A regional carbon emission data monitoring and analyzing system and method

Abstract

The application discloses a regional carbon emission data monitoring and analyzing system and method, and belongs to the technical field of big data analysis and traffic carbon emission monitoring. The method comprises the following steps: acquiring multi-source data of a city road network, wherein the multi-source data at least comprises traffic gate flow data, vehicle GPS track data, weather data and signal lamp timing data; dividing a city area into a plurality of spatial grid units; based on historical vehicle flow form indexes of each grid unit, combining preset business rules reflecting city functional area characteristics, performing dynamic clustering processing on the plurality of grid units to obtain a plurality of carbon emission characteristic grid clusters; wherein the judgment basis of the business rules at least partially originates from land use property data around the grid units; for each carbon emission characteristic grid cluster, according to the correlation characteristics of its vehicle flow form characteristics and event data, selecting a corresponding prediction algorithm to establish a traffic carbon emission prediction sub-model; based on the output of each sub-model, generating a carbon emission hotspot distribution map of the city road network in a future preset time period, and associating and outputting carbon emission contribution degrees and emission attribution analysis results of each characteristic grid cluster. Through the dynamic grid clustering mechanism of "vehicle flow form clustering + city functional area rule verification", the application realizes fine monitoring and accurate tracing of city traffic mobile source emissions, and effectively improves the spatio-temporal resolution and management operability of traffic carbon emission prediction.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and carbon emission monitoring technology, specifically to a method and system for dynamic monitoring and source tracing of urban traffic carbon emission hotspot grids, which is particularly suitable for refined monitoring and dynamic source tracing analysis of high emission hotspot areas such as urban intersections and frequently congested road sections. Background Technology

[0002] Urban road traffic is one of the main sources of carbon emissions, and accurate monitoring of traffic carbon emissions has become a key task in achieving the "dual carbon" goal. Existing methods for predicting traffic carbon emissions are mainly divided into two categories: one is the accounting method based on macroeconomic statistical data, such as the emission factor method, which estimates the total carbon emissions by using regional fuel consumption and average emission factors; the other is the prediction method based on machine learning, such as data-driven models like XGBoost and LSTM, which predict carbon emission trends by mining historical data.

[0003] However, existing methods have significant limitations when dealing with dynamic emission identification at intersection levels. Patents such as CN119941271B and CN119961795A propose a method for predicting carbon emissions from building units. By constructing a multi-step time-series prediction model and integrating static and time-series features, it achieves accurate building carbon emission prediction. This method is highly effective in building scenarios, but its technical architecture is based on a "single object + multi-dimensional features" modeling approach. This means extracting static attribute features (such as building type and area) and time-series features (such as energy consumption data) for a single building unit and performing end-to-end prediction. If this approach is directly applied to urban traffic scenarios, the following problems arise: traffic emissions are mobile sources, and their carbon emission intensity is strongly correlated with event-driven factors such as traffic flow composition, traffic light timing, sudden congestion, and large-scale events. While traditional models can fit historical trends, they lag in responding to emission mutations caused by real-time events and cannot answer questions such as "which intersection emits more" and "why it emits more."

[0004] This invention addresses the aforementioned problems by proposing a "grid-based dynamic clustering" prediction approach. Unlike existing technologies that treat the city as a whole or model individual objects, this invention first divides the urban area into multiple spatial grid units, performs initial clustering based on traffic flow pattern indicators, and then verifies and adjusts this clustering by incorporating urban functional area business rules, resulting in multiple carbon emission characteristic grid clusters. For different characteristic clusters, a corresponding prediction algorithm is selected based on their event response characteristics to establish a sub-model. Through a dynamic clustering mechanism of "data-driven clustering + business rule verification," this invention achieves accurate monitoring of intersection-level traffic carbon emissions and can correlate emission fluctuations with real-time events for tracing their origin, effectively solving the pain point of existing technologies that "can only see high emissions but don't know why they are high." Summary of the Invention

[0005] In a first aspect, embodiments of this disclosure provide a method for dynamic monitoring and source tracing of urban traffic carbon emission hotspot grids, including: Acquire multi-source data of the road network in the target city area; the multi-source data includes at least traffic checkpoint flow data, vehicle GPS trajectory data, meteorological data, and traffic light timing data; The target urban area is divided into multiple spatial grid units according to a preset spatial resolution; Based on the traffic flow data at the traffic checkpoints and the vehicle GPS trajectory data, calculate the traffic flow pattern indicators of each grid unit within the historical period. By combining preset business rules that reflect the characteristics of urban functional areas, the multiple grid units are dynamically clustered to obtain multiple carbon emission characteristic grid clusters; wherein, the judgment basis of the business rules is at least partially derived from the land use data around the grid units. For each carbon emission characteristic grid cluster, based on the correlation characteristics between its traffic flow pattern characteristics and event data, a corresponding prediction algorithm is selected to establish a traffic carbon emission prediction sub-model; Based on the output of each traffic carbon emission prediction sub-model, a carbon emission hotspot distribution map of the target urban area for a future preset time period is generated, and the carbon emission contribution and emission attribution analysis results of each carbon emission characteristic grid cluster are output in conjunction.

[0006] Preferably, the multi-source data further includes at least one of the following: Bus / subway card swipe data obtained from the public transportation management system; Data on freight vehicle entry and exit events obtained from the logistics management platform; Real-time traffic congestion index data obtained from internet map service providers; Air quality data obtained from environmental monitoring stations Preferably, the calculation of traffic flow pattern indicators for each grid cell includes: The first type of indicators characterizing traffic flow fluctuations are extracted, including traffic flow variance, average vehicle speed variance, and peak-to-valley ratio within a preset time period; The second type of indicator characterizing the periodicity of traffic flow is extracted, including the main period intensity and the corresponding period length identified through spectrum analysis; A third type of indicator characterizing the responsiveness of traffic flow events is extracted, including the correlation strength between the traffic flow curve and the specific event occurrence time marked by the event data; the specific events include large-scale events, traffic accidents, and temporary traffic control.

[0007] Preferably, the dynamic clustering processing based on business rules includes: Based on the similarity of the traffic flow pattern indicators, the grid cells are initially clustered. The initial clustering results are validated and adjusted using application business rules, which include: If two grid cells belonging to the same clustering level have different urban functional zone types indicated by their land use data, then they will be assigned to different characteristic grid clusters. If two grid units belonging to different initial clusters have land use data indicating that they belong to the same urban functional area type and are simultaneously affected by the same traffic artery or traffic events, then they will be merged into the same characteristic grid cluster.

[0008] Preferably, the types of urban functional zones include: residential-dominated zones, commercial and office zones, industrial and logistics zones, cultural, educational and medical zones, and transportation hub zones.

