road section statistical mode, combines road topology to adaptively divide differentiated precision grids, disassembles a road network into fine analysis units, solves the problem that a traditional scheme is coarse in granularity and difficult to distinguish local traffic flow differences, extracts three indexes of traffic flow density, passing speed and congestion duration to construct a multi-dimensional feature vector, avoids one-sidedness of single speed judgment, and completely covers congestion full-cycle characteristics; in addition, a situation entropy change algorithm is introduced to calculate grid comprehensive entropy values, converts fuzzy congestion nature judgment into quantifiable comparison values, accurately measures road network passing disorder degree, analyzes time sequence change law by matching a graded threshold value according to the corrected entropy values, can deduce traffic flow aggregation, congestion spread and dissipation trends in advance, realizes prospective congestion prediction, and makes up for the short board that existing technologies can only identify real-time congestion and cannot predict situation evolution.

Description

Technical Field

[0001] This invention relates to the field of traffic data analysis technology, and in particular to a highway traffic data visualization and analysis system. Background Technology

[0002] With the continuous expansion of the national expressway network, the number of sensing terminals such as vehicle detectors, video surveillance, weather stations, and electromechanical maintenance equipment deployed on the roads has increased significantly. During the operation of the road network, massive amounts of heterogeneous data of various types are generated simultaneously, including traffic flow data, vehicle speed monitoring data, road surface video streams, road section meteorological data, and electromechanical equipment maintenance status data, etc. However, the existing expressway traffic monitoring and analysis schemes have several obvious shortcomings in data processing and situation assessment: (1) Poor spatial matching accuracy of traffic data and lack of refined spatiotemporal grid modeling methods: Existing technologies mostly use the entire road section as the smallest statistical unit, which can only output the overall average traffic index of the road section and cannot distinguish the local traffic flow differences of different road sections such as main lines, ramps, and interchanges; without combining road network expansion (1) The construction of a refined spatiotemporal grid model is not possible. Discrete traffic data cannot be accurately bound to the corresponding road spatial units, and the refined quantitative analysis of traffic status in local areas of the road cannot be realized. The spatial analysis granularity is coarse. (2) The traffic congestion situation prediction method is simple and the quantitative judgment reliability is low: The current mainstream congestion judgment method only relies on the instantaneous vehicle speed threshold to simply divide the smooth, slow and congested states. It only relies on a single time series cross section data to make qualitative judgments, without comprehensively considering multiple dimensions of indicators such as traffic density, traffic speed and congestion duration. There is a lack of quantitative judgment model based on information entropy. It is impossible to quantify the degree of disorder in the local traffic order of the road network. It also does not combine spatial neighborhood and time series historical data to correct the judgment results. Instantaneous abnormal data is prone to cause prediction distortion and it is difficult to predict the trend of congestion gathering and spread in advance. Summary of the Invention

[0003] In view of this, the present invention proposes a highway traffic data visualization and analysis system, which can effectively solve the shortcomings of existing technologies, such as the lack of refined spatiotemporal grid modeling methods and the low reliability of quantitative analysis.

[0004] The technical solution of this invention is implemented as follows:

[0005] A highway traffic data visualization and analysis system, comprising:

[0006] The multi-source data acquisition module is used to collect real-time traffic flow data, vehicle speed data, road condition video data, meteorological data, and equipment operation and maintenance data on highways, forming a standardized raw traffic dataset.

[0007] The edge fusion processing module is used to perform heterogeneous data fusion on the standardized original traffic dataset to generate a structured traffic time-series dataset;

[0008] The spatiotemporal grid mapping analysis module is used to construct a spatiotemporal grid model based on the highway network topology, and to map the structured traffic time series dataset to the corresponding road network grid cells based on the spatiotemporal grid model to generate gridded spatiotemporal traffic data.

[0009] The situation entropy change inference module is used to extract the traffic flow density, traffic speed and congestion duration of each road network grid cell based on gridded spatiotemporal traffic data, and generate traffic situation prediction results by combining traffic flow density, traffic speed and congestion duration through the traffic situation entropy change algorithm.

[0010] The visual effects layer decoupling rendering module is used to connect the spatiotemporal grid mapping analysis module and the situation entropy change inference module. It performs layer decoupling and splitting of gridded spatiotemporal traffic data and traffic situation prediction results to complete multi-dimensional visualization rendering output.

[0011] As a further optional solution to the aforementioned highway traffic data visualization and analysis system, the spatiotemporal grid mapping analysis module includes:

[0012] The road network topology parsing unit is used to import highway network vector data and extract road topology features.

[0013] The grid scale adaptive partitioning unit is used to dynamically adjust the grid precision based on the extracted road topology features, combined with road grade and historical traffic activity, to divide the road network grid into differentiated grids.

[0014] The spatiotemporal coordinate registration unit is used to match and calibrate the collection points of the structured traffic time series dataset with the spatial coordinates of the road network grid.

[0015] The data raster binding unit is used to associate and mount the calibrated traffic time-series data to the corresponding road network raster, and establish a raster-time-series data binding relationship;

[0016] The temporal gridded output unit is used to integrate traffic data after binding all road network grids, generate gridded spatiotemporal traffic data, and distribute them to the situation entropy change inference module and the visual effect level decoupling rendering module, respectively.

[0017] As a further optional solution to the aforementioned highway traffic data visualization and analysis system, the raster scale adaptive partitioning unit includes:

[0018] The road network level assessment subunit is used to identify the types of expressway mainlines, ramps, interchanges, and service area road sections based on the extracted road topology elements and match them with the corresponding basic grid reference scale.

[0019] The traffic flow activity measurement subunit is used to statistically analyze the traffic flow frequency, speed fluctuation range and congestion frequency of each road segment type based on historical and real-time structured traffic time series data, and quantify the traffic flow activity level of the road segment.

[0020] The dynamic grid granularity assignment sub-unit is used to perform a two-factor coupled determination based on the road segment network level and the road segment traffic activity level, and to assign grids of different scales to each road segment to complete the differentiated grid division of the entire road network.

[0021] As a further optional solution to the aforementioned highway traffic data visualization and analysis system, the situation entropy change inference module includes:

[0022] The feature parameter extraction unit is used to extract the core parameters of traffic flow density, traffic speed and congestion duration of each road network grid cell from the gridded spatiotemporal traffic data, and to construct a single grid traffic state feature vector.

[0023] The parameter normalization processing unit is used to perform dimensionless normalization processing on the single-grid traffic state feature vector to obtain the normalized single-grid traffic state feature vector.

[0024] The grid situation entropy calculation unit is used to calculate the distribution probability of traffic state parameters of each grid unit based on the normalized single grid traffic state feature vector, and to calculate the single grid traffic situation entropy value, thereby quantifying the degree of local traffic disorder in the road network.

[0025] The spatiotemporal entropy change correction unit is used to combine the spatial correlation and temporal continuous change characteristics of adjacent road network grids to perform spatiotemporal smoothing correction on the traffic situation entropy value of a single grid, so as to obtain the corrected real-time situation entropy value.

[0026] The situation prediction and simulation unit is used to compare the corrected real-time situation entropy value with the preset classification threshold, analyze the evolution law of entropy value rise and fall, deduce the trend of traffic flow fluctuation, congestion accumulation and spread, and generate the final traffic situation prediction result.

[0027] As a further optional solution to the aforementioned highway traffic data visualization and analysis system, the grid situation entropy calculation unit specifically includes:

[0028] The parameter interval determination subunit is used to receive the normalized single-grid traffic state feature vector and determine the value range of three types of feature parameters: traffic density, traffic speed, and congestion duration.

[0029] The state probability statistics subunit is used to calculate the distribution probability of each feature parameter in the corresponding value range, and obtain the single-parameter state distribution probability.

[0030] The single-dimensional entropy calculation sub-unit is used to calculate the independent dimension entropy values ​​corresponding to the three types of feature parameters based on the single-parameter state distribution probability.

