A method and system for constructing a port-to-port voyage level ship emission inventory based on AIS data

Abstract

The application discloses a kind of port-to-port voyage level ship emission list construction method and system based on AIS data, first complete the distributed cleaning of multiple-source AIS and ship basic data Regularization;Ship navigation / park working condition is judged based on speed threshold, combined with spherical distance matching and 17 nautical mile port buffer radius to realize entity port level node identification, through space connectivity rule to suture AIS signal missing broken chain, automatically generate complete "port-port" OD section;Through trajectory interpolation completion, main engine low load correction, emission control area fuel correction realizes 8 types of pollutants point-by-point fine accounting;Finally, distributed aggregation outputs grid emission list and port-to-port OD section emission list.The application improves the ship emission accounting accuracy from national / regional level to entity port and single voyage scale, solves the problem of existing technology section fragmentation, large accounting deviation and high degree of manual intervention, and can support accurate tracking and control of global shipping carbon pollution.

Description

Technical Field

[0001] This invention relates to the field of big data processing technology for transportation navigation safety and environmental protection, specifically to a method and system for constructing an inter-port voyage-level ship emission inventory based on AIS (Automatic Identification System) data. It can be used for refined accounting of air pollutants and carbon emissions at the voyage scale between physical ports worldwide, supporting green shipping management and carbon emission tracking. Background Technology

[0002] With increasingly stringent global requirements for greenhouse gas emission reduction and regional air pollution control, the refined management of the shipping industry, a significant source of air pollutants and greenhouse gas emissions in the transportation sector, has become crucial for achieving green shipping. However, existing ship emission accounting and inventory compilation technologies are largely limited to specific ports or fixed near-shore waters, lacking systematic and dynamic quantitative methods for inter-port shipping activities (origin-destination voyages). Due to limitations in massive trajectory data cleaning, intelligent identification of complex voyages, and coupled accounting of multiple pollutants, current maritime emission inventories generally suffer from insufficient spatiotemporal resolution, significant biases in calculation results, and difficulty in achieving accurate carbon emission tracking at the single voyage scale.

[0003] In the process of quantifying ship pollutant emissions, the "bottom-up" accounting method has been widely used, and existing representative technologies can be divided into three categories: The first category is a regional ship emission calculation technology based on AIS trajectory repair, represented by patent CN112214721A. Its specific steps include: acquiring AIS data and integrating it into a ship information database containing information such as IMO, MMSI, ship name, engine, and fuel type. This also includes classifying data by ship type and length, using the median of complete information data identified in the ship information database as filler for missing information; analyzing the reporting time intervals based on the total or average time intervals of data reported under different navigation states, and taking an appropriate value as the time step for dynamic processing; dynamically reading data and cleaning abnormal and duplicate data such as heading angle, speed, and point distribution; identifying and supplementing the cleaned data with the supplemented database; estimating the displacement of each trajectory point of a single ship using the dynamic method; and manually selecting specific polygonal geographical areas using overlaid map information to statistically analyze regional ship emissions; finally, analyzing the trajectory based on dynamic ship emission characteristics, repairing and supplementing the navigation trajectory according to different missing conditions, and recalculating and correcting the emission characteristics. However, this regional emission calculation method has limited functionality and lacks automatic division logic for inter-port OD segments. The scheme only supports manually delineating polygonal areas to summarize the total emissions. It does not establish an automatic matching mechanism between origin and destination ports, cannot independently separate complete "port-to-port" voyages, and cannot achieve the summary and accurate attribution of emissions for a single inter-port route.

[0004] The second category is route reconstruction and prediction technology based on AIS big data, represented by patent CN112562415A. This technology collects, cleans, transforms, and analyzes massive amounts of historical AIS data to achieve three major functions: route selection and recommendation, navigation trajectory reconstruction, and real-time route prediction. Its specific steps include: First, calculating the spherical distance between the two ports of each candidate route sample and performing multiple rounds of optimization calculations based on the average route distance to eliminate abnormal routes; Second, based on the eliminated route samples, considering factors such as the port of origin, destination, transit points, supply requirements, empty / full load status, ship type, and time range. Historical routes are searched and grouped by waypoints or resupply points. Next, using kernel density estimation, a distance normal distribution is calculated with heading data as a sample and distance as a parameter. An ascending list of port spacings is used as input, and the list indices with the smallest differences at the 25th and 75th quantiles, and the smallest differences in the mean values ​​of the quantiles, are used as outputs to filter and return a recommended route list. Finally, the actual ship trajectories are reconstructed on a map, and the ship's destination is predicted in real time using recommendation and trajectory reconstruction techniques, with the prediction results dynamically updated based on actual operational data. However, this route reconstruction and prediction method only completes the spatial relationship analysis of routes and lacks emission quantification capabilities. This technology only solves the problems of trajectory reconstruction, route selection, and destination prediction; it does not include multi-pollutant and carbon emission accounting modules and cannot support the requirements for compiling shipping emission inventories. The route identification and emission calculation processes are independent of each other.