[0009] Preferably, the selection of the corresponding prediction algorithm includes: For feature grid clusters that remain stable after initial clustering and business rule verification, and whose third-class index is lower than the preset threshold, the time series prediction algorithm is selected as the traffic carbon emission prediction sub-model for the cluster. For feature grid clusters whose third-category indicators are higher than a preset threshold, a machine learning regression algorithm that uses the event data as the key input feature is selected as the traffic carbon emission prediction sub-model for that cluster.

[0010] Preferably, when establishing a predictive sub-model for a machine learning regression algorithm, the method for converting the event data into model input features includes: Transform large-scale events into activity intensity characteristics for future preset time periods; Transform logistics park inbound and outbound events into time-series features that characterize freight activity; Transform traffic light timing data into periodic features that characterize intersection traffic efficiency.

[0011] Preferably, the emission attribution analysis results include: Identify the dominant feature grid clusters that lead to high carbon emissions and their percentage contribution to emissions; The emission composition of the dominant characteristic grid clusters was analyzed, including vehicle type composition analysis, time period distribution analysis, and congestion contribution analysis. Based on the event data, output the key event types and their impact levels that led to the emission anomalies.

[0012] Secondly, this disclosure also provides a dynamic monitoring and tracing system for urban traffic carbon emission hotspot grids, the system comprising: The multi-source data access module is used to synchronize the multi-source data from traffic management departments, meteorological departments, and map service providers. The grid division and feature extraction engine is used to divide the urban area into multiple spatial grid units and calculate the traffic flow pattern indicators of each grid unit. A dual-constraint dynamic clustering engine is used to divide the multiple grid units into multiple carbon emission characteristic grid clusters based on the similarity of the traffic flow pattern indicators and a preset business rule base; wherein, the business rule base defines the logic for determining the type of urban functional area and traffic synergy based on land use data; The cluster prediction modeling module is used to train and predict the traffic carbon emission sub-prediction results for each carbon emission characteristic grid cluster by calling the model algorithm that matches the characteristics of the cluster. The hotspot visualization and source analysis module is used to aggregate the sub-prediction results of each cluster, generate a carbon emission hotspot distribution map, and output the carbon emission contribution and emission attribution analysis results of each characteristic grid cluster.

[0013] Preferably, the hotspot visualization and source tracing analysis module is further configured as follows: When there is a significant deviation between the predicted results and the actual monitoring data, the source analysis process is initiated to locate the main characteristic grid clusters and related events that caused the deviation. It supports interactive querying of carbon emission composition and reasons for changes in a specified region and time period.

[0014] The present invention has the following advantages: This invention proposes a "grid-based dynamic clustering" prediction approach. By dividing urban areas into fine-grained grid units, initial clustering is performed based on traffic flow pattern indicators. This is then validated and adjusted using urban functional area business rules, resulting in carbon emission characteristic grid clusters with different event response characteristics. For different characteristic clusters, differentiated prediction algorithms are selected based on their event response intensity (e.g., the XGBoost model, with event data as the key input, is used for logistics channel clusters with strong event response, while the Prophet time-series model is used for commuter-dominated clusters with strong periodicity and weak event response). Carbon emission hotspot distribution maps and attribution analysis results are generated based on the output of each sub-model. This invention, through a dynamic clustering mechanism of "data-driven clustering + business rule validation," achieves accurate monitoring of intersection-level traffic carbon emissions and can correlate emission fluctuations with real-time events such as large-scale events and traffic accidents, effectively solving the pain point of existing technologies that "can only see high emissions but don't know why they are high." Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating the steps of a method for dynamic monitoring and tracing of carbon emission hotspot grids provided in this embodiment of the invention; Figure 2 This invention provides a dynamic monitoring and tracing system for urban traffic carbon emission hotspot grids. Detailed Implementation

[0017] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings to make the technical solution of the present invention clearer and more complete. It should be noted that the described embodiments are for illustrative purposes only and are not intended to limit the present invention. Other implementation methods that can be made by those skilled in the art based on the content of the present invention without creative effort should all fall within the protection scope of the present invention.

[0018] In this application and its claims, unless otherwise expressly stated, the terms "comprising," "including," and similar expressions should be understood to indicate the presence of the listed items without excluding the presence or addition of other items. The words "an," "a," and similar terms should not be construed as limited to the singular in this application, and may also include multiple items.

[0019] Furthermore, the accompanying drawings in this application are merely illustrative and not necessarily drawn to scale. The same reference numerals denote components with the same or similar functions. To clearly illustrate the present invention, specific details are provided in the following embodiments. Those skilled in the art should understand that these details are not essential for implementing the present invention, and other methods may be used to implement it without affecting the basic idea of ​​the invention.

[0020] Figure 1 The diagram illustrates the entire method flow of an embodiment of the present invention, as detailed below: I. Methods and Procedures for Dynamic Monitoring and Source Tracing of Carbon Emission Hotspot Grids This embodiment uses a core area of ​​a first-tier city's central urban area as the implementation scenario. This area covers approximately 15 square kilometers and includes 32 major intersections, 5 main urban roads, 2 large commercial complexes, 3 schools, and 1 logistics distribution center. Intelligent traffic checkpoint systems, floating car monitoring systems, and traffic light control systems have been deployed within the area, providing a multi-source data foundation for this invention.

[0021] S1: Multi-source traffic data acquisition and spatiotemporal alignment This step utilizes a multi-source data access module to synchronously acquire multi-source data on the road network of the target city area from traffic management departments, meteorological departments, and map service providers. Details are as follows: (1) Traffic flow data collection at traffic checkpoints Traffic flow data at checkpoints is obtained through checkpoint monitoring equipment deployed at major intersections and road sections in the city.

[0022] ① Data Sources and Monitoring Points: A total of 128 checkpoint capture units (model example: Hikvision iDS-TCV900, supporting vehicle type recognition) were deployed at the entrance and exit directions of 32 major intersections in the target area. Each capture unit includes a high-definition camera, supplementary lighting, and a front-end recognition processor.

[0023] ② Data Acquisition and Upload: The checkpoint equipment captures images of passing vehicles in real time. The front-end processor identifies the vehicles based on deep learning algorithms and outputs each vehicle passage record, containing the following fields: pass_time (passage time, accurate to the second), intersection_id (intersection ID), lane_id (lane ID), plate_number (license plate number, encrypted), vehicle_type (vehicle type, coded as: 01-small passenger car, 02-large passenger car, 03-light truck, 04-medium truck, 05-heavy truck), and vehicle_speed (instantaneous speed, unit: km / h). Each vehicle passage record is uploaded in real time to the traffic management department's data center via a dedicated fiber optic network and simultaneously pushed to the system's multi-source data access interface. The interface is built on a Kafka message queue to ensure reliable access for high-concurrency data.

[0024] (2) Vehicle GPS trajectory data collection The vehicle GPS trajectory data comes from the floating car monitoring systems of ride-hailing platforms and taxi companies.

[0025] ①Data source: Real-time GPS location data from approximately 8,000 ride-hailing vehicles and 3,000 taxis operating within the target city area. Sampling frequency is 10 seconds per sampling.