[0031] The weighted entropy fusion subunit is used to perform weighted fusion calculation of the entropy values ​​of each independent dimension according to the influence weight of each feature parameter on the road network traffic situation, so as to obtain the comprehensive situation entropy value of a single grid.

[0032] The situation quantization output sub-unit is used to quantify the degree of traffic order disorder of the road network grid based on the comprehensive situation entropy value of a single grid, and output the quantized situation entropy value of a single grid.

[0033] As a further optional solution to the aforementioned highway traffic data visualization and analysis system, the spatiotemporal entropy change correction unit includes:

[0034] The entropy data receiving subunit is used to receive the single-grid traffic situation entropy value of each independent road network grid output by the grid situation entropy value calculation unit.

[0035] The spatial neighborhood smoothing correction sub-unit is used to select the upstream and downstream adjacent grids and the left and right parallel grids of the target grid, and to carry out spatial dimension correction through the neighborhood entropy mean smoothing algorithm, calculate the spatial correction coefficient, and complete the entropy value spatial smoothing.

[0036] The temporal fluctuation filtering correction subunit is used to retrieve the historical temporal entropy value sequence of the current grid, perform temporal dimension correction using a temporal sliding filter, calculate the temporal correction coefficient, and filter out instantaneous entropy value fluctuations.

[0037] The spatiotemporal coupling correction subunit is used to fuse spatial correction coefficients and temporal correction coefficients to perform spatiotemporal dual-dimensional coupling correction on the single grid situation entropy value, so as to obtain the corrected real-time situation entropy value.

[0038] As a further optional solution to the aforementioned highway traffic data visualization and analysis system, the visual effects layer decoupling rendering module includes:

[0039] The multi-layer data access decoupling unit is used to receive gridded spatiotemporal traffic data and traffic situation prediction results respectively, and to classify and decouple the gridded spatiotemporal traffic data and traffic situation prediction results and remove redundant rendering fields.

[0040] The visualization hierarchy division unit is used to perform three-level visualization hierarchy division of gridded spatiotemporal traffic data and traffic situation prediction results according to the road network base layer, real-time traffic status layer, and situation prediction and early warning layer.

[0041] The differentiated visual effects rendering unit is used to configure corresponding color rules, transparency parameters and dynamic rendering styles for different levels of data, and to perform differentiated visual mapping for normal traffic, slow traffic, congestion and predicted situation.

[0042] The dynamic layer overlay unit is used to spatially register and overlay multiple independent rendering layers, and integrate them to generate a multi-layer overlay visualization layer.

[0043] The real-time output unit is used to continuously and dynamically refresh the multi-layered visualization layers, outputting a full-domain road network visualization that includes real-time traffic conditions and situational predictions.

[0044] A method for visualizing and analyzing highway traffic data, specifically including:

[0045] Real-time collection of highway traffic flow data, vehicle speed data, road condition video data, weather-related data, and equipment operation and maintenance data to form a standardized raw traffic dataset;

[0046] Heterogeneous data fusion is performed on the standardized raw traffic dataset to generate a structured traffic time-series dataset;

[0047] Based on the highway network topology, a spatiotemporal grid model adapted to road orientation is constructed. Based on the spatiotemporal grid model, the structured traffic time series dataset is spatially registered and temporally bound, and mapped one by one to the corresponding grid cells of the road network to generate gridded spatiotemporal traffic data.

[0048] Based on gridded spatiotemporal traffic data, the traffic flow density, traffic speed, and congestion duration of each road network grid unit are extracted. Then, the traffic situation prediction results are generated by combining the traffic flow density, traffic speed, and congestion duration through the traffic situation entropy change algorithm.

[0049] By integrating gridded spatiotemporal traffic data and traffic situation prediction results, layer-level decoupling and splitting are performed on the two types of data. Differentiated rendering rules are configured for each layer, and layer space registration and overlay are completed to achieve multi-dimensional visualization rendering output of highway traffic.

[0050] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described highway traffic data visualization and analysis method.

[0051] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described highway traffic data visualization and analysis method.

[0052] The beneficial effects of this invention are as follows: By building a dedicated spatiotemporal grid model based on the highway network topology, instead of using the entire road segment as the smallest statistical unit, it can adaptively divide the road network grid with differentiated precision according to different road topology elements such as main lines, ramps, interchanges, and service areas. This decomposes the road network space into the smallest analysis unit with controllable granularity, solving the problem of coarse analysis granularity and inability to distinguish traffic flow differences in local road segments in traditional solutions. Secondly, it simultaneously extracts three core indicators—traffic density, traffic speed, and congestion duration—within the grid unit to construct a multi-dimensional feature vector. This comprehensive assessment of the road network traffic status by integrating these three traffic elements avoids judgments based solely on vehicle speed. This approach comprehensively covers the entire lifecycle of congestion formation, persistence, and dissipation. Furthermore, by introducing a traffic situation entropy change algorithm to calculate the comprehensive situation entropy value of a single grid, the entropy value directly quantifies the degree of disorder in local road traffic order, transforming vague qualitative judgments of congestion into comparable and calculable quantitative values, thus achieving precise digital measurement of congestion status. In addition, based on matching hierarchical thresholds with the comprehensive situation entropy value of a single grid, the evolutionary pattern of entropy value fluctuations over time can be analyzed. This allows for the prediction of trends in traffic flow aggregation, congestion spread, and slow-moving traffic dissipation, enabling proactive congestion prediction and addressing the limitation of existing technologies that can only identify current congestion but cannot predict future trends. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a schematic diagram of the composition of a highway traffic data visualization and analysis system according to the present invention;

[0055] Figure 2 This is a flowchart illustrating a method for visualizing and analyzing highway traffic data according to the present invention.

[0056] Figure 3 This is a schematic diagram of the composition of a computing device according to the present invention. Detailed Implementation

[0057] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] refer to Figures 1 to 3A highway traffic data visualization and analysis system includes a multi-source data acquisition module, an edge fusion processing module, a spatiotemporal grid mapping analysis module, a situational entropy change inference module, and a visual effects hierarchy decoupling rendering module, wherein:

[0059] The multi-source data acquisition module is used to collect real-time traffic flow data, vehicle speed data, road condition video data, meteorological data, and equipment operation and maintenance data on highways, forming a standardized raw traffic dataset.

[0060] Specifically, traffic flow and speed data acquisition and analysis: Through RS485 bus and Ethernet dual-channel connection to the coil vehicle detectors and millimeter-wave radar sensing equipment deployed along the entire line, a fixed sampling period of 10 seconds is set to continuously collect raw binary messages of road section traffic flow, instantaneous vehicle speed, lane occupancy, and vehicle type classification; the module is pre-integrated with private protocol parsing plugins of multiple manufacturers such as Hikvision, Dahua, and Qianfang, which automatically identify the device message header identifiers, remove invalid check fields, and extract valid numerical information of traffic flow, vehicle speed, lane number, and device location number.

[0061] Road condition video data retrieval and slicing: Real-time video streams from high-definition road cameras along the route are retrieved via the RTSP streaming media protocol. A single frame of road surface image is automatically captured every 30 seconds as a static road condition sample, and the camera location ID, capture UTC timestamp, and video encoding format metadata are bound simultaneously. For continuously outputting long video streams, a 60-second segmentation slicing strategy is adopted to store video segments in sections, avoiding problems such as stuttering and parsing failures during the transmission of large video files.

[0062] Real-time capture of meteorological data related to road sections: The module connects to small automatic weather stations deployed along the route via the MQTT protocol. Each weather station pushes a set of road environment monitoring reports every minute. The module analyzes five meteorological indicators in the reports in real time, including road surface humidity, road visibility, instantaneous wind speed, rainfall, and road surface freezing point temperature, and simultaneously binds the corresponding road network spatial coordinate information of the weather station.

