A method and system for intelligent detection of abnormal navigation behavior of ships in key waterways

Abstract

This application discloses an intelligent detection method and system for abnormal navigation behavior of vessels in key waterways, relating to the field of smart waterway supervision technology. The method includes: acquiring basic control data of the target key waterway and multi-source navigation perception data of the vessels to be detected; determining abnormal navigation behavior detection rules based on the basic control data; performing vessel target identification and abnormal navigation behavior detection on the multi-source navigation perception data respectively, obtaining multiple preliminary detection results; associating the same vessel target, and fusing and verifying the multiple preliminary detection results based on the association results to obtain the abnormal navigation behavior detection result; continuously tracking and updating the detection results based on other data sources when the data availability status of any data source meets preset abnormal conditions; and outputting the detection result. This application can improve the accuracy of identifying complex navigation behavior in key waterways and the ability for continuous dynamic monitoring of vessels.

Description

Technical Field

[0001] This application relates to the field of intelligent waterway transport supervision technology, and in particular to an intelligent detection method and system for abnormal navigation behavior of ships in key waterways. Background Technology

[0002] With the continuous increase in inland waterway traffic volume, the demand for navigation safety supervision in key waterways such as bridge areas, ferry crossings, reservoir areas, and shoals is increasing. These waterways are typically characterized by complex channel conditions, dense vessel traffic, significant differences in navigation rules, and high risks of unforeseen events. Vessel speed, course, track, mooring position, and traffic status in these waterways can all affect navigation safety. To improve waterway traffic supervision capabilities, existing intelligent inland waterway monitoring systems typically utilize information technology such as Automatic Identification System (AIS), BeiDou positioning, video surveillance, and electronic nautical charts to collect, display, and analyze vessel dynamics, assisting supervisors in understanding vessel navigation status.

[0003] In related technologies, inland waterway vessel navigation monitoring typically relies on vessel positioning data, trajectory data, or video image data to identify and alert on behaviors such as speeding, deviation from course, illegal mooring, abnormal berthing and departure, and track deviation. However, the actual navigation environment in key waterways is complex. Vessel dynamic data is easily affected by factors such as terminal equipment status, signal obstruction, data latency, and dense vessel distribution. Video image data is also easily affected by factors such as nighttime lighting, rain and fog, obstructions, and shooting angles, leading to missed or false detections in abnormal behavior identification. Furthermore, navigation conditions, regulatory rules, and risk types vary across different key waterways. Existing detection methods primarily focus on judging routine navigation status and lack the adaptability to complex navigation behaviors, special waterway control requirements, and continuous vessel dynamic tracking, making it difficult to reflect abnormal navigation risks in key waterways in a timely and accurate manner.

[0004] Therefore, in the intelligent detection of abnormal navigation behavior of vessels in key inland waterways, the reliability of vessel dynamic data is easily affected by the scene environment, the accuracy of complex navigation behavior identification is insufficient, the adaptability to the differences in navigation control in different waterways is weak, and the ability to continuously track vessel dynamics is insufficient, which have become urgent problems to be solved. Summary of the Invention

[0005] This application provides an intelligent detection method and system for abnormal navigation behavior of vessels in key waterways, aiming to solve the problems in intelligent detection of abnormal navigation behavior of vessels in key inland waterways, such as the reliability of vessel dynamic data being easily affected by the scene environment, insufficient accuracy in identifying complex navigation behavior, weak adaptability to differences in navigation control in different waterways, and insufficient ability to continuously track vessel dynamics.

[0006] Firstly, a method for intelligent detection of abnormal navigation behavior of vessels in key waterways, the method comprising: Acquire basic control data for key target waters and multi-source navigation perception data of vessels to be inspected; Based on the aforementioned basic control data, abnormal navigation behavior detection rules corresponding to the target key waterways are determined; The multi-source navigation perception data were used to perform ship target identification and abnormal navigation behavior detection, resulting in several preliminary detection results. The same ship target in the multi-source navigation perception data is associated, and the multiple preliminary detection results are fused and verified based on the association results to obtain the abnormal navigation behavior detection results of the ship to be detected; If the data availability status of any data source in the multi-source navigation perception data meets the preset abnormal conditions, the vessel to be detected is continuously tracked based on other data sources, and the abnormal navigation behavior detection result is updated according to the continuous tracking result. Output the abnormal navigation behavior detection result or the updated abnormal navigation behavior detection result.