[0026] ② Data format: Each trajectory point record contains the following fields: device_id (device ID, anonymized), timestamp (timestamp, Unix timestamp format), longitude (longitude, WGS84 coordinate system), latitude (latitude, WGS84 coordinate system), speed (instantaneous speed, unit km / h), direction (driving direction angle, 0-360 degrees), and status (vehicle status, 0-unloaded, 1-loaded).

[0027] ③ Data Access: GPS trajectory data is pushed to the data access layer of this system in real time via the MQTT protocol, with the subscription topic being / city / gps / raw.

[0028] (3) Meteorological data collection Meteorological data is used to correct emission factors under different temperature and humidity conditions.

[0029] ①Data source: Access to national meteorological station observation data released by the meteorological bureau of the target city, as well as data from three micro meteorological stations deployed in the area.

[0030] ② Data content: including temperature (temperature, unit °C), humidity (relative humidity, %), wind speed (wind speed, unit m / s), wind direction (wind direction), and weather type (weather phenomenon, such as sunny, rainy, snowy).

[0031] ③ Data collection method: The latest meteorological data is obtained by calling the RESTful API interface provided by the meteorological bureau (in this embodiment, the meteorological subscription service interface from Alibaba Cloud is used) at regular intervals of 5 minutes.

[0032] (4) Traffic light timing data acquisition Traffic light timing data is used to characterize the periodic characteristics of intersection traffic efficiency.

[0033] ①Data source: Timing scheme data from the traffic signal control system of the area.

[0034] ② Data content: including intersection_id (intersection ID), phase_id (phase ID), start_time (phase start time), end_time (phase end time), duration (phase duration in seconds), and cycle_length (total cycle duration in seconds).

[0035] ③ Data Acquisition Method: The timing scheme change event is acquired in real time through the data synchronization agent service deployed on the signal control system side and pushed to this system via HTTP POST.

[0036] (5) Event data collection Event data is used to characterize specific events that may lead to sudden changes in carbon emissions.

[0037] ① Large-scale event data: Obtain event scheduling information from cultural and tourism bureaus and large venues, including venue_id, event_name, start_time, end_time, and expected_attendance.

[0038] ②Inbound and outbound event data of logistics parks: Access inbound and outbound records from the regional logistics management platform, including warehouse_id (warehouse ID), truck_plate (license plate number, encrypted), entry_time (inbound time), exit_time (outbound time), and goods_type (goods type).

[0039] ③ Traffic accident data: Accident information is obtained from the traffic control center's emergency response system, including accident_id (accident ID), location (location coordinates), report_time (alarm time), clear_time (time of completion of handling), and severity (1-minor, 2-moderate, 3-serious).

[0040] (6) Unified data access and spatiotemporal alignment After acquiring the above data, all raw data collected through the aforementioned channels is sent to the unified data access layer (this example uses Apache Kafka). Each type of data is published as a publisher to its corresponding Kafka topic: topic_traffic_capture, topic_gps_raw, topic_weather, topic_signal_timing, and topic_events.

[0041] A data normalization microservice acts as a consumer, consuming messages from these topics. This service processes data from different sources as follows: ① Cleaning: Remove obviously abnormal data, such as GPS coordinates outside the city limits, negative vehicle speeds, and vehicle passage times that are in the future.

[0042] ② Timestamp unification: Convert the timestamps of all data to ISO8601 format in UTC+8 time zone.

[0043] ③ Spatial Index Encoding: For each piece of data with spatial location (vehicle passing record, GPS point, accident point), it is encoded based on the Geohash algorithm with a precision of 7 bits (corresponding to a grid of approximately 150 meters × 150 meters), which facilitates subsequent aggregation by grid.

[0044] Finally, all data is organized according to time series and written into a distributed time series database (such as an InfluxDB cluster), providing a unified and standardized data foundation for the subsequent processing step S2.

[0045] S2: Spatial grid division and traffic flow pattern index calculation This step, based on the historical data collected and stored in S1, first divides the target urban area into several spatial grid units. Then, it performs quantitative analysis on the traffic flow sequence of each grid unit, extracting three types of morphological indicators that characterize its inherent patterns. This process is executed by a periodic batch processing job (running monthly in this embodiment), with the analysis window being the most recent consecutive 90 calendar days.

[0046] (1) Spatial grid division The target urban area is divided into multiple spatial grid units according to a preset spatial resolution.

[0047] ① Grid granularity determination: Taking into account road network density, calculation accuracy, and computing resources, this embodiment adopts a grid granularity of 100 meters × 100 meters. This granularity is sufficient to distinguish the traffic flow characteristics of different intersections and road sections, while avoiding the data sparsity problem caused by excessively fine grids.

[0048] ② Grid Coding: Assign a unique identifier Grid_ID to each grid cell, with the coding rule being "city code_row number_column number", such as "XC_0123_0456". Simultaneously record the coordinates (latitude and longitude) of the grid's center point and the coordinate range of its four corners.

[0049] ③ Grid Attribute Labeling: Based on urban land use planning data and point-of-interest data, the dominant functional area type is labeled for each grid unit, including: residential dominant area, commercial and office area, industrial and logistics area, cultural, educational and medical area, transportation hub area, green space and water area, and others. The labeling results are stored in the grid_profile table.

[0050] (2) Data preprocessing First, query the time series database for all raw data related to the target area within the specified historical period (past 90 days).

[0051] ① Grid-based aggregation of traffic flow data: Each vehicle passage record and GPS trajectory point is mapped to a corresponding grid cell based on its spatial coordinates. For each grid cell, the following time series is generated by aggregating data with a 15-minute statistical window: Traffic flow sequence Q(t): The total number of vehicles passing through this grid within the window.

[0052] Average vehicle speed sequence V(t): the arithmetic mean of the instantaneous speeds of vehicles passing through this grid within the window.

[0053] Heavy vehicle proportion sequence H(t): The proportion of heavy trucks and large buses passing through the grid within the window to the total number of vehicles.

[0054] Congestion index sequence C(t): calculated based on the ratio of average vehicle speed to free-flow vehicle speed, with the free-flow vehicle speed being the 85th percentile of the historical speed of the same period (same week, same time period) for that grid.

[0055] ② Data cleaning: Clean the time series of each grid, use linear interpolation to fill in individual data missing due to equipment failure (no more than 3 consecutive missing time points), and use statistical methods based on the 3σ principle to remove obvious outliers (such as vehicle speed exceeding 120km / h or traffic flow increasing by more than 10 times).

[0056] ③ Data alignment: Ensure that all time series in all grids have the same start and end time, as well as the same time interval (15 minutes), to form a unified time series data matrix.

[0057] (3) Calculation of traffic flow pattern indicators For the time series sequence of each preprocessed grid cell, the following three types of indices are calculated: ① First type of indicator: Volatility indicators These indicators reflect the dispersion and fluctuation of traffic flow.

[0058] Traffic flow variance: characterizes the degree of fluctuation in traffic flow over time.