[0063] Electromechanical equipment operation and maintenance data interface polling collection: Connect to the open API interface of the high-speed electromechanical equipment operation and maintenance management platform through HTTP timed polling. The polling period is set to 5 minutes to batch obtain the online status, fault code, power supply voltage, historical maintenance records, and maintenance point information of all front-end sensing devices such as cameras, vehicle detectors, and weather stations, and to fully collect the operation and maintenance data of the entire life cycle of the equipment.

[0064] Multi-source data is uniformly standardized: Each data point is uniformly appended with a globally unique road network location ID and a standard UTC timestamp, and a unified spatiotemporal identification format is used; numerical data such as traffic flow, vehicle speed, temperature, and humidity are uniformly retained to two decimal places; video and image data are uniformly transcoded into the H.264 universal encoding format; fault and maintenance text information are uniformly mapped to the built-in fault code dictionary, and text fields are standardized; null values, garbled characters, and out-of-range dirty data generated by offline equipment are automatically identified and removed, and invalid noise data is filtered out; after all the standardization is completed, the five types of standardized data are integrated to form a standardized raw traffic dataset in a unified format, which is pushed to the downstream edge fusion processing module through a message queue.

[0065] In this way, by simultaneously collecting four types of multi-dimensional information—vehicle flow and speed, road surface video, road section weather, and equipment operation and maintenance—it breaks through the limitation of traditional data collection systems that only collect traffic flow data. It can simultaneously record the basic conditions of traffic flow that cause congestion, environmental factors such as rain and snow and low visibility, and equipment failure data, providing a complete and multi-dimensional original data foundation for subsequent heterogeneous data fusion, root cause analysis of road network congestion, and situational entropy variable analysis. Secondly, all collected data are uniformly attached with a unique road network location ID and a standard timestamp, establishing spatial and temporal correlations from the data layer. The downstream spatiotemporal grid mapping analysis module can directly complete the data and road network grid coordinate registration based on the location ID, avoiding data misalignment and matching deviation problems caused by the lack of spatial identifiers in the original data, and solving the defects of poor spatial matching accuracy of existing traffic data from the source.

[0066] The edge fusion processing module is used to perform heterogeneous data fusion on the standardized original traffic dataset to generate a structured traffic time-series dataset.

[0067] Specifically, the system receives standardized raw traffic datasets in real time via a distributed message queue. An internal data verification unit performs multi-dimensional compliance verification: based on the dual identifier matching verification of road network location ID and UTC standard timestamp, it directly filters out incomplete and invalid data with missing spatial identifiers and timestamps; it configures reasonable industry threshold ranges for numerical fields such as traffic flow, vehicle speed, rainfall, and visibility, and automatically removes extreme abnormal data that exceeds the reasonable range of road network operation; it verifies the encoding integrity of road condition videos and captured image files, discarding damaged media files caused by transmission interruptions or missing data packets; after completing the verification, it selects compliant and valid raw data and transmits it to the noise filtering unit for noise reduction processing.

[0068] The noise filtering unit is configured with dedicated noise reduction strategies for five types of data sources, removing data noise in layers: Traffic flow and vehicle speed time-series data: a 3rd-order moving average filtering algorithm is used to eliminate impulse noise generated by instantaneous jumps from millimeter-wave radar and loop detectors; Road condition capture video frames: an image difference comparison algorithm is used to remove blurry and invalid capture images caused by strong light or rain and snow; Meteorological monitoring data: a 5-minute continuous observation judgment mechanism is set up, and meteorological values ​​with only single sudden changes and no continuous evidence are judged as noise and removed; Equipment operation and maintenance text data: matching the built-in fault code dictionary, deleting garbled text information with no corresponding code or incorrect format; After noise reduction, clean and well-organized multi-source data is output and sent to the time-series alignment unit to unify the time dimension.

[0069] The time-series alignment unit sets 10 seconds as the globally unified time-series slice granularity, and completes the time-axis alignment operation of all data based on the UTC standard timestamps inherent in all data: the traffic flow and speed data with a sampling period of 10 seconds, the video capture data of 30 seconds, the meteorological monitoring data of 1 minute, and the operation and maintenance polling data of 5 minutes are uniformly mapped to the 10-second standard time-series slice; a dual primary key index is established using "road network location ID + time-series slice timestamp" to collect and bind the five types of data under the same space and time series; for short-term time-series data gaps, a linear interpolation algorithm is used to fill in the missing indicators; and the time-series misalignment problem caused by inconsistent sampling periods of different sensing devices is eliminated, realizing multi-dimensional data time synchronization under the same road segment and the same time.

[0070] The heterogeneous data fusion unit relies on the dual primary keys of location ID and time-series slice to achieve integrated fusion of multiple types of heterogeneous data: traffic numerical indicators such as traffic density and traffic speed, meteorological environmental indicators such as rainfall and visibility, and operation and maintenance indicators such as equipment online status and fault codes are integrated into the same structured array; road condition videos and captured images only store file index paths, without repeatedly storing large-capacity media files, reducing the local storage pressure on edge nodes; a data association mapping relationship is established, and meteorological environmental parameters and equipment fault status are used as auxiliary association fields of traffic operation indicators to achieve deep binding between traffic congestion causes and traffic status data; after fusion, a unified association data table is generated and sent to the regularization output unit.

[0071] The output unit uniformly encapsulates the fused related data tables to generate a standardized structured traffic time-series dataset: it adopts a unified JSON structured storage format and divides the dataset into four fixed partitions: traffic basic fields, meteorological related fields, equipment operation and maintenance fields, and media index fields; it retains the global road network point ID and time-series slice timestamp as the top-level core index fields to adapt to the coordinate registration and raster mounting requirements of the downstream spatiotemporal raster mapping analysis module; it performs lightweight compression processing on the dataset and pushes it to the spatiotemporal raster mapping analysis module through the edge local area network channel to complete the entire heterogeneous data fusion process.

[0072] Thus, through two pre-processing steps—intactness verification and layered differential noise filtering—incomplete, abrupt, and distorted dirty data are batch-removed, significantly reducing the inflow of abnormal noise data into the downstream spatiotemporal grid modeling and grid situation entropy calculation stages. This avoids single-point abnormal values ​​causing sudden changes in single-grid entropy values ​​and distortion of congestion prediction, effectively alleviating the shortcomings of existing technologies in predicting traffic congestion situations, such as their simplistic approach and low reliability of quantitative analysis. Secondly, addressing the industry challenge of inconsistent sampling frequencies for vehicle flow, video, weather, and maintenance equipment, time-series alignment is achieved with a uniform granularity of 10 seconds. This aggregates heterogeneous data from different periods into the same spatiotemporal slice, enabling the same local area of ​​the road network to be aligned with the same... Simultaneous matching of multi-dimensional data on traffic, environment, and equipment provides a complete and time-ordered data foundation for downstream refined spatiotemporal grid modeling, helping to improve the spatial matching accuracy between traffic data and road network grids. In addition, it abandons the traditional model of storing various types of data independently and without correlation, binding traffic flow and speed, road condition images, weather conditions, and equipment failure data to the same structured time-series dataset, and establishing a strong correlation between external factors causing congestion such as rain, snow, low visibility, and equipment failure and traffic operation indicators. The subsequent situational entropy change inference module can comprehensively quantify the degree of road network traffic disorder by integrating multi-dimensional factors, breaking through the limitations of traditional qualitative judgment of congestion based solely on a single indicator of instantaneous vehicle speed.

[0073] A spatiotemporal raster mapping analysis module is used to construct a spatiotemporal raster model based on the highway network topology, and to map structured traffic time-series datasets to corresponding road network raster cells based on the spatiotemporal raster model, generating gridded spatiotemporal traffic data; in some embodiments, the spatiotemporal raster mapping analysis module includes:

[0074] The road network topology parsing unit is used to import highway network vector data and extract road topology features.