[0005] The third category comprises journal articles on national-scale shipping carbon emission assessments that integrate AIS and international trade flows: WANG XT, LIU H, et al. Trade-linked shipping CO2emissions [J]. Nature Climate Change, 2021, 11 (11): 945-951. This paper constructs a comprehensive global shipping carbon emission assessment framework, integrating high-frequency AIS trajectories and international trade data to achieve international voyage identification and emission calculation. First, it uses minute-level AIS speed and heading data to analyze ship motion status, using crossing national territorial sea baselines as the OD (Original Distance) voyage division standard, and calculates the carbon emission intensity of each voyage segment by combining ship attribute parameters; at the same time, it builds a standardized global shipping route network, uses trade volume weighting to process related trade data, and characterizes the overall pattern of global shipping carbon emissions at the national scale. However, the above voyage division is coarse-grained, with OD units only at the national level. The method in this paper uses the national territorial sea baseline as the standard for navigation range demarcation, does not identify the anchoring, berthing, and navigation operations of ships in various physical ports, and the emission analysis granularity cannot be refined to the physical port, making it impossible to distinguish the differences in route emissions between different ports within the same country.

[0006] In summary, existing technologies have not yet formed a complete technical system that integrates automatic extraction of port-level OD segments, refined emission accounting throughout the entire process, and multi-dimensional inventory output, and cannot meet the needs of precise control of shipping carbon emissions at the port scale. Summary of the Invention

[0007] To address the technical problems in current maritime emission inventory compilation, such as insufficient spatial resolution accuracy, inability to attribute to physical ports, fragmentation of shipping segments due to missing AIS signals, large emission calculation deviations, and inability to output multi-dimensional lists in an integrated manner, this invention provides a method and system for constructing inter-port voyage-level vessel emission inventories based on AIS data. Based on AIS trajectory data processing technology, it comprehensively constructs an inter-port shipping emission calculation technology system. By establishing port buffer matching, voyage segment status identification, and chain break reconstruction techniques, it improves the accuracy of port-level emission positioning and OD (Original Distance) voyage resolution, refines emission calculation accuracy to the physical port and single voyage scale, accurately identifies the complete origin-destination voyage of a vessel, and completes the quantitative calculation of carbon emissions, thereby improving the overall technical accuracy of inter-port shipping emission calculation and vessel emission inventory compilation.

[0008] This invention discloses a method for constructing an inter-port voyage-level vessel emission inventory based on AIS data. It utilizes the Spark distributed computing framework to build an integrated system for trajectory analysis and emission accounting, including the following steps: Step 1: Distributed Acquisition and Preprocessing of Basic Data: Parallel reading of global ship AIS dynamic trajectory data, ship static attribute database, emission factor database, and global port center point latitude and longitude dataset; distributed cleaning of raw AIS data, extraction of core fields such as ship unique identifier (MMSI), timestamp, latitude and longitude, and instantaneous speed, and removal of invalid records; sorting of individual ship trajectory points in ascending order by timestamp using MMSI as the grouping key to complete time series normalization; Step 2, Determining Ship Operating Conditions and Generating Initial OD Segments: Traverse the time-series trajectory points of a single ship, dynamically mark the navigation / berthing status based on the speed threshold, correct abnormal short segments, and filter valid berthing segments; Step 3, High-precision dynamic identification of port nodes: Based on the pre-loaded coordinates of global port center points, the port of origin is determined for each valid berthing segment by calculating the spherical distance and matching the buffer radius; the initial OD segment is generated based on the ports of origin of adjacent valid berthing segments before and after the navigation segment. Step 4, Disconnected Link Segment Reconstruction: For disconnected links caused by missing AIS signals, adjacent segments are merged based on spatial connectivity rules to generate a complete closed-loop inter-port OD segment dataset. Step 5, Trajectory Interpolation and Refined Emission Calculation: Spatiotemporal interpolation is performed on adjacent AIS trajectory points where the time interval exceeds the interpolation threshold. The main engine load factor is calculated based on the ship's static parameters and low load correction is performed. The emission factors of the three types of emission sources, namely main engine, auxiliary engine and auxiliary boiler, are matched and matched. The fuel oil correction is performed by superimposing emission control area rules. The emissions of various air pollutants and greenhouse gases are calculated for each interpolation point. Step 6: Multi-dimensional aggregation and output of emission data: Discrete emission points are aggregated separately according to the set spatial resolution grid and origin and destination port pairs, and the global gridded ship emission inventory and inter-port OD segment emission inventory are output.