[0059] The calculation process is as follows: For each day, firstly, calculate the traffic flow sequence for 96 time points (24 hours × 4 points / hour) of the grid. Secondly, calculate the average daily traffic volume for that day. Then, calculate the 96 days. Sample variance of values Finally, the daily variance over 90 days was analyzed. Calculate the arithmetic mean to obtain the traffic flow variance index for this grid.

[0060] Average vehicle speed variance: characterizes the stability of vehicle speed and indirectly reflects congestion fluctuations.

[0061] The calculation process is similar to that of traffic flow variance, based on the average vehicle speed sequence. calculate.

[0062] Peak-to-valley ratio: characterizes the fluctuation range of traffic flow throughout the day.

[0063] The calculation process is as follows: For each day, first, find the daily traffic flow sequence. maximum value and minimum value Secondly, calculate the average daily traffic volume for that day. Then, calculate the peak-to-valley ratio for that day. Finally, for each day within 90 days... Calculate the arithmetic mean to obtain the peak-to-valley ratio index for this grid.

[0064] ② Second category of indicators: Periodic indicators These indicators identify repetitive patterns hidden in traffic flow sequences.

[0065] Detrending of Serials: Complete 90-day Traffic Flow Series By fitting a linear trend line using the least squares method and subtracting the trend, the detrended sequence is obtained. This eliminates the influence of long-term trends on cycle analysis.

[0066] Fast Fourier Transform: for The spectrum is obtained by applying the Fast Fourier Transform. .

[0067] Power spectrum calculation: Calculate the power spectral density of the spectrum. .

[0068] Main cycle identification: After excluding the DC component (zero frequency), within a reasonable frequency range (corresponding to a cycle from 30 minutes to 7 days), search for... The maximum value point. The frequency corresponding to this point. This is the dominant frequency.

[0069] The indicators are calculated as follows: Main cycle: (Unit: hour or day). For example, if the dominant frequency corresponds to 2 cycles per day, then the main cycle is 12 hours.

[0070] Principal period intensity: the intensity at the point of maximum value. The value, after normalization, serves as an intensity indicator; the larger the value, the more significant the periodicity.

[0071] ③ Third category of indicators: Event responsiveness indicators These metrics quantify the sensitivity of traffic flow to specific events.

[0072] Response intensity for large-scale events: Extract all large-scale activity events that occurred within the specified grid or surrounding grids (within 500 meters) during the analysis period from the event database, and denot this set as the event set. Each event It has a start time and end time .

[0073] For each event : Calculate the time 1 hour before the event ( Average traffic flow within ) .

[0074] Calculate the event period ( Average traffic flow .

[0075] Calculate the traffic flow response intensity of this event. .

[0076] Calculate all of this grid arithmetic mean As an indicator of the intensity of response to large-scale events.

[0077] Traffic accident response intensity: Extract all traffic accident events that occurred in this grid or adjacent grids during the analysis period from the event database, and denote them as the set. Each accident It has its own time of occurrence and clearing time .

[0078] For each accident : Calculate the 30 minutes before the accident ( Average speed within ) .

[0079] Calculate the period of the accident ( average vehicle speed .

[0080] Calculate the vehicle speed response intensity of the accident. (A positive value indicates a decrease in vehicle speed).

[0081] Calculate all of this grid arithmetic mean , as an indicator of the intensity of response to traffic accidents.

[0082] School dismissal incident response intensity (for the grid surrounding the school): For grids marked as education, culture, and medical areas, extract the school dismissal time on weekday afternoons (e.g., 16:00-18:00) as periodic events.

[0083] Calculate the average traffic flow during non-school dismissal periods .

[0084] Calculate the average traffic flow during school dismissal time .

[0085] Calculate response strength .

[0086] (4) Result storage After the calculation is complete, each grid cell will receive a morphological index vector containing the aforementioned specific values. This vector is stored in the `grid_morphology_features` table of a relational database (such as PostgreSQL), containing the following fields: `grid_id`, `variance_traffic`, `variance_speed`, `pvr`, `dominant_period`, `period_strength`, `event_response_even`, `event_response_accident`, `event_response_school`, and `update_time`.

[0087] This result is used in the S3 clustering process.

[0088] S3: Dynamic Mesh Clustering Based on Dual Constraints This step aims to group grid cells with similar traffic flow patterns and carbon emission characteristics into one category, laying the foundation for subsequent differentiated modeling. Its core is to integrate two constraints: "data-driven morphological similarity" and "knowledge-driven functional area rules," to perform dynamic clustering. This process is executed by a dual-constraint dynamic clustering engine and consists of three stages: data clustering, rule verification, and dynamic adjustment.

[0089] (1) Initial clustering based on traffic flow pattern indicators In this stage, based on the quantitative morphological indicators calculated by S2, an unsupervised machine learning algorithm is used for preliminary grouping.

[0090] ① Feature vector construction A multidimensional feature vector is constructed for each grid cell. To balance the dimensions of the various indices, Z-score standardization is applied to each morphological index of all cells.

[0091] For example, for a grid cell Grid_XC_0123_0456, its eigenvector V_grid can be represented as:

[0092] ② Clustering algorithm execution The K-Means++ algorithm from the scikit-learn machine learning library is used. This algorithm effectively avoids local optima caused by improper initial center point selection. In this embodiment, the elbow rule is used to automatically determine the optimal K value. The specific logic is as follows:

[0093] Key parameter: Determination of the preset number of clusters K. In this embodiment, calculations show a clear inflection point when K=5; therefore, the initial number of clusters is determined to be 5. Assume five initial clusters are obtained:

[0094] (2) Validation and adjustment based on business rule base This stage introduces domain knowledge to refine the purely data-driven clustering results, making them more aligned with the realities of urban functional area management. Business rules are stored in a structured manner in a rule base (in this embodiment, the Drools rule engine is used for management).

[0095] ① Example of rule base definition:

[0096] ② Rule execution and calculation logic: Execute rule "R001" (Separate functional areas): For each grid cell in Initial_Cluster_1, all grid pairs are traversed, and their functional area types are checked. For example, Initial_Cluster_1 contains Grid_P (residential-dominated area) and Grid_Q (industrial logistics area). According to rule R001, the functional area types of the two are significantly different (residential and industrial belong to different major categories), triggering a separation action. The system marks Grid_P and Grid_Q as "to be separated" and removes them from Initial_Cluster_1.

[0097] Implement Rule “R002” (Traffic Corridor Consolidation): For grid cells distributed along the same urban arterial road (such as Jiefang Road), these grids may belong to different initial clusters (e.g., some in Cluster_2, some in Cluster_3). Using a traffic accident on Jiefang Road as the baseline event, the event synchronization drive intensity (SYNC) of these grids is calculated. The calculation shows that the proportion of grids along Jiefang Road experiencing a synchronized decrease in vehicle speed during a traffic accident exceeds 0.7, which is higher than the threshold of 0.6. According to rule R002, these grids are determined to be "synchronously driven by the same traffic corridor," triggering a merging action to combine the grids along the road into a new candidate cluster.