[0007] Optionally, the above scheme may involve acquiring basic management data for the target key water areas, including: Electronic navigation map data and geographic information data of the target key water area are obtained to obtain basic spatial data of the water area; Based on the aforementioned basic spatial data of the water area, the spatial control range of the target key water area is determined, and the spatial control data of the water area is obtained. Obtain navigation restriction information, passage rule information, and risk area information corresponding to the target key waterway to obtain waterway navigation control data; Obtain the static file information of the vessel, and associate the static file information of the vessel with the water space control data and the water navigation control data to obtain the basic control data.

[0008] Optionally, in the above scheme, based on the basic control data, abnormal navigation behavior detection rules corresponding to the target key waterway are determined, including: Based on the water space control data, the water type of the target key water area is determined, and the water type identification result is obtained; Based on the water area type identification results, the corresponding candidate detection rules are matched from the preset rule base to obtain a candidate rule set; The rule parameters in the candidate rule set are configured based on the waterway navigation control data to obtain the rule parameter configuration result; Based on the rule parameter configuration results, abnormal navigation behavior detection rules corresponding to the target key water area are determined.

[0009] Optionally, in the above scheme, if the water type identification result indicates that the target key water area is a bridge area, an abnormal navigation behavior detection rule corresponding to the target key water area is determined, including: Obtain navigation channel information, bridge area boundary information, and bridge area traffic restriction information corresponding to the water area of ​​the bridge area to obtain bridge area control and constraint data; Based on the bridge area control and constraint data, the standard passage status of vessels in the bridge area waters is determined, and the standard passage status information of the bridge area is obtained. Based on the bridge area standard traffic status information, configure at least one detection rule among anchoring, U-turn, overtaking, speeding, wrong navigation channel, reverse navigation, sudden navigation, and encounter risk to obtain the bridge area rule configuration result; Based on the bridge area rule configuration results, abnormal navigation behavior detection rules for the bridge area are generated, and the navigation status of ships in the bridge area is detected in parallel according to the abnormal navigation behavior detection rules for the bridge area to obtain abnormal navigation behavior detection results for the bridge area.

[0010] Optionally, in the above scheme, if the water type identification result indicates that the target key water area is a ferry crossing, a reservoir, or a shallow water area, an abnormal navigation behavior detection rule corresponding to the target key water area is determined, including: Based on the water area type identification results, the corresponding scene control object is determined, and scene control object information is obtained; Based on the information of the scene control object, the corresponding scene abnormal behavior type is determined, and a set of scene abnormal behaviors is obtained; Based on the waterway navigation control data, corresponding judgment conditions are configured for the set of abnormal behaviors in the scenario to obtain scenario-based detection rules; Based on the aforementioned scenario-based detection rules, the navigation status of ships in the target key waters is detected, and the abnormal navigation behavior detection results for different waters are obtained.

[0011] In the above scheme, optionally, ship target identification and abnormal navigation behavior detection are performed on the multi-source navigation perception data to obtain multiple preliminary detection results, including: The multi-source navigation perception data is classified into data source types to obtain positioning navigation data and image navigation data. Based on the positioning-type navigation data, at least one navigation state feature among the ship's position, speed, heading, and trajectory is extracted to obtain the first navigation state feature; Based on the abnormal navigation behavior detection rules and the first navigation state features, abnormal navigation behavior is detected to obtain a first preliminary detection result. Based on the image-type navigation data, ship target identification and motion state analysis are performed to obtain the second navigation state characteristics; Based on the abnormal navigation behavior detection rules and the second navigation state characteristics, abnormal navigation behavior is detected to obtain a second preliminary detection result.

[0012] Optionally, in the above scheme, associating the same ship target in the multi-source navigation perception data includes: Based on the positioning navigation data, the spatiotemporal state information of the first vessel target is determined, and the description information of the first target is obtained; Based on the image-type navigation data, the image recognition features of the second ship target are determined, and the description information of the second target is obtained; Based on the first target description information and the second target description information, candidate ship target matching relationships are generated to obtain candidate association results; The consistency evaluation of the candidate association results is performed to obtain the target association evaluation result; The association results of the same ship target in different data sources are determined based on the target association evaluation results.