[0075] The grid scale adaptive partitioning unit is used to dynamically adjust the grid precision based on the extracted road topology features, combined with road grade and historical traffic activity, to divide the road network grid into differentiated grids.

[0076] The spatiotemporal coordinate registration unit is used to match and calibrate the collection points of the structured traffic time series dataset with the spatial coordinates of the road network grid.

[0077] The data raster binding unit is used to associate and mount the calibrated traffic time-series data to the corresponding road network raster, and establish a raster-time-series data binding relationship;

[0078] The temporal gridded output unit is used to integrate traffic data after binding all road network grids, generate gridded spatiotemporal traffic data, and distribute them to the situation entropy change inference module and the visual effect level decoupling rendering module, respectively.

[0079] Specifically, the road network topology parsing unit imports standard SHP vector geographic data of the expressway network to fully parse the topological elements of the entire road network: distinguishing road type labels and identifying four types of road topological elements: mainline, interchange ramps, service area connecting lines, and toll station access roads; extracting static attributes for each road, including road grade, road centerline latitude and longitude coordinates, road segment start and end points, road length, and number of lanes; extracting the spatial adjacency relationship of the road network, marking the upstream and downstream connected grids and the left and right parallel adjacent road segments, and constructing a spatial association topology index for the road network; caching the parsed road topology elements to local memory and transferring them to the grid scale adaptive partitioning unit.

[0080] The raster scale adaptive division unit is based on road topology elements and dynamically adjusts the raster accuracy by coupling road level and historical traffic activity as two factors.

[0081] The spatiotemporal coordinate registration unit receives structured traffic time-series datasets and global raster spatial information, and performs coordinate matching calibration: it extracts the road network point IDs and sensing device latitude and longitude coordinates inherent in the structured traffic time-series data, and performs spatial overlay calculations with the spatial coordinate range of the raster units; it uses a planar Euclidean distance matching algorithm to bind each acquisition point to the road network raster unit with the nearest spatial distance, and records the corresponding mapping relationship between the point and the raster; for long-range millimeter-wave radar cross-section equipment deployed across raster units, it allocates the cross-section data to multiple adjacent raster units according to the radar coverage range weight, avoiding local data loss caused by binding the cross-section data to a single raster; it uniformly converts all raster units and acquisition points to the WGS84 standard geographic coordinate system to eliminate matching deviations caused by the inconsistency between the coordinate systems of vector data and equipment acquisition data; it outputs the point-raster mapping relationship after coordinate calibration and inputs it into the data raster binding unit.

[0082] The data raster binding unit, relying on the mapping relationship obtained by coordinate registration, attaches standardized time-series data to the corresponding raster: using "raster ID + 10s time-series slice timestamp" as the joint primary key, and reusing the unified UTC time-series slice identifier, it collects all dimensions of data such as traffic density, traffic speed, congestion duration, weather, and equipment operation and maintenance at the same time and location to the corresponding raster; it constructs a raster-specific structured data table, and each raster record synchronously stores spatial boundary coordinates, time-series identifier, traffic indicators, weather-related parameters, and equipment status index, forming a strong binding relationship between raster and time-series data; after all raster data is bound, it is pushed to the time-series gridded output unit.

[0083] The temporal gridded output unit integrates all time-series traffic data bound to all grids across the entire domain, completing a unified encapsulation and output: it is uniformly encapsulated as gridded spatiotemporal traffic data, with each data point carrying grid spatial attributes, a unified temporal identifier, and multi-dimensional traffic operation indicators, while fully preserving the meteorological and equipment-related fields from the previous fusion process; it splits into two data distribution channels: one pushes to the situation entropy change inference module for grid-level traffic situation entropy variable calculation; the other pushes to the visual effects layer decoupled rendering module as the basic data source for the visualization base map layer; it adopts a block compression transmission mechanism to distribute grid data in batches according to road areas, reducing the amount of data transmitted in a single batch and ensuring the real-time data reading efficiency of downstream modules.

[0084] Thus, abandoning the traditional single statistical unit model for the entire road segment, the grid size is dynamically adjusted based on road grade and traffic activity. High-congestion interchanges and ramps use ultra-fine precision grids, while remote, low-traffic sections use coarse-grained grids. This allows for precise differentiation of traffic flow differences between mainlines, ramps, and interchanges, refining the smallest analytical unit of traffic data to grids of hundreds or even tens of meters, significantly improving the matching accuracy between traffic data and road space, and enabling refined analysis of local areas of the road network. Secondly, a unified conversion to a standard geographic coordinate system is used, employing distance matching and cross-sectional weight allocation for dual calibration to resolve incompatibility issues between sensing points across grids and vector data coordinate systems. This ensures that each set of time-series traffic data is accurately mapped to the corresponding road space grid, addressing the pain point of insufficient spatial matching accuracy in existing technologies and providing a precise spatial basis for subsequent grid entropy quantification. Furthermore, using the grid as... The smallest unit aggregates indicators across all dimensions, including traffic density, traffic speed, congestion duration, weather, and equipment operation and maintenance. Each grid forms an independent and complete traffic state feature vector. The downstream situation entropy change inference module can directly extract multi-dimensional parameters from a single grid to calculate entropy values. Relying on refined grid units, the degree of traffic disorder can be independently quantified in different regions, breaking through the limitations of traditional qualitative judgment based on a single indicator for the entire road segment. This provides data support for multi-dimensional quantitative judgment and improves the reliability of congestion prediction. In addition, the encapsulated gridded spatiotemporal traffic data is synchronously distributed to the situation entropy change inference module and the visual effect layer decoupled rendering module. A set of standardized grid data simultaneously meets the business needs of both algorithm calculation and layer visualization. The grid comes with complete spatial coordinates and layered indicators. The visualization module can directly render the road network base map and traffic state layer layer by layer without repeating spatial matching and transformation, improving the efficiency of multi-dimensional visualization rendering.

[0085] In some embodiments, the raster scale adaptive partitioning unit includes:

[0086] The road network level assessment subunit is used to identify the types of expressway mainlines, ramps, interchanges, and service area road sections based on the extracted road topology elements and match them with the corresponding basic grid reference scale.

[0087] The traffic flow activity measurement subunit is used to statistically analyze the traffic flow frequency, speed fluctuation range and congestion frequency of each road segment type based on historical and real-time structured traffic time series data, and quantify the traffic flow activity level of the road segment.

[0088] The dynamic grid granularity assignment sub-unit is used to perform a two-factor coupled determination based on the road segment network level and the road segment traffic activity level, and to assign grids of different scales to each road segment to complete the differentiated grid division of the entire road network.

[0089] Specifically, the road network level assessment subunit receives the topological elements of the entire road network extracted by the road network topology analysis unit, and performs road segment classification and binding with the baseline grid scale:

[0090] Based on the road type labels within the topological elements, road segments are distinguished and the road network is divided into five categories: expressway mainline, interchange ramps, interchange areas, service area connecting lines, and toll station access roads.

[0091] Pre-configured basic grid reference sizes for various road sections: 50m for the main line of the expressway, 20m for the interchange ramps and toll station access roads, 10m for the core area of ​​the interchange, and 100m for the connecting lines of remote service areas.

[0092] Each road segment's topology element is assigned a road segment type identifier and a basic grid reference scale label, generating a road segment basic scale mapping table, which is then simultaneously distributed to the traffic flow activity calculation subunit.

[0093] The traffic flow activity measurement subunit retrieves a structured traffic time-series dataset, including historical time-series data from the past 7 days and real-time 10-second slice time-series data, to quantify traffic flow activity from multiple dimensions:

[0094] Indicator statistics: The statistics are divided into three core indicators: daily average traffic frequency, daily speed standard deviation fluctuation range, and weekly total congestion duration.