[0009] As a further improvement of the present invention, the distributed cleaning of the original AIS data in step 1 specifically includes: The distributed computing partition is dynamically set according to the original file size, and the core fields of MMSI, timestamp, longitude, latitude and instantaneous speed are extracted. Records with coordinates exceeding the range of [-180, 180] and [-90, 90] are removed, as are records with missing core fields and redundant trajectory points with duplicate MMSI and timestamp. All trajectory points of a single ship are aggregated using a distributed aggregation operator with MMSI as the grouping key, and time-series normalization is completed by sorting the timestamps in ascending order, and the data is converted into a distributed columnar storage structure.

[0010] As a further improvement of the present invention, step 2, which involves determining the ship's operating condition and correcting abnormal sailing segments, specifically includes: The instantaneous speed of each trajectory point is compared with the speed determination threshold of 1.0 knots. Points with speed ≤ threshold are marked as anchored and points with speed > threshold are marked as sailing. Based on this, the continuous trajectory is divided into alternating anchored sub-segments and sailing sub-segments. Abnormal short navigation segments with a duration ≤ the first preset time (300 seconds) will be corrected to a berthing state; berthing sub-segments with a duration ≥ the second preset time (3600 seconds) will be selected as valid berthing segments, and short berthing sub-segments with a duration < the second preset time (3600 seconds) will be corrected to a navigation state.

[0011] As a further improvement of the present invention, the high-precision dynamic identification of port nodes in step 3 specifically includes: Step 31: During the module initialization phase, preload a dataset containing the latitude and longitude coordinates of the center points of 2033 physical ports worldwide and enable a fast spatial location indexing and caching mechanism. Step 32: Perform latitude and longitude range verification on the input trajectory points. After successful verification, calculate the spherical distance between the trajectory points and the center points of each port using the Haversine formula. The specific calculation formula is as follows: After converting the input coordinates from degrees to radians, calculate the difference between longitude and radians: Difference in dimensional radians: Calculate intermediate parameters for distance: Calculate the central angle: ) Finally, the spherical distance is obtained: Where R is the average radius of the Earth (calibrated as 6,371,000 meters), and a fixed conversion factor of 1,852 is used to convert the unit of meter into the international nautical standard unit of "nautical mile (nm)".

[0012] Step 33: Set a port buffer radius of 17 nautical miles and include ports whose spherical distance from the trajectory point is ≤ buffer radius into the candidate matching set; Step 34: If the candidate matching set is empty, return a result of no port matching; if the candidate set contains only one port, directly determine the port to which the point belongs; if the candidate set contains multiple ports, sort them in ascending order of spherical distance, and select the port with the smallest distance as the unique port to which the trajectory point belongs. Step 35: For the port matching results of all trajectory points within the same berthing segment, a mode voting mechanism is used to select the port with the highest frequency of occurrence as the final destination port for that berthing segment.

[0013] As a further improvement of the present invention, step 4, reconstructing the interrupted link segment, specifically includes: Extract the end point and beginning point of two temporally adjacent segments belonging to the same vessel, and calculate the spherical distance between the two points using the Haversine formula. If the distance is ≤100 nautical miles, the two segments are merged into a complete OD segment, inheriting the vessel attribute parameters and average speed of the original segment, and removing the empty values ​​of unknown ports in the middle, to obtain a closed-loop inter-port OD segment dataset with clear origin and destination ports.