[0098] Execute rule "R003" (forced clustering at special locations): For the grids (Grid_S1, Grid_S2, Grid_S3) containing the three schools in the area, although they may be assigned to different initial clusters due to differences in traffic flow patterns, according to rule R003, the system found that all three are educational and medical area grids and are relatively close in space (less than 500 meters). Therefore, the three are forcibly merged into the "school sensitive cluster".

[0099] ③ Secondary clustering and cluster merging After initial verification and adjustment by the rule engine, some grids were marked as "to be separated" and needed to be re-clustered; some grids were marked as "to be merged" and needed to be merged into new clusters.

[0100] The system performs the following operations: First, all marked grid cells are removed from their original clusters. Second, for grid cells marked "to be separated," K-Means clustering is re-executed (with the K value dynamically adjusted), assigning them to the most similar existing cluster or forming new smaller clusters. Then, for grid cells marked "to be merged," they are directly merged into new feature clusters and assigned new cluster IDs. Finally, the cluster size is checked; if a cluster has too few grid cells (e.g., less than 3), it is merged with its most similar neighboring clusters.

[0101] (3) Generate the final carbon emission characteristic grid cluster After adjustments and secondary optimizations by the rule engine, a final stable clustering result is formed. The system assigns a descriptive name and a unique ID to each final cluster and records all grid cell members within the cluster. The resulting final clustering result is shown in Table 1.

[0102] Table 1 Final Clustering Results

[0103] The clustering results are persistently stored in the system configuration database and serve as key inputs for independently selecting and training prediction models for each cluster in step S4. This dual-constraint clustering mechanism of "data + rules" is one of the core innovations of this invention in improving the targeting and accuracy of subsequent prediction models.

[0104] S4: Construction and Training of Clustered Traffic Carbon Emission Prediction Sub-model This step involves selecting a prediction algorithm, constructing a feature engineering pipeline, and training the model for each carbon emission feature grid cluster divided by S3, based on its unique traffic flow patterns and event response characteristics. This results in the creation of a dedicated traffic carbon emission prediction sub-model for each cluster.

[0105] (1) Carbon emission accounting benchmark Before building the predictive model, the traffic flow data first needs to be converted into carbon emissions. This embodiment uses a method based on the COPERT model emission factor for calculation.

[0106] ① Construction of a basic emission factor database Based on vehicle type and driving conditions, a basic emission factor library is constructed and stored in the emission_factors table of the database, the structure of which is shown in Table 2.

[0107] Table 2 Example of data in the emission_factors table

[0108] ② Meteorological correction factor Temperature and humidity have a significant impact on emission factors. In this embodiment, a weather correction coefficient table is constructed and stored in the weather_correction table, the structure of which is shown in Table 3.

[0109] Table 3 Example of data in the weather_correction table

[0110] ③ Grid carbon emission calculation For each grid cell, the carbon emission E_grid(t) is calculated in each 15-minute time period t as shown in formula (1): (1) in: The number of vehicles of a certain type passing through this grid within time period t; The average road segment length within the grid (in this embodiment, it is taken as 1 / 2 of the grid diagonal length, i.e., approximately 70 meters); Emission factors are based on average vehicle speed and vehicle type; Meteorological correction factors based on temperature and humidity (2) Model selection strategy and decision logic The core basis for model selection is the third type of indicator (event responsiveness indicator) calculated in S2, with reference to other morphological characteristics. The system presets a response intensity threshold. It is used for quantitative decision-making.

[0111] ① For the commuter-dominant cluster (F_001): Judgment: The average value of the "Large-scale event response intensity" index for this cluster is 0.12, which is below the threshold. The average "Traffic Accident Response Intensity" was 0.08, which is also below the threshold. However, its "Main Period Intensity" index was as high as 0.85, showing a very significant periodicity.

[0112] Decision: Select a classic time series forecasting model. This embodiment uses the Facebook Prophet model because it can automatically handle trends, seasonality, and holiday effects, and its assumptions about the periodicity of the series are highly consistent with the characteristics of this cluster.

[0113] ②For the logistics channel cluster (F_002): Judgment: The average value of the "Large-scale event response intensity" index for this cluster is 0.08, but the "traffic accident response intensity" is 0.42, which is higher than the threshold. More importantly, this cluster of grids has a strong correlation with the inbound and outbound events of the logistics park, with an event synchronization driving strength of 0.78.

[0114] Decision: Select a machine learning regression model that uses event-indicating data as the key explanatory variable. This embodiment uses the Extreme Gradient Boosting Regressor (XGBoost Regressor) because it can effectively handle nonlinear relationships and feature interactions, and is robust to missing values.

[0115] ③ For the commercially active cluster (F_003): Judgment: The cluster's "large-scale event response intensity" index is as high as 0.65, far exceeding the threshold. Weekend traffic is significantly higher than weekday traffic and is highly correlated with shopping mall promotions and large-scale events.

[0116] Decision: Select a machine learning regression model centered on event data. XGBoostRegressor will also be used, but its input features will focus on constructing features related to the event itself.

[0117] ④ For the school sensitive cluster (F_004): Judgment: The "School Dismissal Event Response Strength" index of this cluster is as high as 0.81, far exceeding the threshold. Its emission peaks exhibit a strong temporal regularity (specific times during weekday afternoons).

[0118] Decision: A hybrid model combining time rules and machine learning is selected. This embodiment uses the Prophet model as a base, and adds the "school dismissal event" as an external regression factor (add_regressor) to the model.

[0119] ⑤ For the transportation hub cluster (F_005): Judgment: The cluster's "large-scale event response intensity" is 0.35, slightly above the threshold. Its emission fluctuations are strongly correlated with train arrival and departure timetables.

[0120] Decision: Select a machine learning model that uses train timetable events as key input. In this embodiment, XGBoostRegressor is chosen.

[0121] (3) Feature engineering and model input construction For different types of models, construct differentiated feature inputs.

[0122] ① Data preparation for the Prophet model for cluster F_001: The input to the Prophet model is a DataFrame with two columns: ds (time stamp) and y (target value).

[0123] Data preparation: The historical carbon emission data of all grid cells within the cluster are summed by timestamps to obtain the cluster-level historical total carbon emission sequence y.

[0124] Convert timestamps (ds) to Pandas datetime format with a frequency of 15 minutes.

[0125] Add holiday features: Pass information such as statutory holidays and adjusted workdays into the model as the holidays parameter.

[0126] Add additional regression factors: optional, such as "average temperature of the day" can be added as add_regressor.

[0127] ② Feature engineering for the XGBoost model of the F_002 cluster: Core event feature transformation Characteristics of logistics activity: Obtain the logistics park inbound and outbound plans for the next 24 hours at 15-minute intervals from the logistics management platform. For each forecast period t, calculate the total number of planned inbound and outbound vehicles for all logistics parks within the cluster within the time window [t-2h, t+2h], as a continuous feature.