[0095] Dimensionless hierarchical processing: Normalize the three types of indicators to the range of 0 to 1 respectively, and sum them with weights of 0.4, 0.35 and 0.25 to obtain the comprehensive score of road segment activity;

[0096] The classification rules are as follows: a comprehensive score of ≥0.7 is considered high activity, 0.3~0.7 is medium activity, and <0.3 is low activity. A traffic activity level label is attached to each road segment, and the basic scale and activity level of the road segment are pushed to the dynamic grid granularity assignment sub-unit simultaneously.

[0097] The dynamic grid granularity assignment sub-unit integrates the road segment network level and traffic flow activity level as two factors for coupled judgment, dynamically corrects the actual grid size, and divides the entire road network into grids:

[0098] Preset two-factor coupling correction rule:

[0099] High-speed mainline + high activity: The baseline 50m is adjusted downward to a fine 20m grid; High-speed mainline + medium activity: The standard 50m grid is maintained; High-speed mainline + low activity: The coarse 80m grid is adjusted upward.

[0100] Interchange ramps / toll plaza approach roads + high activity level: the baseline 20m grid is lowered to a 10m ultra-fine grid; medium and low activity levels maintain the 20m baseline grid.

[0101] The core area of ​​the interchange uniformly adopts a 10m ultra-fine grid, which will not be adjusted according to the activity level;

[0102] The service area connection line's high activity level has been reduced to 50m, while the medium-low activity level remains at 100m with a coarse grid.

[0103] Cut the grid boundaries along the centerline of each road segment, ensuring that the grid boundaries strictly fit the lane range to avoid data misalignment caused by grids crossing road segments or lanes.

[0104] Each generated raster is assigned a globally unique raster ID, and the spatial latitude and longitude boundaries, road segment type, actual raster size, and corresponding activity level are recorded synchronously to generate a complete set of differentiated road network rasteres, which are then output to the downstream spatiotemporal coordinate registration unit.

[0105] Thus, by classifying road types such as highway mainlines, ramps, interchanges, and service areas through road network level analysis sub-units, differentiated basic grid benchmarks are configured for different functional road sections, aligning with the traffic flow distribution characteristics of different areas of the highway network. This addresses the shortcomings of traditional solutions that use a uniform grid size across the entire road section, resulting in coarse granularity in congested interchange areas and wasted computing power in remote sections. Secondly, relying on massive amounts of time-series traffic data, the activity level is quantified by comprehensively considering multiple dimensions such as traffic frequency, speed fluctuations, and congestion duration, rather than simply relying on static road level to divide the grid. The grid fineness can be dynamically adjusted according to real-time traffic changes on road sections, automatically switching to ultra-fine grids for congested road sections during peak hours and automatically enlarging the grid granularity for low-traffic road sections during off-peak hours, balancing local analysis accuracy with overall computing power consumption. Furthermore, differentiated grid sizes are generated by integrating both static road attributes and dynamic traffic flow status, with 10m grids used for high-congestion interchanges and ramps. The 20m ultra-fine grid refines traffic data to the smallest analytical unit at the local lane level, replacing the coarse mode of treating the entire road segment as a single statistical unit. Each grid corresponds to a precise local road space, and subsequent time-series data mounting and grid entropy calculation can accurately locate local congestion points, significantly improving the matching accuracy between traffic data and road network space, and enabling refined decomposition and quantitative analysis of local traffic conditions in the road network. In addition, the output differentiated fine grids can collect single-grid-specific traffic flow density, traffic speed, and congestion duration feature vectors in the subsequent data grid binding process. The downstream grid situation entropy calculation unit can independently quantify the degree of traffic disorder for each grid, and the spatiotemporal entropy change correction unit can accurately retrieve adjacent grids based on the subdivided grids for spatial smoothing correction, avoiding the masking of local traffic flow changes by large-granularity road segment average data, and effectively alleviating the problems of the single congestion prediction method and low reliability of quantitative analysis in existing technologies.

[0106] The situation entropy change inference module is used to extract traffic flow density, traffic speed, and congestion duration from each road network grid cell based on gridded spatiotemporal traffic data, and to generate traffic situation prediction results by combining traffic flow density, traffic speed, and congestion duration through a traffic situation entropy change algorithm; in some embodiments, the situation entropy change inference module includes:

[0107] The feature parameter extraction unit is used to extract the core parameters of traffic flow density, traffic speed and congestion duration of each road network grid cell from the gridded spatiotemporal traffic data, and to construct a single grid traffic state feature vector.

[0108] The parameter normalization processing unit is used to perform dimensionless normalization processing on the single-grid traffic state feature vector to obtain the normalized single-grid traffic state feature vector.

[0109] The grid situation entropy calculation unit is used to calculate the distribution probability of traffic state parameters of each grid unit based on the normalized single grid traffic state feature vector, and to calculate the single grid traffic situation entropy value, thereby quantifying the degree of local traffic disorder in the road network.

[0110] The spatiotemporal entropy change correction unit is used to combine the spatial correlation and temporal continuous change characteristics of adjacent road network grids to perform spatiotemporal smoothing correction on the traffic situation entropy value of a single grid, so as to obtain the corrected real-time situation entropy value.

[0111] The situation prediction and simulation unit is used to compare the corrected real-time situation entropy value with the preset classification threshold, analyze the evolution law of entropy value rise and fall, deduce the trend of traffic flow fluctuation, congestion accumulation and spread, and generate the final traffic situation prediction result.

[0112] In some embodiments, the grid situation entropy calculation unit specifically includes:

[0113] The parameter interval determination subunit is used to receive the normalized single-grid traffic state feature vector and determine the value range of three types of feature parameters: traffic density, traffic speed, and congestion duration.

[0114] The state probability statistics subunit is used to calculate the distribution probability of each feature parameter in the corresponding value range, and obtain the single-parameter state distribution probability.

[0115] The single-dimensional entropy calculation sub-unit is used to calculate the independent dimension entropy values ​​corresponding to the three types of feature parameters based on the single-parameter state distribution probability.

[0116] The weighted entropy fusion subunit is used to perform weighted fusion calculation of the entropy values ​​of each independent dimension according to the influence weight of each feature parameter on the road network traffic situation, so as to obtain the comprehensive situation entropy value of a single grid.

[0117] The situation quantization output sub-unit is used to quantify the degree of traffic order disorder of the road network grid based on the comprehensive situation entropy value of a single grid, and output the quantized situation entropy value of a single grid.

[0118] In some embodiments, the spatiotemporal entropy change correction unit includes:

[0119] The entropy data receiving subunit is used to receive the single-grid traffic situation entropy value of each independent road network grid output by the grid situation entropy value calculation unit.

[0120] The spatial neighborhood smoothing correction sub-unit is used to select the upstream and downstream adjacent grids and the left and right parallel grids of the target grid, and to carry out spatial dimension correction through the neighborhood entropy mean smoothing algorithm, calculate the spatial correction coefficient, and complete the entropy value spatial smoothing.

[0121] The temporal fluctuation filtering correction subunit is used to retrieve the historical temporal entropy value sequence of the current grid, perform temporal dimension correction using a temporal sliding filter, calculate the temporal correction coefficient, and filter out instantaneous entropy value fluctuations.

[0122] The spatiotemporal coupling correction subunit is used to fuse spatial correction coefficients and temporal correction coefficients to perform spatiotemporal dual-dimensional coupling correction on the single grid situation entropy value, so as to obtain the corrected real-time situation entropy value.

[0123] Specifically, the feature parameter extraction unit receives gridded spatiotemporal traffic data and extracts core parameters grid by grid, using a unique raster ID and a 10-second standard time-series slice as the retrieval key.

[0124] For each differentiated granularity road network grid, three core indicators are extracted from the structured time-series data bound to the grid: real-time traffic density within the grid (pcu / km), average cross-sectional traffic speed (km / h), and cumulative duration of continuous congestion (s).