[0014] As a further improvement of the present invention, step 5, trajectory interpolation and refined emission calculation, specifically includes: Step 51: Traverse all adjacent AIS trajectory points of a single ship, and use linear spatiotemporal interpolation based on time scaling factor for adjacent trajectory points with time interval > 600 seconds to generate a continuous trajectory sequence with uniform time step. Step 52: Extract the ship's design speed Vdesign from the ship's static attribute library, and calculate the design speed V based on the instantaneous speed V at the current trajectory point using the formula LF=(V / Vdesign). 3 / 100 Calculate the host load factor LF; when LF<0.2, trigger low load correction and match the corresponding low load emission correction coefficient to calibrate the host base emission; Step 53: Match the real-time operating power of auxiliary engines and auxiliary boilers according to the current operating conditions of the ship, and match the corresponding emission factors of 8 types of pollutants in combination with the ship type, construction year and fuel type. The 8 types of pollutants include PM, NOx, SOx, CO, HC, CO2, N2O and CH4. Step 54: Use the global emission control area vector polygon to determine the spatial intersection of each trajectory point. If the trajectory point falls within the emission control area, trigger the fuel switching simulation logic and use the fuel correction coefficient to perform a secondary conversion and correction of SOx and PM emissions. Step 55: Calculate the emissions of each pollutant point by point using the unified formula E=(P×LF×EF×Δt) / C, where P is the rated power of the corresponding emission source, EF is the pollutant emission factor, Δt is the time step of the trajectory point, and C is the dimensionless conversion constant 3.6×10 9 This is used to convert the mass unit of power work calculation results from grams to tons and the time unit from seconds to hours.

[0015] As a further improvement of the present invention, the multi-dimensional aggregation output of emission data in step 6 specifically includes: The latitude and longitude values ​​that were stored using integer scaling during the calculation process to optimize storage are restored to the true geographic latitude and longitude values ​​according to the formulas Longitude=scale_lon×res and Latitude=scale_lat×res, where res is a preset spatial resolution of 0.1°; Construct a distributed key-value pair dataset with grid latitude and longitude index, starting port name, and ending port name as composite keys, and 8 types of pollutant emission vectors as values; The distributed reduction operator is invoked to accumulate the emissions of each pollutant in parallel according to the spatial grid dimension and the origin and destination port pair dimension, respectively, and outputs a global gridded ship emission inventory with a resolution of 0.1°, as well as an inter-port OD segment emission statistics table with the physical port pair dimension. The output results can be directly connected to the geographic information system for spatial visualization.

[0016] As a further improvement of the present invention, the spatial location fast index caching mechanism in step 31 is specifically as follows: the latitude and longitude of the trajectory point are rounded to 4 decimal places to generate a unique cache index. When the cache is hit, the pre-stored port matching result is returned directly. When the cache is not hit, the full spherical distance is calculated and the matching result is stored in the cache.

[0017] This invention also discloses a port-based, range-department-level vessel emission inventory construction system based on AIS data, used to execute the above-described construction method, including: The data preprocessing module is used to perform the basic data distributed acquisition and preprocessing operations described in step 1 above; The operating condition determination module is used to perform the ship operating condition determination and initial OD segment generation operations described in step 2 above; The port matching module is used to perform the high-precision dynamic identification operation of port nodes described in step 3 above. The segment reconstruction module is used to perform the broken link segment reconstruction operation described in step 4 above. The emissions accounting module is used to perform the trajectory interpolation and refined emissions calculation operations described in step 5 above; The aggregation output module is used to perform the multi-dimensional aggregation output operation of emission data described in step 6 above.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Significantly improved OD identification accuracy: This invention relies on a global database of 2,033 physical port locations and a precise buffer radius of 17 nautical miles. Combining the minimum distance selection rule and the mode voting noise reduction mechanism, it improves the accuracy of ship origin and destination point identification from the national / coarse regional level of existing technologies to the physical port level. It can accurately locate the 57% of ship CO2 emissions to the physical port route, solve the problems of regional matching ambiguity and trajectory drift noise in dense port clusters, and achieve accurate attribution of emission responsibility.

[0019] 2. Complete voyage segment reconstruction capability: Through dynamic judgment of navigation / berthing conditions and spatial connectivity chain stitching mechanism, this invention can automatically identify the complete voyage logic of "berthing-navigation-berthing", automatically merge fragmented voyage segments caused by AIS signal blind spots in the ocean, and generate complete "port-port" closed-loop voyages without the need for manual delineation of geographical areas, thus solving the problems of segment fragmentation and high degree of manual intervention in existing technologies.