[0013] Optionally, in the above scheme, the multiple preliminary detection results are fused and verified based on the correlation results to obtain the abnormal navigation behavior detection results of the vessel to be detected, including: Based on the association results, the first preliminary detection result and the second preliminary detection result are mapped to the same ship target to obtain a target detection result set; A consistency judgment is made on the abnormal behavior type, abnormal occurrence area, and abnormal occurrence time in the target detection result set to obtain a consistency judgment result; If the consistency judgment result meets the preset confirmation conditions, it is confirmed that the same ship target has corresponding abnormal navigation behavior, and an abnormal event is confirmed. If the consistency judgment result does not meet the preset confirmation condition, a target verification data source is determined from the data source corresponding to the multi-source navigation perception data, a verification instruction is generated for the target verification data source, and the processing result of the suspected abnormal event is determined according to the verification result returned by the target verification data source. Based on the processing results of the confirmed abnormal events and the suspected abnormal events, the detection results of the abnormal navigation behavior of the vessel to be detected are obtained.

[0014] Optionally, in the above scheme, if the data availability status of any data source in the multi-source navigation perception data meets a preset anomaly condition, the vessel to be detected is continuously tracked based on other data sources, and the abnormal navigation behavior detection result is updated according to the continuous tracking result, including: Availability monitoring is performed on each data source in the multi-source navigation perception data to obtain the data source availability status; If any data source meets a preset abnormal condition as indicated by the data source availability status, a data source abnormal event is generated. Based on the abnormal events of the data source, a target data source that can be used for continuous tracking is determined, and the tracking data source selection result is obtained; Based on the tracking data source selection results, the vessel to be detected is identified and its navigation status is continuously analyzed to obtain continuous tracking results. The abnormal navigation behavior detection results are updated based on the continuous tracking results, and graded alarm information is generated based on the updated abnormal navigation behavior detection results.

[0015] Secondly, an intelligent detection system for abnormal navigation behavior of vessels in key waterways, the system comprising: The data acquisition module is used to acquire basic control data for key target waters and multi-source navigation perception data of vessels to be monitored. The rule determination module is used to determine the abnormal navigation behavior detection rules corresponding to the target key water area based on the control basic data; The preliminary detection module is used to perform ship target identification and abnormal navigation behavior detection on the multi-source navigation perception data, and obtain multiple preliminary detection results; The target association module is used to associate the same ship target in the multi-source navigation perception data to obtain the association result; The fusion verification module is used to perform fusion verification on the multiple preliminary detection results based on the correlation results to obtain the abnormal navigation behavior detection results of the vessel to be detected; The continuous tracking module is used to continuously track the vessel to be detected based on other data sources when the data availability status of any data source in the multi-source navigation perception data meets the preset abnormal conditions, and update the abnormal navigation behavior detection results according to the continuous tracking results. The result output module is used to output the abnormal navigation behavior detection result or the updated abnormal navigation behavior detection result.

[0016] Compared with the prior art, this application has at least the following beneficial effects: This application, based on further analysis and research of existing technical problems, recognizes that in the intelligent detection of abnormal navigation behavior of vessels in key inland waterways, the reliability of vessel dynamic data is easily affected by the scene environment, the accuracy of complex navigation behavior identification is insufficient, the adaptability to differences in navigation control in different waterways is weak, and the ability to continuously track vessel dynamics is insufficient. By acquiring basic control data of the target key waterway and multi-source navigation perception data of the vessels to be detected, the detection of abnormal navigation behavior can simultaneously combine waterway control requirements and the real-time navigation status of vessels. By determining abnormal navigation behavior detection rules based on the basic control data, the detection rules can be adapted to the actual navigation conditions of the target key waterway. By separately performing vessel target identification and abnormal navigation behavior detection on multi-source navigation perception data, preliminary detection basis from different data sources can be formed. Furthermore, through… By associating the same vessel target and fusing and verifying multiple preliminary detection results, false positives and false negatives caused by the influence of environment, equipment status, or data quality on a single data source can be reduced. Simultaneously, when the data availability status of any data source meets preset abnormal conditions, the vessel under test can be continuously tracked and the abnormal navigation behavior detection results updated through other data sources. This allows for continuous tracking of vessel dynamics even when some data sources are unavailable, thereby improving the accuracy of complex navigation behavior identification, the adaptability of detection in different key waterways, and the continuity of abnormal navigation supervision. This addresses the problems in intelligent detection of abnormal navigation behavior in key inland waterways where the reliability of vessel dynamic data is easily affected by the scene environment, the accuracy of complex navigation behavior identification is insufficient, the adaptability to differences in navigation control in different waterways is weak, and the ability to continuously track vessel dynamics is inadequate. Attached Figure Description