[0128] Heavy vehicle proportion characteristics: Based on historical data, the predicted value of the heavy vehicle proportion of each grid cell in the future period is calculated, which can be estimated using time series models or simple averaging methods.

[0129] Traffic accident risk characteristics: Based on historical accident data, and combined with factors such as weather and time of day, accident risk probability characteristics are constructed. For example, if the predicted time period is rainy at night, a higher accident risk coefficient is assigned.

[0130] Spatiotemporal context features include one-hot encoding of "hour cycle" (0-23), one-hot encoding of "day of the week" (1-7), "whether it is a holiday" flag, and "whether it is a workday" flag.

[0131] Historical lag characteristics: Carbon emissions of the cluster in the past 1 hour, 3 hours, and 24 hours (at the same period point) are added as characteristics.

[0132] Each row of the final training feature matrix X_F002 corresponds to a historical time point and contains all the feature columns mentioned above.

[0133] ③ Feature engineering for the XGBoost model of the F_003 cluster: Core event feature transformation: Large-scale event characteristics: Information on mall promotions, large-scale performances, sports events, and other events scheduled for the next 24 hours is obtained from the event management platform. For each predicted time period t, the number of ongoing events within the cluster or within a 500-meter radius is counted within the time window [t-2h, t+2h], and weighted by event size (e.g., number of participants / 1000) to obtain the event intensity characteristics.

[0134] Weekend / Holiday Characteristics: Due to the significant weekend effect in commercially active clusters, key characteristics include "whether it is a weekend", "whether it is the day before a holiday", and "whether it is the last day of a holiday".

[0135] Meteorological characteristics: Access the meteorological API to obtain the forecast temperature, forecast humidity, and weather phenomena (rain / snow / sunny) for the next 24 hours.

[0136] Historical and periodic characteristics: Same as F_002 cluster, including historical lag characteristics and spatiotemporal context characteristics.

[0137] ④ Feature engineering for the hybrid model of the F_004 cluster: Prophet base model: Similarly, the cluster-level historical total carbon emission sequence y is used as the target, and ds is used as the timestamp.

[0138] External regression factors: The "school dismissal event" is included as an external regression factor. Construction method: For each weekday afternoon dismissal period (16:00-18:00), a dummy variable `school_event_flag` is defined, set to 1 during this period and 0 at other times. This variable is then passed as `add_regressor` to the Prophet model.

[0139] (4) Model training and hyperparameter optimization ① Data partitioning Historical data from the past 90 consecutive days was used as the training set. For the XGBoost model, the data was divided chronologically, with the first 80% used for training and the last 20% for validation. For the Prophet model, all data was used for fitting, but the performance was evaluated through cross-validation.

[0140] ② Training process For the XGBoost model: Using the xgboost library, set the initial parameters as follows: learning_rate=0.05, max_depth=6. n_estimators=200, subsample=0.8, colsample_bytree=0.8. Training is performed using the root mean square error (RMSE) as the loss function. Early stopping (early_stopping_rounds=20) is used with the validation set to prevent overfitting. Grid search is further employed to optimize the hyperparameter space, with the search range including learning_rate (0.01, 0.05, 0.1), max_depth (4, 6, 8), and subsample (0.7, 0.8, 0.9).

[0141] For the Prophet model: Using the prophet library, call the fit() method to perform the fitting. Adjust the seasonality scale parameter seasonality_prior_scale (set to 10 to enhance seasonality fitting ability) and the holiday scale parameter holidays_prior_scale (set to 10). For the F_004 cluster, add school_event_flag as an additional regression factor.

[0142] ③ Model Evaluation The performance of each model was evaluated using a validation set, with key metrics including: Root Mean Square Error (RMSE): Measures the deviation between the predicted value and the actual value. Mean Absolute Percentage Error (MAPE): measures relative error Coefficient of determination ( ): Measure the goodness of fit of the model The performance of each cluster model was evaluated and is shown in Table 4: Table 4 Performance Evaluation Table for Each Cluster Model

[0143] ④ Model saving The best-performing model, after training and validation, along with its corresponding feature engineering pipeline (including normalizers, encoders, etc.), is serialized and saved via MLflow and registered in a model repository. Each feature cluster has its own independent model storage path and version number, for example:

[0144] At this point, each carbon emission feature grid cluster has a tailor-made, trained prediction sub-model, ready for S5's integrated prediction.

[0145] S5: Integrated Carbon Emission Prediction and Hotspot Source Analysis This step, as the output stage of the prediction process, is responsible for coordinating and scheduling the prediction sub-models of each feature cluster trained in S4, generating a unified regional carbon emission hotspot distribution map, and performing attribution analysis to quantify the expected contribution of each grid cluster to future carbon emissions, answering the two core questions of "where are the emissions the most" and "why are the emissions the most."

[0146] (1) Predictive task scheduling and execution This step is automatically triggered daily at a preset time (e.g., 17:00 UTC+8) by the predictive scheduling service in the system (implemented in this embodiment based on Apache Airflow). Its core is to execute an atomic predictive pipeline job.

[0147] ① Data preparation before forecasting The scheduling service first accesses the external data interface to obtain the latest future information necessary for performing the prediction: Obtain information from traffic management departments regarding planned events such as traffic control and road construction for the following day (00:00-23:45).

[0148] Obtain the schedule of major events and promotional activities for the following day from the Bureau of Culture and Tourism and shopping malls.

[0149] Obtain hourly forecasts of outdoor temperature, humidity, and weather phenomena for the following day from the meteorological service interface.

[0150] Obtain the next day's inbound and outbound plans for the logistics park from the logistics management platform.

[0151] Obtain the train / passenger vehicle arrival and departure timetable for the next day from the railway / passenger transport department.

[0152] Meanwhile, historical actual carbon emission sequences of each feature cluster are extracted from the internal time series database for a certain window period (such as the most recent 72 hours) before the prediction start time, and used for the construction of lag features required by some models.

[0153] ② Clustered model parallel inference Based on the clustering configuration determined by S3, the scheduling service creates parallel prediction tasks for the five feature clusters F_001 to F_005 respectively.

[0154] Each prediction task loads the prediction sub-model and feature engineering pipeline that are trained and saved in S4 for the corresponding feature cluster.

[0155] Real-time feature construction: Each task, based on the type of its cluster, uses the acquired future and historical data to dynamically construct the model input feature matrix for the next 24 hours (96 points at 15-minute intervals). This process reuses the feature engineering logic for each cluster defined in S4.

[0156] For F_001: Construct a DataFrame containing ds (96 time points for the next day), and pass in holiday information.

[0157] For F_002: Construct a feature matrix that includes logistics activity characteristics, heavy vehicle proportion characteristics, traffic accident risk characteristics, spatiotemporal characteristics, and historical lag characteristics.