[0020] 3. Significantly improved emission accounting accuracy: This invention completes the discrete trajectory through 600-second step trajectory interpolation, introduces main engine low load correction and emission control zone fuel switching correction, covers three types of emission sources (main engine, auxiliary engine, and boiler) and eight types of pollutants, and solves the accounting deviation problem caused by incomplete combustion at low ship speeds and compliant fuel replacement in emission control zones.

[0021] 4. Integrated Distributed Computing Architecture: This invention achieves parallel computing throughout the entire process based on a distributed framework, and simultaneously outputs grid-level spatial emission data and port-to-port level route emission data. It can directly connect to GIS systems to generate emission heat maps and inter-port emission correlation matrices, supporting the dual needs of spatial control and route emission reduction analysis, and can support the rapid processing of global annual-level AIS data. Attached Figure Description

[0022] Figure 1 This is a flowchart of the method for constructing an inter-port range-class vessel emission inventory based on AIS data disclosed in this invention; Figure 2 This is a matrix diagram showing the distribution of flight distances by country as disclosed in this invention; Figure 3 This is a national-level CO2 emission distribution matrix diagram disclosed in this invention; Figure 4 This is a spatial distribution map of CO2 emissions from a typical inter-port OD route disclosed in this invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. 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.

[0024] The present invention will now be described in further detail with reference to the accompanying drawings: This invention discloses a method and system for extracting port-level OD (Original Distance) routes and calculating multiple pollutants and carbon emissions per voyage based on massive AIS trajectory data. It addresses the problem that existing technologies lack a complete technical system integrating port-level OD segment extraction and refined emission accounting throughout the entire process. This invention utilizes the Spark distributed computing framework to build an integrated system for trajectory analysis and emission accounting: a spherical buffer zone is constructed based on the port's center point to match AIS trajectory points with the actual port; trajectory routes are segmented using speed thresholds and minimum berthing durations to distinguish between navigation and berthing states; spatial connectivity rules are introduced to splice broken links caused by missing signals, constructing complete inter-port OD routes; discrete trajectories are completed using temporal interpolation, and multiple types of air pollutants and carbon emissions are calculated point-by-point, ultimately achieving emission estimation per AIS signal point at the single-voyage scale.

[0025] Example This embodiment provides a method for constructing an inter-port range-based vessel emission inventory based on AIS data. Experiments were conducted using real global vessel AIS trajectory data, a vessel static attribute database, and a global port coordinate dataset. The method was strictly implemented according to the steps outlined in this invention, as follows: Step 1: Distributed Acquisition and Preprocessing of Basic Data. Four types of basic data sources are read in parallel: minute-level AIS dynamic trajectory data of global ships, a static attribute library of ships including rated power, design speed, construction year, and fuel type, an emission factor set containing emission factors for 8 types of pollutants categorized by ship type / year / fuel, and a dataset of latitude and longitude coordinates of the center points of 2033 ports worldwide. The system dynamically allocates distributed computing partitions based on the original file size, extracting core fields such as MMSI, timestamp, longitude, latitude, and instantaneous speed; cleaning rules are set: records with coordinates exceeding the range [-180, 180] and latitude exceeding the range [-90, 90] are removed; missing records with any empty MMSI, timestamp, or latitude / longitude field are removed; redundant trajectory points with duplicate MMSI and timestamps are removed. Distributed aggregation is performed using MMSI as the grouping key. Trajectories within a single ship are sorted and normalized by ascending timestamp, and converted into a distributed columnar storage structure, providing normalized input for subsequent steps. The overall process of this step is shown in the attached figure. Figure 1 As shown.

[0026] Step 2: Determining Ship Operating Conditions and Generating Initial OD Segments. Read the preprocessed single-ship time-series trajectory, using 1.0 knots as the threshold for determining navigation / berthing status, and mark the ship's operating conditions point by point; set abnormal segment correction rules: abnormal short navigation segments with a duration ≤300 seconds are uniformly corrected to berthing status, and short berthing segments with a total duration <3600 seconds are uniformly corrected to navigation status; only valid berthing segments with a berthing duration ≥3600 seconds are retained for the port matching process. For valid berthing segments that have completed port matching, the preceding valid berthing port of adjacent navigation sub-segments is recorded as the starting point O, and the following valid berthing port is recorded as the ending point D, generating initial fragmented OD segments.