[0125] Reuse the unified UTC time sequence identifier, collect the three sets of indicators of the same grid by single time sequence slice, and remove the null value missing records caused by device offline in the grid;

[0126] The extracted traffic density, traffic speed, and congestion duration are combined to construct a fixed-dimensional single-grid traffic state feature vector. ,in For traffic density, For traffic speed, To determine the congestion duration, the feature vectors of all grid cells are pushed in batches to the parameter normalization processing unit.

[0127] The parameter normalization unit performs a unified dimensionless transformation on three types of parameters with different dimensions and numerical ranges, eliminating the interference of magnitude differences on entropy calculation.

[0128] The minimum-maximum normalization algorithm is used to determine the global maximum and minimum values ​​of three types of indicators based on the full 7-day historical raster data of the entire road network: traffic density range [0,120], traffic speed range [0,120], and congestion duration range [0,600].

[0129] Normalization calculation formula: All indicators are uniformly mapped to the [0,1] interval;

[0130] Output the normalized and standardized feature vector To ensure that the three indicators of traffic flow density, vehicle speed, and congestion duration have equal calculation weights, they are fed into the grid situation entropy value calculation unit.

[0131] The grid situation entropy calculation unit incorporates parameter interval determination, probability statistics, one-dimensional entropy solving, and weighted fusion sub-units, and completes quantization calculation based on normalized feature vectors.

[0132] Parameter interval partitioning: The normalized interval [0,1] is uniformly divided into 5 discrete state intervals. The frequency of each parameter falling into different intervals within the past 10 minutes of a single grid is counted, and the state distribution probability of each parameter type is calculated. ;

[0133] Solving for one-dimensional information entropy: using the Shannon information entropy formula. Solve for the traffic flow density entropy respectively. Traffic rate entropy Congestion duration entropy ;

[0134] Weighted fusion: Preset weight coefficients , , Calculate the overall situation entropy value of a single grid cell: ;

[0135] Entropy value physical definition: The higher the comprehensive entropy value, the more intense the traffic flow fluctuation and the higher the degree of traffic disorder in the grid; the original single grid situation entropy value is output and transmitted to the spatiotemporal entropy change correction unit.

[0136] The spatiotemporal entropy change correction unit relies on the pre-stored road network grid adjacency topology to perform smoothing and noise reduction correction on the original entropy value:

[0137] Spatial neighborhood smoothing correction: Search for upstream and downstream, left and right parallel adjacent grids of the target grid, and take the average entropy value of the neighboring grids as the spatial correction coefficient to eliminate the entropy pulse change caused by instantaneous anomalies of single-point devices.

[0138] Temporal sliding filter correction: retrieve the historical entropy value sequence of the previous 5 consecutive 10-second time slices of the grid, and use the first-order moving average filter to obtain the temporal correction coefficient to filter out instantaneous random fluctuation errors.

[0139] Spatiotemporal coupling correction: The original comprehensive entropy value is weighted and corrected by integrating spatial correction coefficients and temporal correction coefficients to obtain the corrected real-time situation entropy value; avoiding the distortion of judgment caused by single instantaneous abnormal data, the corrected entropy value is sent into the situation prediction and inference unit.

[0140] The situation prediction and extrapolation unit incorporates a three-level entropy threshold system (low disorder threshold 0.25, medium disorder threshold 0.6, and high disorder threshold 0.85) to perform trend extrapolation.

[0141] Traffic saturation classification: Adjusted entropy value < 0.25 indicates smooth traffic; 0.25~0.6 indicates slow traffic; 0.6~0.85 indicates mild congestion; ≥ 0.85 indicates severe congestion.

[0142] Time-series evolution analysis: Comparing the current time series with the slope of the grid entropy value change in the previous 3 minutes, a continuous increase in entropy value indicates a congestion accumulation trend; a continuous decrease in entropy value indicates a congestion dissipation trend.

[0143] Spatial spread prediction: By combining the synchronous entropy changes of adjacent grids, the range of congestion spreading to upstream and downstream and parallel ramps can be predicted;

[0144] The system integrates and outputs structured traffic situation prediction results, which are then simultaneously pushed to the visual effects layer decoupled rendering module for layered rendering and display of visualization layers.

[0145] Thus, the Shannon information entropy algorithm is used to transform the ambiguous road congestion status into a calculable and comparable continuous entropy value. The magnitude of the entropy value directly quantifies the degree of disorder in local traffic order of the road network, achieving refined and digital quantitative assessment and overcoming the shortcoming of existing technologies that cannot quantitatively assess the degree of road network congestion disorder. Secondly, based on the grid adjacency topology relationship pre-generated by the upstream grid module, spatial neighborhood mean smoothing and temporal sliding filtering are carried out simultaneously for dual correction: on the one hand, eliminating the entropy value mutation caused by instantaneous anomalies of single-point equipment, and on the other hand, filtering short-term random traffic flow waves. The system features a dual correction mechanism to avoid prediction distortion caused by instantaneous dirty data and impulse noise, effectively mitigating the core defect of low reliability in existing congestion quantitative analysis technologies. In addition, relying on differentiated fine grid units, it can independently calculate entropy values ​​and predict the situation for different local areas of interchanges, ramps, and main lines, accurately locating local congestion points at the hundred-meter level. At the same time, it combines the temporal changes of entropy values ​​and the synchronous changes of adjacent grids to deduce the trend of congestion accumulation, spread, and dissipation, achieving forward-looking congestion warnings and breaking through the limitations of traditional coarse-grained analysis of the entire road section and the inability to predict the evolution of congestion.

[0146] A visual effects layer decoupling rendering module is used to connect the spatiotemporal grid mapping analysis module and the situation entropy change inference module. It performs layer-level decoupling and decomposition on gridded spatiotemporal traffic data and traffic situation prediction results to complete multi-dimensional visualization rendering output. In some embodiments, the visual effects layer decoupling rendering module includes:

[0147] The multi-layer data access decoupling unit is used to receive gridded spatiotemporal traffic data and traffic situation prediction results respectively, and to classify and decouple the gridded spatiotemporal traffic data and traffic situation prediction results and remove redundant rendering fields.

[0148] The visualization hierarchy division unit is used to perform three-level visualization hierarchy division of gridded spatiotemporal traffic data and traffic situation prediction results according to the road network base layer, real-time traffic status layer, and situation prediction and early warning layer.

[0149] The differentiated visual effects rendering unit is used to configure corresponding color rules, transparency parameters and dynamic rendering styles for different levels of data, and to perform differentiated visual mapping for normal traffic, slow traffic, congestion and predicted situation.

[0150] The dynamic layer overlay unit is used to spatially register and overlay multiple independent rendering layers, and integrate them to generate a multi-layer overlay visualization layer.

[0151] The real-time output unit is used to continuously and dynamically refresh the multi-layered visualization layers, outputting a full-domain road network visualization that includes real-time traffic conditions and situational predictions.

[0152] Specifically, the multi-layer data access decoupling unit establishes two independent data receiving channels, which respectively connect to the two major upstream modules:

[0153] The first channel receives gridded spatiotemporal traffic data, which carries global grid IDs, grid spatial coordinates, traffic density, traffic speed, weather, and equipment operation and maintenance indicators across all dimensions. The second channel receives traffic situation prediction results, including real-time situation entropy values ​​after correction for each grid, congestion level, congestion aggregation / dissipation evolution trend, and predicted diffusion range labels. The two channels fully reuse the same grid IDs, 10-second time slices, and UTC timestamp spatiotemporal identifiers, eliminating the need for secondary coordinate matching and transformation.

[0154] Classification and decoupling: The two types of data are split into three independent data pools: static road network basic data, real-time traffic operation indicators, and congestion situation prediction indicators. This isolates the static base map data from the dynamic real-time updated data and avoids data read and write conflicts.