[0017] Figure 1 A flowchart illustrating an intelligent detection method for abnormal navigation behavior of vessels in key waterways, provided as an embodiment of this application; Figure 2 This is an overall flowchart of an intelligent detection method for abnormal navigation behavior of vessels in key waterways provided in one embodiment of this application; Figure 3 This is a schematic diagram of a dual-engine multi-source fusion collaborative detection architecture provided in one embodiment of this application; Figure 4 This is a flowchart of a bridge area multi-rule collaborative detection algorithm provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] like Figure 1 and Figure 2 As shown in this embodiment, an intelligent detection method for abnormal navigation behavior of vessels in key waterways is provided. This method can be applied to inland waterway intelligent water transport supervision systems, key waterway traffic supervision platforms, bridge area navigation safety supervision platforms, or vessel dynamic supervision platforms. The key waterways may include at least one of bridge area waterways, ferry crossing waterways, reservoir waterways, and shallow waterways, and may also include other inland waterways requiring special navigation control. The vessels to be detected may be vessels entering or about to enter the target key waterway.

[0020] In one possible embodiment, basic control data for the target key waterway and multi-source navigation perception data of the vessel to be monitored are first acquired. The basic control data may include the spatial extent of the target key waterway, navigation boundaries, navigation restrictions, risk areas, prohibited areas, traffic rules, and vessel static profile information. The multi-source navigation perception data may include positioning-based navigation data and image-based navigation data. Positioning-based navigation data may originate from AIS data, BeiDou positioning data, or other vessel positioning data, while image-based navigation data may originate from CCTV video surveillance systems or other video acquisition equipment. AIS is an abbreviation for Automatic Identification System; CCTV is an abbreviation for Closed-Circuit Television.

[0021] After acquiring the basic control data, abnormal navigation behavior detection rules are determined based on this data for the target key waterways. Specifically, the control targets and abnormal behavior types for different types of key waterways are not entirely the same. For example, in bridge areas, the focus can be on whether vessels are mistaking navigation channels, sailing backwards, changing course abruptly, encountering risks, speeding, overtaking, anchoring, or turning around; in ferry crossing areas, the focus can be on illegal ferry operations and ferries sailing beyond their permitted areas; in reservoir areas, the focus can be on passenger vessel route compliance and illegal mooring; and in shoal areas, the focus can be on large-tonnage vessels illegally entering or mooring in prohibited areas. Therefore, corresponding abnormal navigation behavior detection rules can be determined based on the spatial boundaries, control requirements, and vessel types of the target key waterways.

[0022] After obtaining the abnormal navigation behavior detection rules, ship target identification and abnormal navigation behavior detection are performed on the multi-source navigation perception data to obtain multiple preliminary detection results. For example, the current position, speed, heading, and historical trajectory of the ship can be identified through positioning-based navigation data, and combined with the abnormal navigation behavior detection rules to obtain a first preliminary detection result; simultaneously, ship targets, ship movement trends, ship nameplate information, and traversed areas in video footage can be identified through image-based navigation data, and combined with the abnormal navigation behavior detection rules to obtain a second preliminary detection result. The multiple preliminary detection results can represent the judgment results of abnormal navigation behavior of the same ship or different ships from different data sources.

[0023] Furthermore, the same vessel target in the multi-source navigation perception data is associated, and the multiple preliminary detection results are fused and verified based on the association results to obtain the abnormal navigation behavior detection results of the vessel to be detected. Specifically, vessel targets in different data sources can be matched and bound according to features such as time information, spatial information, vessel size, vessel type, vessel outline, and vessel name recognition results to determine whether the vessel target in the positioning navigation data and the vessel target in the image navigation data are the same vessel. After the association is completed, the preliminary detection results output from different data sources are mapped to the same vessel target, and the abnormal behavior type, abnormal occurrence time, abnormal occurrence area, and vessel target identity are fused and verified to obtain more reliable abnormal navigation behavior detection results.