[0027] Step 3: High-precision dynamic identification of port nodes. During initialization, the port identification module preloads the latitude and longitude coordinates of the center points of 2033 ports and enables a fast spatial location index cache: the latitude and longitude of the trajectory points are rounded to four decimal places to generate a cache index. If the cache is hit, the matching result is returned directly; if no cache record is found, the complete spherical distance is calculated and stored in the cache. The input trajectory coordinates are first validated for latitude and longitude range. The Haversine formula is used to calculate the spherical distance between the trajectory point and the center of all ports point by point. A port buffer radius of 17 nautical miles is set, and only ports with a distance ≤ 17 nautical miles are included in the candidate matching set. If the candidate set is empty, no matching result is returned. If the candidate set contains only one port, it is directly determined as the assigned port. If the candidate set contains multiple ports, they are sorted in ascending order of spherical distance, and the port with the smallest distance is selected as the assigned port for that point. For the matching results of all trajectory points within the same berthing segment, a mode voting mechanism is used to select the port with the highest frequency of occurrence as the final assigned port for that berthing segment.

[0028] Step 4: Reconstructing Disconnected Ship Segments. Extract the start and end coordinates of two adjacent time-series ship segments. Calculate the spherical distance between the two points using the Haversine formula. Set a connectivity threshold of 100 nautical miles; if the distance is ≤100 nautical miles, the segment is considered spatially connected, and segment merging is performed. After merging, the average speed and ship attribute parameters of the two segments are inherited, and unknown port values ​​are removed to reconstruct a complete closed-loop OD (Original Destination) segment dataset with clearly defined origin and destination ports. The resulting matrix of voyages between the top 40 shipping countries worldwide after this step is attached. Figure 2 As shown, the frequency of shipping routes between East Asia, Europe, and Southeast Asia is significantly higher than that in other sea areas.

[0029] Step 5: Trajectory Interpolation and Refined Emission Calculation. Traverse all adjacent AIS trajectory points for a single ship. When the time difference between two points is >600 seconds, perform linear spatiotemporal interpolation based on time ratio to generate a continuous trajectory sequence with uniform time steps. Extract the design speed Vdesign from the ship's static database and calculate it using the formula LF=(V / Vdesign). 3 The main engine load factor LF is calculated, and LF=0.2 is set as the low load threshold. When LF<0.2, the corresponding low load emission correction factor is matched to amplify the main engine's basic emissions. The operating power of auxiliary engines and auxiliary boilers is matched according to the ship's operating conditions. The emission factors for eight pollutants (PM, NOx, SOx, CO, HC, CO2, N2O, CH4) are matched based on the ship type, year of construction, and fuel type. A polygonal vector map layer of the global ECA emission control area is overlaid, and spatial intersection is determined for each trajectory point. When the trajectory falls within the ECA range, the fuel correction factor FCF is activated to perform a secondary conversion correction on SOx and PM emissions. The emissions of each pollutant are calculated point by point using the unified formula E=(P×LF×EF×Δt) / C, where the dimensional conversion constant C is fixed at 3.6×10⁻⁶. 9 This is used to convert the mass unit of the power-dos-work calculation results from grams to tons and the time unit from seconds to hours. The CO2 emission matrix among the top 40 shipping countries worldwide, calculated in this step, is shown in the attached figure. Figure 3 As shown, the carbon emissions from air routes between China, the United States, Singapore, Japan, and other countries are significantly higher.

[0030] Step 6: Multi-dimensional aggregation and output of emission data. The latitude and longitude coordinates, which were scaled to integers during the calculation process to optimize storage, are restored to true geographic coordinates using the formulas Longitude=scale_lon×0.1 and Latitude=scale_lat×0.1, with a fixed spatial resolution of 0.1°. A distributed key-value pair is constructed, with grid latitude and longitude indexes, origin port, and destination port as composite keys, and the emission amounts of eight types of pollutants as values. The distributed reduceByKey reduction operator is called to accumulate the emission amounts of each pollutant in parallel along both the grid dimension and port pair dimension. The cluster summarizes the calculation results of each partition, outputting two types of standardized datasets: one is a global gridded ship emission inventory at 0.1° resolution, and the other is a port pair dimension inter-port OD segment emission statistics table. The output results can be directly integrated into a GIS system to draw shipping emission heat maps and inter-port emission correlation maps. The spatial distribution results of CO2 emissions at the inter-port OD scale are attached. Figure 4 As shown.