[0155] Redundant field removal: Delete intermediate calculation parameters (such as single-dimensional entropy values ​​and spatial / temporal correction coefficients, which are only used for algorithm calculation and have no visualization requirements) in the process of raster entropy calculation and time-series correction. Only retain the raster space boundary, traffic indicator values, congestion level, and evolution trend labels required for layer rendering. Compress the volume of data to be rendered to reduce the computing power cost of graphics rendering. After decoupling and cleaning, push the data into three categories of data pools to the visualization level division unit.

[0156] The visualization hierarchy is divided into units based on business display logic. The decoupled dataset is then split into three layers with completely independent and uncoupled rendering logic for each layer.

[0157] Road network base layer: Extract static road network vectors, raster space boundaries, road topology outlines, and static data of interchange / ramp / service area geographical boundaries as the bottom fixed base map layer. It is only loaded during system initialization and has no high-frequency refresh operation.

[0158] Real-time traffic status layer: Extract dynamic indicators of vehicle flow density, traffic speed, road weather, and equipment online status from gridded spatiotemporal traffic data, bind them to the corresponding grid spatial range, and use them as a high-frequency dynamic refresh layer. The refresh cycle is synchronized with the upstream 10-second time-series slice.

[0159] Situation prediction and early warning layer: Extract the comprehensive entropy value, congestion level, congestion spread trend and prediction and early warning range data from the traffic situation prediction results, and use them as an overlay early warning layer, which is updated synchronously with the real-time traffic status layer;

[0160] After splitting, three independent layer datasets are output, each with its own raster space coordinate index, which are then sent to the differentiated visual effects rendering unit for individual rendering.

[0161] The differentiated visual effects rendering unit configures independent color schemes, transparency, and dynamic rendering styles for each of the three-level layers to achieve differentiated visual distinctions for different traffic conditions:

[0162] Road network base layer: uniformly uses light gray road outlines, white background, fixed transparency of 100%, static with no dynamic animation effects, as the bottom base layer;

[0163] Real-time traffic status layer: Basic road color scheme is configured based on grid traffic rate: green for smooth traffic (speed > 60 km / h), yellow for slow traffic (30~60 km / h), and red for congested traffic (< 30 km / h); a semi-transparent blue mask is overlaid on road sections with low visibility due to rain or snow, and gray dashed border marks are added to the corresponding grids of faulty equipment; the layer transparency is uniformly set to 80%.

[0164] Situation prediction and early warning layer: Based on the corrected real-time situation entropy value, configure the early warning dynamic style: add a slow yellow flashing animation for medium disorder slow traffic warning, and add a high red high-frequency flashing animation for high disorder severe congestion prediction; the predicted congestion spread range is filled with a gradient red semi-transparent fill, and the layer transparency is set to 50% to avoid obscuring the underlying traffic flow status layer;

[0165] Each layer independently completes pixel rendering and animation generation, outputs independent layer rendering textures, and sends them to the dynamic layer overlay unit.

[0166] The dynamic layer overlay unit relies on the unified WGS84 raster space coordinates across the entire chain to achieve precise registration and overlay of multiple layers:

[0167] Unified coordinate alignment: All three levels of layers reuse the WGS84 standard geographic coordinates uniformly converted by the raster mapping module, using the latitude and longitude of the raster boundary as the alignment reference, automatically correcting layer pixel offset, and eliminating layer misalignment and boundary mismatch problems.

[0168] The hierarchical stacking order is fixed: bottom layer road network base layer → middle layer real-time traffic status layer → top layer situation prediction and early warning layer, strictly following the display logic from static to bottom layer dynamic and early warning top layer;

[0169] Layer blending calculation: Pixel alpha channel blending is performed based on the preset transparency parameters of each layer, color mixing in overlapping areas of layers is automatically handled, and layer occlusion conflicts are eliminated; after integration, a single multi-layer overlay visualization layer is generated and pushed to the interface real-time output unit.

[0170] The real-time output unit of the interface refreshes synchronously according to a 10-second time-sequence slice, continuously iterating and updating the overlaid visualization layers:

[0171] Reuse the upstream unified 10-second update frequency, and re-render and overlay layers every time a new batch of gridded spatiotemporal traffic data and situation prediction results are received, dynamically refreshing the entire road network screen.

[0172] The output visualization includes a complete road network geographic outline, real-time traffic flow color of each grid, weather / equipment status markers, flashing warning for congestion prediction, and congestion spread prediction area. It also binds grid ID interactive indexes, allowing users to click on a grid to retrieve its historical entropy value and traffic flow time series data.

[0173] The visualized images are simultaneously pushed to the monitoring screen and the web management terminal, completing the entire multi-dimensional visualization rendering output process.

[0174] In this way, data classification and decoupling are completed before layer rendering, and redundant calculation fields are removed, retaining only the fields necessary for visualization, significantly reducing the amount of data to be rendered and memory usage. At the same time, static base maps and dynamically updated data are isolated, static layers do not need to be refreshed repeatedly, and only dynamic traffic flow and warning layers are updated frequently, solving the shortcomings of traditional coupled rendering solutions such as synchronous refresh of all layers, excessive resource consumption, and screen lag. Secondly, the traditional integrated coupled rendering architecture is abandoned, and the road network base map, real-time traffic flow, and situation warning are separated into three completely independent rendering logics, with each layer's color scheme, animation, and refresh cycle not interfering with each other; each layer can be individually switched on or off as needed. The layers support multiple display needs, such as viewing only the basic road network, only real-time traffic flow, and only congestion prediction, solving the pain points of traditional visualization layers being stacked and obscured, information being mixed, and unable to be viewed in layers. In addition, relying on the standardized raster spatial coordinates output by adaptive differentiated raster division, multiple layers are registered and superimposed based on the same set of geographic benchmarks, and there will be no layer misalignment or boundary offset in ultra-fine raster areas of interchanges and ramps. It can accurately display the traffic flow status and congestion prediction of local raster at the level of hundreds of meters. Combined with the refined spatiotemporal raster modeling solution mentioned above, it solves the defects of existing technologies such as poor spatial matching accuracy and inability to display local refined visualization.

[0175] A method for visualizing and analyzing highway traffic data, specifically including:

[0176] Real-time collection of highway traffic flow data, vehicle speed data, road condition video data, weather-related data, and equipment operation and maintenance data to form a standardized raw traffic dataset;

[0177] Heterogeneous data fusion is performed on the standardized raw traffic dataset to generate a structured traffic time-series dataset;

[0178] Based on the highway network topology, a spatiotemporal grid model adapted to road orientation is constructed. Based on the spatiotemporal grid model, the structured traffic time series dataset is spatially registered and temporally bound, and mapped one by one to the corresponding grid cells of the road network to generate gridded spatiotemporal traffic data.

[0179] Based on gridded spatiotemporal traffic data, the traffic flow density, traffic speed, and congestion duration of each road network grid unit are extracted. Then, the traffic situation prediction results are generated by combining the traffic flow density, traffic speed, and congestion duration through the traffic situation entropy change algorithm.

[0180] By integrating gridded spatiotemporal traffic data and traffic situation prediction results, layer-level decoupling and splitting are performed on the two types of data. Differentiated rendering rules are configured for each layer, and layer space registration and overlay are completed to achieve multi-dimensional visualization rendering output of highway traffic.

[0181] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described highway traffic data visualization and analysis method.

[0182] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described highway traffic data visualization and analysis method.