[0024] If the availability of any data source in the multi-source navigation sensing data meets preset anomaly conditions, the vessel under inspection is continuously tracked based on other data sources, and the abnormal navigation behavior detection results are updated according to the continuous tracking results. The preset anomaly conditions may include interruption of positioning-type navigation data, AIS / BeiDou terminal offline, abnormal shutdown of AIS / BeiDou terminal, missing positioning signal, abnormal jumps in positioning trajectory, video image obstruction, and video recognition confidence level lower than a preset value. When positioning-type navigation data is unavailable, image-type navigation data can be switched to or enhanced for vessel identification and continuous navigation status analysis; when image-type navigation data is unavailable, positioning-type navigation data can be enhanced for continuous trajectory analysis. Therefore, even if some data sources are abnormal, continuous monitoring of the vessel under inspection can be maintained.

[0025] Finally, the abnormal navigation behavior detection results or updated abnormal navigation behavior detection results are output. The abnormal navigation behavior detection results may include vessel identification, abnormal behavior type, time of occurrence, location of occurrence, duration of occurrence, severity of occurrence, detection basis, and alarm level. The detection results can be output to the monitoring system for visualization, historical review, alarm push notifications, and report export. They can also serve as the basis for subsequent manual review, enforcement actions, or risk warnings.

[0026] This embodiment acquires basic control data and multi-source navigation perception data for key target waters, enabling abnormal navigation behavior detection to simultaneously combine waterway control requirements and vessel dynamic status. By determining detection rules based on basic control data, the detection process can adapt to the differences in navigation control in different key waters. By detecting separately and then associating and merging the same vessel target for verification, it can reduce missed detections and false detections caused by the influence of environmental or equipment status on a single data source. By continuously tracking based on other data sources when a data source is abnormal, it can improve the ability to continuously track vessel dynamics, thereby enhancing the accuracy, stability, and regulatory continuity of abnormal navigation behavior detection in key waters.

[0027] In one possible embodiment, acquiring basic data on the control of the target key water area may include the following process.

[0028] First, electronic navigation chart data and geographic information data for the target key waterways are acquired to obtain the basic spatial data of the waterway. The electronic navigation chart data may include information such as the channel centerline, channel boundary lines, bridge locations, pier locations, shoal areas, wharf locations, ferry crossing locations, navigation mark locations, and navigation channel locations. The geographic information data can be sourced from a GIS (Geographic Information System). This geographic information data may include latitude and longitude coordinates, shoreline data, waterway boundaries, administrative region boundaries, regulatory region boundaries, and spatial coordinates of geographical features. By performing unified coordinate processing on the electronic navigation chart data and geographic information data, the basic spatial data of the waterway can be obtained.

[0029] Then, based on the aforementioned basic spatial data of the water area, the spatial control range of the target key water area is determined, resulting in water area spatial control data. The spatial control range can be represented in the form of a set of latitude and longitude coordinates, a polygonal region, an electronic fence region, a centerline buffer zone, a prohibited area, or a risk area. For example, for bridge areas, the bridge control range can be determined based on the bridge location, the navigation channel location, and the upstream and downstream control distances of the bridge area; for ferry crossings, the ferry control range can be determined based on the ferry crossing's operating range, ferry routes, and berthing areas; for reservoir areas, the reservoir control range can be determined based on the reservoir's channel boundaries, passenger routes, and restricted berthing areas; and for shoal areas, the shoal control range can be determined based on shoal boundaries, prohibited areas, and navigational restriction areas.

[0030] Next, navigation restriction information, traffic rules information, and risk area information corresponding to the target key waterways are obtained to obtain waterway navigation control data. The navigation restriction information may include bridge navigation aperture size, navigation clearance height, navigation clearance width, permitted vessel types, maximum permitted tonnage, speed limits, encounter restrictions, and overtaking restrictions; the traffic rules information may include vessel traffic direction, prescribed routes, operating range, berthing restrictions, and prohibited entry requirements; the risk area information may include prohibited entry areas in shoals, risk areas near bridge piers, high-risk encounter areas, risk areas at ferry crossings, and key monitored areas in reservoir areas. By structuring the above information, waterway navigation control data can be obtained.