[0158] For F_003: Construct a feature matrix that includes features of large-scale events, weekends, weather, and historical lag.

[0159] For F_004: Construct a DataFrame containing ds and set school_event_flag to the value of the corresponding time period.

[0160] For F_005: Construct a feature matrix containing train time event features.

[0161] Model inference: Input the constructed feature matrix into the corresponding model and execute the predict method. Each model outputs a vector of length 96. ,in The predicted carbon emissions of feature cluster i in the next t-th 15-minute period are expressed in kilograms of carbon dioxide equivalent. ).

[0162] ③ Grid-level spatial mapping of carbon emissions The prediction result of each feature cluster is the total number of clusters. It needs to be allocated to each grid cell within the cluster in order to generate a spatial distribution map.

[0163] Allocation method: Allocation is carried out proportionally based on the carbon emission proportion of each grid unit in the same historical period (such as the same week or the same period).

[0164] Suppose that cluster i contains n grids, and the proportion of carbon emissions from grid j to the total emissions of the cluster during the same historical period is given by... Then the predicted carbon emissions for grid j in time period t are: = × .

[0165] If historical data for the forecast period is unavailable, the average percentage of the same period in the most recent week can be used as a substitute.

[0166] (2) Generation of carbon emission hotspot distribution map ① Aggregation of grid-level prediction results The prediction results of all grids are organized according to the time dimension to form a three-dimensional prediction data cube containing 96 time slices and approximately 1500 grids.

[0167] ② Hotspot Level Classification For each time slice t, sort all the predicted carbon emission values ​​of the grids from high to low, and classify the hotspot levels according to the following rules: Hotspot Level 1 (Red): Top 5% of the grid Hotspot Level 2 (Orange): 5%-15% of the grid Hotspot Level 3 (Yellow): 15%-30% of the grid Normal area (green): the rest of the grid ③ Heatmap generation Based on GIS engines (such as Mapbox and Leaflet), the grid prediction results are rendered into dynamic heat maps. The heat maps support sliding playback along the timeline, intuitively displaying the spatiotemporal evolution of carbon emission hotspots within 24 hours of the following day.

[0168] ④Result persistence and publishing The generated hotspot distribution map data (predicted values ​​for each grid and each time period) is written into the prediction results database (such as the spatiotemporal database PostGIS), recording metadata such as prediction generation time, prediction target date, and data source (model version).

[0169] The prediction results are simultaneously pushed to a message bus (such as Kafka) to notify downstream data visualization services, alarm services, and decision support systems.

[0170] (3) Calculation and attribution analysis of carbon emission contribution Contribution analysis and attribution analysis are among the core innovations of this invention, aiming to answer the two core questions of "where are they ranked higher" and "why are they ranked higher".

[0171] ① Cluster-level contribution calculation For each feature cluster i, the predicted total carbon emissions for the next day By analyzing the predicted sequence Summing yields: (unit: ).

[0172] Calculate the predicted total emissions for the region .

[0173] Calculate the percentage contribution of each feature cluster i. .

[0174] ② Hotspot grid attribution analysis For grids marked as "hotspot level 1", the system automatically performs attribution analysis to pinpoint the causes of their high emissions.

[0175] Attribution analysis logic is implemented based on a rule engine. Example rule:

[0176] ③Result formatting and visualization Attribution analysis results are stored as structured data (JSON format) and correlated with hotspot distribution maps. The system visualization module (based on ECharts and Mapbox) automatically reads the latest prediction and attribution data, generating and displaying the following views: Main view: A dynamic map of carbon emission hotspot distribution for the next 24 hours, supporting timeline scrolling playback. Clicking on any hotspot grid will display detailed information about that grid.

[0177] Cluster-level contribution dashboard: Visually displays the percentage contribution of each feature cluster in pie chart format. Each sector in the pie chart represents a feature cluster, and the size of the sector corresponds to its contribution.

[0178] Hotspot Grid Details: Lists all grids marked as hotspot level 1 and level 2, including the following fields: grid ID, location description (e.g., "intersection of Jiefang Road and Zhongshan Road"), predicted emissions (…). (15 min), hotspot level, and attribution results.

[0179] Attribution Statistics Dashboard: Statistics on the attribution distribution of hot grids for the day, for example: "Today there are 45 hot grids, of which 23 are caused by traffic congestion, 12 by heavy vehicle concentration, 6 by large-scale events, and 4 by school dismissal."

[0180] (4) Deviation source analysis The system is also equipped with a deviation tracing function. When there is a significant deviation between the actual monitoring data and the prediction results (such as the actual emissions exceeding the predicted value by 20%), the source tracing analysis process will be automatically started.

[0181] ① Deviation detection The real-time monitoring module continuously receives actual traffic flow and carbon emission data for the following day and compares it with the predicted values. When the cumulative deviation exceeds a preset threshold for three consecutive time periods (45 minutes), deviation tracing is triggered.

[0182] ② Deviation positioning The source analysis process first identifies the main feature clusters causing the deviation. Then, it calculates the absolute value of the deviation between the actual and predicted values ​​for each feature cluster, sorts them by their contribution to the deviation, and identifies the cluster with the largest contribution.

[0183] ③Retrospective analysis of related events For the cluster with the largest contribution to the bias, backtrack the actual events that occurred in that cluster during the bias period and compare them with the input features used in the prediction to identify the key events that caused the bias. For example, if the F_003 cluster has the largest bias, and backtracking reveals that an unpredictable temporary promotional activity actually occurred, then the source conclusion is output: "The sudden increase in emissions in the commercial area was caused by an unrecorded temporary promotional activity."

[0184] ④ Model Adaptive Hints The system records the deviation events and attribution results in the model training log, prompting the model administrator to consider adding new event types or adjusting feature weights in the next training cycle.

[0185] This contribution analysis and source tracing mechanism enables urban traffic managers not only to grasp the overall emission trends, but also to clearly identify the main sources and causes of emissions, providing direct quantitative evidence and decision support for implementing precise traffic control and emission reduction measures (such as optimizing signal timing, adjusting freight hours, and strengthening traffic management around events).

[0186] II. Dynamic Monitoring and Source Tracing System for Carbon Emission Hotspot Grids Reference Figure 2 This disclosure also provides a dynamic monitoring and tracing system for urban traffic carbon emission hotspot grids, the system comprising: The multi-source data access module is used to synchronize the multi-source data from traffic management departments, meteorological departments, and map service providers; the multi-source data includes at least traffic checkpoint flow data, vehicle GPS trajectory data, meteorological data, and traffic light timing data; A grid partitioning and feature extraction engine is used to divide an urban area into multiple spatial grid units and calculate traffic flow pattern indicators for each grid unit based on the multi-source data. The traffic flow pattern indicators include a first type of indicator representing volatility, a second type of indicator representing periodicity, and a third type of indicator representing event responsiveness. A dual-constraint dynamic clustering engine is used to divide the multiple grid units into multiple carbon emission characteristic grid clusters based on the similarity of the traffic flow pattern indicators and a preset business rule base; wherein, the business rule base defines the logic for determining the type of urban functional area and traffic synergy based on land use data; The cluster prediction modeling module is used to train and predict the traffic carbon emission sub-prediction results for each carbon emission characteristic grid cluster by calling the model algorithm that matches the characteristics of the cluster. The hotspot visualization and source analysis module is used to aggregate the sub-prediction results of each cluster, generate a carbon emission hotspot distribution map, and output the carbon emission contribution and emission attribution analysis results of each characteristic grid cluster.