[0183] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A highway traffic data visualization and analysis system, characterized in that, include: The multi-source data acquisition module is used to collect real-time traffic flow data, vehicle speed data, road condition video data, meteorological data, and equipment operation and maintenance data on highways, forming a standardized raw traffic dataset. The edge fusion processing module is used to perform heterogeneous data fusion on the standardized original traffic dataset to generate a structured traffic time-series dataset; The spatiotemporal grid mapping analysis module is used to construct a spatiotemporal grid model based on the highway network topology, and to map the structured traffic time series dataset to the corresponding road network grid cells based on the spatiotemporal grid model to generate gridded spatiotemporal traffic data. The situation entropy change inference module is used to extract the traffic flow density, traffic speed and congestion duration of each road network grid cell based on gridded spatiotemporal traffic data, and generate traffic situation prediction results by combining traffic flow density, traffic speed and congestion duration through the traffic situation entropy change algorithm. The visual effects layer decoupling rendering module is used to connect the spatiotemporal grid mapping analysis module and the situation entropy change inference module. It performs layer decoupling and splitting of gridded spatiotemporal traffic data and traffic situation prediction results to complete multi-dimensional visualization rendering output.

2. The highway traffic data visualization and analysis system according to claim 1, characterized in that, The spatiotemporal raster mapping analysis module includes: The road network topology parsing unit is used to import highway network vector data and extract road topology features. The grid scale adaptive partitioning unit is used to dynamically adjust the grid precision based on the extracted road topology features, combined with road grade and historical traffic activity, to divide the road network grid into differentiated grids. The spatiotemporal coordinate registration unit is used to match and calibrate the collection points of the structured traffic time series dataset with the spatial coordinates of the road network grid. The data raster binding unit is used to associate and mount the calibrated traffic time-series data to the corresponding road network raster, and establish a raster-time-series data binding relationship; The temporal gridded output unit is used to integrate traffic data after binding all road network grids, generate gridded spatiotemporal traffic data, and distribute them to the situation entropy change inference module and the visual effect level decoupling rendering module, respectively.

3. The highway traffic data visualization and analysis system according to claim 2, characterized in that, The raster scale adaptive partitioning unit includes: The road network level assessment subunit is used to identify the types of expressway mainlines, ramps, interchanges, and service area road sections based on the extracted road topology elements and match them with the corresponding basic grid reference scale. The traffic flow activity measurement subunit is used to statistically analyze the traffic flow frequency, speed fluctuation range and congestion frequency of each road segment type based on historical and real-time structured traffic time series data, and quantify the traffic flow activity level of the road segment. The dynamic grid granularity assignment sub-unit is used to perform a two-factor coupled determination based on the road segment network level and the road segment traffic activity level, and to assign grids of different scales to each road segment to complete the differentiated grid division of the entire road network.

4. The highway traffic data visualization and analysis system according to claim 1, characterized in that, The situation entropy change inference module includes: The feature parameter extraction unit is used to extract the core parameters of traffic flow density, traffic speed and congestion duration of each road network grid cell from the gridded spatiotemporal traffic data, and to construct a single grid traffic state feature vector. The parameter normalization processing unit is used to perform dimensionless normalization processing on the single-grid traffic state feature vector to obtain the normalized single-grid traffic state feature vector. The grid situation entropy calculation unit is used to calculate the distribution probability of traffic state parameters of each grid unit based on the normalized single grid traffic state feature vector, and to calculate the single grid traffic situation entropy value, thereby quantifying the degree of local traffic disorder in the road network. The spatiotemporal entropy correction unit is used to combine the spatial correlation and temporal continuous change characteristics of adjacent road network grids to perform spatiotemporal smoothing correction on the traffic situation entropy value of a single grid, so as to obtain the corrected real-time situation entropy value. The situation prediction and simulation unit is used to compare the corrected real-time situation entropy value with the preset classification threshold, analyze the evolution law of entropy value rise and fall, deduce the trend of traffic flow fluctuation, congestion accumulation and spread, and generate the final traffic situation prediction result.

5. The highway traffic data visualization and analysis system according to claim 4, characterized in that, The grid situation entropy calculation unit specifically includes: The parameter interval determination subunit is used to receive the normalized single-grid traffic state feature vector and determine the value range of three types of feature parameters: traffic density, traffic speed, and congestion duration. The state probability statistics subunit is used to calculate the distribution probability of each feature parameter in the corresponding value range, and obtain the single-parameter state distribution probability. The single-dimensional entropy calculation sub-unit is used to calculate the independent dimension entropy values ​​corresponding to the three types of feature parameters based on the single-parameter state distribution probability. The weighted entropy fusion subunit is used to perform weighted fusion calculation of the entropy values ​​of each independent dimension according to the influence weight of each feature parameter on the road network traffic situation, so as to obtain the comprehensive situation entropy value of a single grid. The situation quantization output sub-unit is used to quantify the degree of traffic order disorder of the road network grid based on the comprehensive situation entropy value of a single grid, and output the quantized situation entropy value of a single grid.

6. The highway traffic data visualization and analysis system according to claim 4, characterized in that, The spatiotemporal entropy change correction unit includes: The entropy data receiving subunit is used to receive the single-grid traffic situation entropy value of each independent road network grid output by the grid situation entropy value calculation unit. The spatial neighborhood smoothing correction sub-unit is used to select the upstream and downstream adjacent grids and the left and right parallel grids of the target grid, and to carry out spatial dimension correction through the neighborhood entropy mean smoothing algorithm, calculate the spatial correction coefficient, and complete the entropy value spatial smoothing. The temporal fluctuation filtering correction subunit is used to retrieve the historical temporal entropy value sequence of the current grid, perform temporal dimension correction using a temporal sliding filter, calculate the temporal correction coefficient, and filter out instantaneous entropy value fluctuations. The spatiotemporal coupling correction subunit is used to fuse spatial correction coefficients and temporal correction coefficients to perform spatiotemporal dual-dimensional coupling correction on the single grid situation entropy value, so as to obtain the corrected real-time situation entropy value.

7. The highway traffic data visualization and analysis system according to claim 1, characterized in that, The visual effects layer decoupling rendering module includes: The multi-layer data access decoupling unit is used to receive gridded spatiotemporal traffic data and traffic situation prediction results respectively, and to classify and decouple the gridded spatiotemporal traffic data and traffic situation prediction results and remove redundant rendering fields. The visualization hierarchy division unit is used to perform three-level visualization hierarchy division of gridded spatiotemporal traffic data and traffic situation prediction results according to the road network base layer, real-time traffic status layer, and situation prediction and early warning layer. The differentiated visual effects rendering unit is used to configure corresponding color rules, transparency parameters and dynamic rendering styles for different levels of data, and to perform differentiated visual mapping for normal traffic, slow traffic, congestion and predicted situation. The dynamic layer overlay unit is used to spatially register and overlay multiple independent rendering layers, and integrate them to generate a multi-layer overlay visualization layer. The real-time output unit is used to continuously and dynamically refresh the multi-layered visualization layers, outputting a full-domain road network visualization that includes real-time traffic conditions and situational predictions.

8. A method for visualizing and analyzing highway traffic data, characterized in that, Specifically, it includes: Real-time collection of highway traffic flow data, vehicle speed data, road condition video data, weather-related data, and equipment operation and maintenance data to form a standardized raw traffic dataset; Heterogeneous data fusion is performed on the standardized raw traffic dataset to generate a structured traffic time-series dataset; Based on the highway network topology, a spatiotemporal grid model adapted to road orientation is constructed. Based on the spatiotemporal grid model, the structured traffic time series dataset is spatially registered and temporally bound, and mapped one by one to the corresponding grid cells of the road network to generate gridded spatiotemporal traffic data. Based on gridded spatiotemporal traffic data, the traffic flow density, traffic speed, and congestion duration of each road network grid unit are extracted. Then, the traffic situation prediction results are generated by combining the traffic flow density, traffic speed, and congestion duration through the traffic situation entropy change algorithm. By integrating gridded spatiotemporal traffic data and traffic situation prediction results, layer-level decoupling and splitting are performed on the two types of data. Differentiated rendering rules are configured for each layer, and layer space registration and overlay are completed to achieve multi-dimensional visualization rendering output of highway traffic.

9. A computing device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the highway traffic data visualization and analysis method of claim 8.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the highway traffic data visualization and analysis method of claim 8.

Images