[0031] Finally, the static file information of the vessel is obtained and correlated with the water space control data and waterway navigation control data to obtain the basic control data. The static file information may include basic vessel information, vessel name, vessel type, vessel length, vessel width, vessel tonnage, shipping company affiliation information, operating attributes, approved routes, and safety supervision attributes. By correlating the static file information with the water space control data and waterway navigation control data, unified and standardized basic control data can be formed, providing fundamental data support for subsequent identification of key target waterways, determination of detection rules, vessel identity verification, and judgment of abnormal behavior.

[0032] This embodiment determines the basic spatial data of the water area through electronic waterway chart data and geographic information data, and further forms water area spatial control data, water area navigation control data and ship static file association data. This enables the subsequent detection of abnormal navigation behavior to no longer rely solely on real-time sensing data, but to be supported by clear water area boundaries, navigation rules and ship basic attributes, thereby improving the accuracy of detection rule configuration and the standardization of key water area supervision.

[0033] In one possible embodiment, determining abnormal navigation behavior detection rules corresponding to the target key waters based on basic control data may include the following process.

[0034] First, the water type of the target key water area is determined based on the water space management data, resulting in water type identification results. Specifically, the target key water area can be identified as a bridge area, ferry crossing area, reservoir area, shoal area, or other preset key water type based on its spatial management scope, area identification, regulatory attributes, geographical features, and navigation functions. The water type identification results can be used to indicate the regulatory scenario category of the target key water area.

[0035] Then, based on the water area type identification results, corresponding candidate detection rules are matched from a preset rule base to obtain a candidate rule set. The preset rule base may include abnormal navigation behavior detection rule templates corresponding to different water area types. For example, candidate detection rules for bridge areas may include detection rules for anchoring, U-turns, overtaking, speeding, mistaking navigation channels, reverse navigation, sudden navigation changes, and encountering risks; candidate detection rules for ferry crossings may include detection rules for illegal ferry operations and ferries sailing beyond their permitted range; candidate detection rules for reservoir areas may include detection rules for passenger vessel route compliance and illegal mooring; and candidate detection rules for shoals may include detection rules for large-tonnage vessels illegally entering and mooring in prohibited areas.

[0036] Next, the rule parameters in the candidate rule set are configured based on the waterway navigation control data to obtain the rule parameter configuration results. These rule parameters may include area boundaries, speed thresholds, heading change thresholds, dwell time thresholds, encounter distance thresholds, trajectory deviation thresholds, permitted vessel types, permitted tonnage, navigable passage number, designated route range, operating range, and alarm level thresholds, etc. Different target key waterways can be configured with different rule parameters according to actual navigation conditions and regulatory requirements. For example, different bridges have different navigable passage sizes and speed limits; therefore, the navigable passage boundaries, speed thresholds, and traffic directions in the bridge area detection rules can be configured based on the waterway navigation control data.

[0037] Finally, based on the rule parameter configuration results, abnormal navigation behavior detection rules corresponding to the target key waterways are determined. These abnormal navigation behavior detection rules can be a set of one or more rules used to detect the navigation status of vessels within the target key waterways. These rules can be enabled, disabled, have thresholds adjusted, and dynamically updated to adapt to the regulatory needs of different regulatory areas, different navigation times, and different vessel types.

[0038] This embodiment identifies water area types through water space management data, matches candidate detection rules from a preset rule base, and configures rule parameters by combining water navigation management data. This allows the abnormal navigation behavior detection rules to match the actual navigation conditions of the target key water areas, avoiding insufficient scenario adaptation caused by using uniform rules, and improving the pertinence and accuracy of abnormal navigation behavior identification in different key water areas.

[0039] In one possible embodiment, such as Figure 4 As shown, when the water area type identification results indicate that the target key water area is a bridge area, the abnormal navigation behavior detection rules can be determined according to the navigation characteristics of the bridge area.