[0031] This invention provides a port-based, distance-class vessel emission inventory construction system based on AIS data, used to execute the aforementioned construction method, including: The data preprocessing module is used to perform the basic data distributed acquisition and preprocessing operations described in step 1 above; The operating condition determination module is used to perform the ship operating condition determination and initial OD segment generation operations described in step 2 above; The port matching module is used to perform the high-precision dynamic identification operation of port nodes described in step 3 above. The segment reconstruction module is used to perform the broken link segment reconstruction operation described in step 4 above. The emissions accounting module is used to perform the trajectory interpolation and refined emissions calculation operations described in step 5 above; The aggregation output module is used to perform the multi-dimensional aggregation output operation of emission data described in step 6 above.

[0032] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for constructing an inter-port range-class vessel emission inventory based on AIS data, characterized in that, Building an integrated system for trajectory analysis and emission accounting based on the Spark distributed computing framework includes the following steps: Step 1: Distributed Acquisition and Preprocessing of Basic Data: Parallel reading of global ship AIS dynamic trajectory data, ship static attribute database, emission factor database, and global port center point latitude and longitude dataset; distributed cleaning of raw AIS data, extraction of core fields such as ship unique identifier (MMSI), timestamp, latitude and longitude, and instantaneous speed, and removal of invalid records; sorting of individual ship trajectory points in ascending order by timestamp using MMSI as the grouping key to complete time series normalization; Step 2, Determining Ship Operating Conditions and Generating Initial OD Segments: Traverse the time-series trajectory points of a single ship, dynamically mark the navigation / berthing status based on the speed threshold, correct abnormal short segments, and filter valid berthing segments; Step 3, High-precision dynamic identification of port nodes: Based on the pre-loaded coordinates of global port center points, the port of origin is determined for each valid berthing segment by calculating the spherical distance and matching the buffer radius; the initial OD segment is generated based on the ports of origin of adjacent valid berthing segments before and after the navigation segment. Step 4, Disconnected Link Segment Reconstruction: For disconnected links caused by missing AIS signals, adjacent segments are merged based on spatial connectivity rules to generate a complete closed-loop inter-port OD segment dataset. Step 5, Trajectory Interpolation and Refined Emission Calculation: Spatiotemporal interpolation is performed on adjacent AIS trajectory points where the time interval exceeds the interpolation threshold. The main engine load factor is calculated based on the ship's static parameters and low load correction is performed. The emission factors of the three types of emission sources, namely main engine, auxiliary engine and auxiliary boiler, are matched and matched. The fuel oil correction is performed by superimposing emission control area rules. The emissions of various air pollutants and greenhouse gases are calculated for each interpolation point. Step 6: Multi-dimensional aggregation and output of emission data: Discrete emission points are aggregated separately according to the set spatial resolution grid and origin and destination port pairs, and the global gridded ship emission inventory and inter-port OD segment emission inventory are output.

2. The construction method according to claim 1, characterized in that, The distributed cleaning of raw AIS data in step 1 specifically includes: The distributed computing partition is dynamically set according to the original file size, and the core fields of MMSI, timestamp, longitude, latitude and instantaneous speed are extracted. Records with coordinates exceeding the range of [-180, 180] and [-90, 90] are removed, as are records with missing core fields and redundant trajectory points with duplicate MMSI and timestamp. All trajectory points of a single ship are aggregated using a distributed aggregation operator with MMSI as the grouping key, and time-series normalization is completed by sorting the timestamps in ascending order, and the data is converted into a distributed columnar storage structure.

3. The construction method according to claim 1, characterized in that, Step 2, which involves determining the ship's operating conditions and correcting abnormal voyage segments, specifically includes: The instantaneous speed of each trajectory point is compared with the speed determination threshold. Points with speed ≤ threshold are marked as anchored and points with speed > threshold are marked as sailing. Based on this, the continuous trajectory is divided into alternating anchored sub-segments and sailing sub-segments. Abnormal short navigation segments with a duration ≤ the first preset time are corrected to a berthing state; berthing sub-segments with a duration ≥ the second preset time are selected as valid berthing segments, and short berthing sub-segments with a duration < the second preset time are corrected to a navigation state.