[0187] The hotspot visualization and source tracing analysis module is also configured to: initiate the source tracing analysis process when there is a significant deviation between the prediction results and the actual monitoring data, locate the main feature grid clusters and related events that caused the deviation; and support users to interactively query the carbon emission composition and reasons for changes in a specified area and time period.

[0188] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0189] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0190] In this disclosure, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The block diagrams of devices, apparatuses, devices, and systems involved in this disclosure are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as "comprising," "including," "having," etc., are open-ended terms meaning "including but not limited to," and are used interchangeably with them. The terms "or" and "and" as used herein refer to the terms "and / or," and are used interchangeably with them unless the context clearly indicates otherwise. The term "such as" as used herein refers to the phrase "such as but not limited to," and is used interchangeably with it.

[0191] Additionally, as used herein, the "or" used in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0192] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0193] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0194] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0195] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for dynamic monitoring and source tracing of urban traffic carbon emission hotspot grids, characterized in that, include: Acquire multi-source data of the road network in the target city area; the multi-source data includes at least traffic checkpoint flow data, vehicle GPS trajectory data, meteorological data, and traffic light timing data; The target urban area is divided into multiple spatial grid units according to a preset spatial resolution; Based on the traffic flow data at the traffic checkpoints and the vehicle GPS trajectory data, calculate the traffic flow pattern indicators of each grid unit within the historical period. By combining preset business rules that reflect the characteristics of urban functional areas, the multiple grid units are dynamically clustered to obtain multiple carbon emission characteristic grid clusters; wherein, the judgment basis of the business rules is at least partially derived from the land use data around the grid units. For each carbon emission characteristic grid cluster, based on the correlation characteristics between its traffic flow pattern characteristics and event data, a corresponding prediction algorithm is selected to establish a traffic carbon emission prediction sub-model; Based on the output of each traffic carbon emission prediction sub-model, a carbon emission hotspot distribution map of the target urban area for a future preset time period is generated, and the carbon emission contribution and emission attribution analysis results of each carbon emission characteristic grid cluster are output in conjunction.

2. The method according to claim 1, characterized in that, The multi-source data also includes at least one of the following: Bus / subway card swipe data obtained from the public transportation management system; Data on freight vehicle entry and exit events obtained from the logistics management platform; Real-time traffic congestion index data obtained from internet map service providers; Air quality data obtained from environmental monitoring stations.

3. The method according to claim 1, characterized in that, The calculation of traffic flow pattern indicators for each grid cell includes: The first type of indicators characterizing traffic flow fluctuations are extracted, including traffic flow variance, average vehicle speed variance, and peak-to-valley ratio within a preset time period; The second type of indicator characterizing the periodicity of traffic flow is extracted, including the main period intensity and the corresponding period length identified through spectrum analysis; A third type of indicator characterizing the responsiveness of traffic flow events is extracted, including the correlation strength between the traffic flow curve and the specific event occurrence time marked by the event data; the specific events include large-scale events, traffic accidents, and temporary traffic control.

4. The method according to claim 1, characterized in that, The dynamic clustering process based on business rules includes: Based on the similarity of the traffic flow pattern indicators, the grid cells are initially clustered. The initial clustering results are validated and adjusted using application business rules, which include: If two grid cells belonging to the same clustering level have different urban functional zone types indicated by their land use data, then they will be assigned to different characteristic grid clusters. If two grid units belonging to different initial clusters have land use data indicating that they belong to the same urban functional area type and are simultaneously affected by the same traffic artery or traffic events, then they will be merged into the same characteristic grid cluster.

5. The method according to claim 4, characterized in that, The types of urban functional zones include: residential-dominated zones, commercial and office zones, industrial and logistics zones, cultural, educational and medical zones, and transportation hub zones.

6. The method according to claim 1, characterized in that, The selected corresponding prediction algorithm includes: For feature grid clusters that remain stable after initial clustering and business rule verification, and whose third-class index is lower than the preset threshold, the time series prediction algorithm is selected as the traffic carbon emission prediction sub-model for the cluster. For feature grid clusters whose third-category indicators are higher than a preset threshold, a machine learning regression algorithm that uses the event data as the key input feature is selected as the traffic carbon emission prediction sub-model for that cluster.

7. The method according to claim 6, characterized in that, When establishing a predictive sub-model for a machine learning regression algorithm, methods for transforming the event data into model input features include: Transform large-scale events into activity intensity characteristics for future preset time periods; Transform logistics park inbound and outbound events into time-series features that characterize freight activity; Transform traffic light timing data into periodic features that characterize intersection traffic efficiency.

8. The method according to claim 1, characterized in that, The results of the emission attribution analysis include: Identify the dominant feature grid clusters that lead to high carbon emissions and their percentage contribution to emissions; The emission composition of the dominant characteristic grid clusters was analyzed, including vehicle type composition analysis, time period distribution analysis, and congestion contribution analysis. Based on the event data, output the key event types and their impact levels that led to the emission anomalies.

9. A dynamic monitoring and source tracing system for urban traffic carbon emission hotspot grids, characterized in that, The system for implementing the method of any one of claims 1-8 comprises: The multi-source data access module is used to synchronize the multi-source data from traffic management departments, meteorological departments, and map service providers. The grid division and feature extraction engine is used to divide the urban area into multiple spatial grid units and calculate the traffic flow pattern indicators of each grid unit. A dual-constraint dynamic clustering engine is used to divide the multiple grid units into multiple carbon emission characteristic grid clusters based on the similarity of the traffic flow pattern indicators and a preset business rule base; wherein, the business rule base defines the logic for determining the type of urban functional area and traffic synergy based on land use data; The cluster prediction modeling module is used to train and predict the traffic carbon emission sub-prediction results for each carbon emission characteristic grid cluster by calling the model algorithm that matches the characteristics of the cluster. The hotspot visualization and source analysis module is used to aggregate the sub-prediction results of each cluster, generate a carbon emission hotspot distribution map, and output the carbon emission contribution and emission attribution analysis results of each characteristic grid cluster.

10. The system according to claim 9, characterized in that, The hotspot visualization and source tracing analysis module is also configured as follows: When there is a significant deviation between the predicted results and the actual monitoring data, the source analysis process is initiated to locate the main characteristic grid clusters and related events that caused the deviation. It supports interactive querying of carbon emission composition and reasons for changes in a specified region and time period.

Images