[0040] First, information on navigation channels, bridge area boundaries, and traffic restrictions in the bridge area is obtained to generate bridge area control and constraint data. The navigation channel information may include the location, number, clear width, clear height, direction of navigation, permitted vessel types, permitted vessel tonnage, and centerline of each navigation channel. The bridge area boundary information may include the upstream and downstream control boundaries of the bridge, the pier protection area, and the area covered by the electronic fence. The traffic restrictions may include speed limits, prohibitions on mooring, prohibitions on U-turns, prohibitions on overtaking, meeting restrictions, and traffic restrictions corresponding to different navigation channels.

[0041] Then, based on the bridge area control and constraint data, the standard passage status of vessels within the bridge area is determined, resulting in bridge area standard passage status information. This information may include the navigation channels a vessel should enter, permitted navigation direction, permitted speed range, permitted track range, permitted encounter distance, permitted dwell time, and the range of normal track variations within the bridge area. For vessels entering the bridge area, the standard passage status they must adhere to can be determined based on their vessel type, size, tonnage, course, and current position.

[0042] Next, based on the standard traffic status information of the bridge area, at least one detection rule is configured for anchoring, U-turn, overtaking, speeding, wrong-way navigation, reverse navigation, sudden change of course, and encounter risk, resulting in the bridge area rule configuration result. Specifically, the anchoring detection rule can be configured based on the state where the vessel's speed within the bridge area is consistently below a preset threshold and its position change is less than a preset range; the U-turn detection rule can be configured based on the angle of change of course and trajectory turning characteristics; the overtaking detection rule can be configured based on the relative position, relative speed, and distance changes between vessels traveling in the same direction; the speeding detection rule can be configured based on the vessel's real-time speed and the bridge area speed limit threshold; the wrong-way navigation detection rule can be configured based on the deviation between the vessel's actual passage through the navigation channel and the standard navigation channel; the reverse navigation detection rule can be configured based on the reverse relationship between the vessel's course and trajectory direction; the sudden change of course detection rule can be configured based on the degree of sudden change in course, speed, or trajectory curvature within a short period; and the encounter risk detection rule can be configured based on the distance between different vessels, the course intersection relationship, relative speed, and the expected encounter position.

[0043] Finally, based on the bridge area rule configuration results, abnormal navigation behavior detection rules for the bridge area are generated. These rules are then used to perform parallel detection of vessel navigation status within the bridge area, yielding the detection results. During the detection process, multiple types of bridge area detection rules can be run in parallel, and suspected abnormal behaviors matched by a single type of rule can be cross-verified. For example, when a vessel exhibits track deviation, mismatched navigation channels, or abrupt changes in course, a comprehensive assessment can be made to determine if it has taken the wrong navigation channel or exhibited abrupt navigation behavior. Similarly, when the relative distance between two vessels continuously decreases within the bridge area and their expected meeting point is within the bridge area's control zone, a determination can be made to identify any potential encounter risks.

[0044] This embodiment determines the standard passage status of the bridge area through bridge area control constraint data, and configures multiple abnormal navigation behavior detection rules for the bridge area accordingly. This makes bridge area detection no longer limited to a single speeding or deviation judgment, but can perform parallel detection and comprehensive judgment of complex behaviors such as anchoring, U-turn, overtaking, speeding, wrong navigation channel, reverse navigation, sudden navigation, and encountering risks, thereby improving the completeness and accuracy of abnormal behavior identification in complex navigation scenarios in the bridge area.

[0045] In one possible embodiment, if the water type identification result indicates that the target key water area is a ferry crossing, a reservoir, or a shallow water area, abnormal navigation behavior detection rules can be determined based on the regulatory focus of the corresponding water area.

[0046] First, based on the water area type identification results, the corresponding scene control objects are determined, and scene control object information is obtained. For ferry crossing waters, the scene control objects may include ferries, ferry routes, ferry crossing operating areas, berthing areas, and areas at risk of illegal ferry operations; for reservoir waters, the scene control objects may include passenger ships, designated routes, berthing areas, no-stopping areas, and reservoir channel boundaries; for shoal waters, the scene control objects may include large-tonnage vessels, shoal boundaries, prohibited areas, navigation-restricted areas, and areas at risk of grounding.