4. The construction method according to claim 1, characterized in that, The high-precision dynamic identification of port nodes in step 3 specifically includes: Step 31: During the module initialization phase, preload a dataset containing the latitude and longitude coordinates of the center points of 2033 physical ports worldwide and enable a fast spatial location indexing and caching mechanism. Step 32: Perform latitude and longitude range verification on the input trajectory points. After the verification is passed, use the Haversine formula to calculate the spherical distance between the trajectory points and the center points of each port. Step 33: Set the port buffer radius and include ports whose spherical distance from the trajectory point is ≤ the buffer radius into the candidate matching set; Step 34: If the candidate matching set is empty, return a result of no port matching; if the candidate set contains only one port, directly determine the port to which the point belongs; if the candidate set contains multiple ports, sort them in ascending order of spherical distance, and select the port with the smallest distance as the unique port to which the trajectory point belongs. Step 35: For the port matching results of all trajectory points within the same berthing segment, a mode voting mechanism is used to select the port with the highest frequency of occurrence as the final destination port for that berthing segment.

5. The construction method according to claim 1, characterized in that, Step 4, the reconstruction of the interrupted link segment, specifically includes: Extract the end point and beginning point of two temporally adjacent segments belonging to the same vessel, and calculate the spherical distance between the two points using the Haversine formula. If the distance is less than or equal to the spatial connectivity threshold, merge the two segments into a complete OD segment, inherit the vessel attribute parameters and average speed of the original segment, remove the empty values ​​of unknown ports in the middle, and obtain a closed-loop inter-port OD segment dataset with clear origin and destination ports.

6. The construction method according to claim 1, characterized in that, Step 5, specifically the trajectory interpolation and refined emission calculation, includes: Step 51: Traverse all adjacent AIS trajectory points of a single ship, and use linear spatiotemporal interpolation based on time scaling factor for adjacent trajectory points with time interval > 600 seconds to generate a continuous trajectory sequence with uniform time step. Step 52: Extract the ship's design speed Vdesign from the ship's static attribute library, and calculate the design speed V based on the instantaneous speed V at the current trajectory point using the formula LF=(V / Vdesign). 3 / 100 Calculate the host load factor LF; when LF<0.2, trigger low load correction and match the corresponding low load emission correction coefficient to calibrate the host base emission; Step 53: Match the real-time operating power of auxiliary engines and auxiliary boilers according to the current operating conditions of the ship, and match the corresponding emission factors of 8 types of pollutants in combination with the ship type, construction year and fuel type. The 8 types of pollutants include PM, NOx, SOx, CO, HC, CO2, N2O and CH4. Step 54: Use the global emission control area vector polygon to determine the spatial intersection of each trajectory point. If the trajectory point falls within the emission control area, trigger the fuel switching simulation logic and use the fuel correction coefficient to perform a secondary conversion and correction of SOx and PM emissions. Step 55: Calculate the emissions of each pollutant point by point according to the unified formula E=(P×LF×EF×Δt) / C, where P is the rated power of the corresponding emission source, EF is the pollutant emission factor, Δt is the time step of the trajectory point, and C is the dimension conversion constant.

7. The construction method according to claim 1, characterized in that, The multi-dimensional aggregation output of emission data in step 6 specifically includes: The latitude and longitude that were stored using integer scaling during the calculation process to optimize storage are restored to the true geographic latitude and longitude according to the formulas Longitude=scale_lon×res and Latitude=scale_lat×res, where res is the preset spatial resolution; Construct a distributed key-value pair dataset with grid latitude and longitude index, starting port name, and ending port name as composite keys, and 8 types of pollutant emission vectors as values; The distributed reduction operator is invoked to accumulate the emissions of each pollutant in parallel according to the spatial grid dimension and the origin and destination port pair dimension, respectively, and outputs a global gridded ship emission inventory with a resolution of 0.1°, as well as an inter-port OD segment emission statistics table with the physical port pair dimension. The output results can be directly connected to the geographic information system for spatial visualization.

8. The construction method according to claim 4, characterized in that, The spatial location fast index caching mechanism in step 31 is as follows: the latitude and longitude of the trajectory point are rounded to 4 decimal places to generate a unique cache index. When the cache is hit, the pre-stored port matching result is returned directly. When the cache is not hit, the full spherical distance is calculated and the matching result is stored in the cache.

9. A system for constructing an inter-port range-class vessel emission inventory based on AIS data, used to execute the construction method described in claim 1, characterized in that, include: The data preprocessing module is used to perform distributed acquisition and preprocessing operations on basic data; The operating condition determination module is used to perform ship operating condition determination and initial OD segment generation operations; The port matching module is used to perform high-precision dynamic identification of port nodes; The segment reconstruction module is used to perform segment reconstruction operations for disconnected links; The emissions accounting module is used to perform trajectory interpolation and refined emissions calculation operations; The aggregation output module is used to perform multi-dimensional aggregation output operations on emission data.

Images