[0047] Then, based on the information of the scene control objects, the corresponding scene abnormal behavior types are determined, resulting in a scene abnormal behavior set. For ferry crossing waters, the scene abnormal behavior set may include illegal ferry operations, ferries sailing beyond their designated areas, and non-ferry vessels entering high-risk ferry areas; for reservoir waters, the scene abnormal behavior set may include passenger vessels violating route regulations, passenger vessels deviating from their designated routes, and illegal berthing; for shoal waters, the scene abnormal behavior set may include large-tonnage vessels illegally entering shoal areas, vessels berthing in prohibited areas, and vessels entering areas at risk of running aground.

[0048] Next, based on the waterway navigation control data, corresponding judgment conditions are configured for the set of abnormal behaviors in the scenario, resulting in scenario-based detection rules. These judgment conditions may include spatial boundary judgment conditions, vessel type judgment conditions, vessel tonnage judgment conditions, route deviation judgment conditions, operating range judgment conditions, berthing duration judgment conditions, and prohibited area judgment conditions. For example, for ferries sailing beyond their permitted range, judgment can be made based on the positional relationship between the ferry terminal's operating range and the ferry's trajectory; for passenger vessels violating route regulations, judgment can be made based on the deviation distance between the passenger vessel's trajectory and the prescribed route; for large-tonnage vessels illegally entering the area, judgment can be made based on the spatial relationship between the vessel's tonnage information and the prohibited area in the shoal.

[0049] Finally, based on scenario-based detection rules, the navigation status of vessels in key target waterways is detected, yielding abnormal navigation behavior detection results for each waterway. These results may include waterway type, vessel target, abnormal behavior type, abnormal location, abnormal time, and rule hit information, which are used for subsequent fusion verification, tiered alerts, and regulatory display.

[0050] This embodiment identifies the scene control objects based on the water area type identification results, and further determines the scene abnormal behavior set and scene-based detection rules, so that different key water areas can be detected differently according to their respective navigation characteristics and regulatory priorities. This improves the identification ability of subdivided scenes such as illegal ferry crossings, reservoir area route violations, and illegal entry into shallow waters, and improves the problem of insufficient adaptation of general detection rules to the differences in key water areas.

[0051] In one possible embodiment, such as Figure 3 As shown, multi-source navigation perception data can be processed separately by different detection engines to obtain multiple preliminary detection results.

[0052] First, the multi-source navigation sensing data is categorized by data source type, resulting in positioning-based navigation data and image-based navigation data. Positioning-based navigation data can include AIS data, BeiDou positioning data, or data uploaded by other positioning terminals, specifically including ship position, speed, heading, trajectory, timestamp, equipment status, and ship identification. Image-based navigation data can include CCTV video surveillance data, video frame images, video target detection results, ship motion trajectories, and ship nameplate recognition results. By classifying the data source types, appropriate detection models or rule-based processing flows can be invoked for each data type.

[0053] Then, at least one navigation state feature from the ship's position, speed, heading, and trajectory is extracted from the positioning-based navigation data to obtain the first navigation state feature. For example, features such as changes in ship speed, heading, trajectory deviation, dwell time, trajectory curvature, and positional relationship relative to bridge areas or restricted areas can be calculated based on continuous positioning points. For AIS / BeiDou data, the terminal device status can also be extracted simultaneously to determine whether there are situations such as device offline, abnormal shutdown, or data interruption in the positioning-based navigation data.

[0054] Next, abnormal navigation behavior is detected based on the abnormal navigation behavior detection rules and the first navigation state characteristics to obtain the first preliminary detection results. For example, it is possible to determine whether there is speeding based on speed characteristics and speed limit thresholds; whether there is misdirection of navigation channel based on the relationship between the trajectory and the navigation channel boundary; whether there is reverse navigation based on the relationship between the trajectory direction and the ship's heading; whether there is illegal entry based on whether the ship has entered a restricted area; and whether there is an abnormal shutdown event of the positioning terminal based on the equipment status.

[0055] Furthermore, ship target recognition and motion state analysis are performed based on image-based navigation data to obtain second navigation state features. Specifically, target detection models can be used to identify ship targets in video footage, such as using the YOLOv7 target detection framework to detect ships; character recognition networks can be used to extract ship nameplate information, such as using a CRNN character recognition network to identify ship names in video footage. YOLO is an abbreviation for You Only Look Once, which can be understood as a single-stage target detection algorithm; CRNN is an abbreviation for Convolutional Recurrent Neural Network. By tracking ship targets in continuous video frames, second navigation state features such as ship motion direction, video trajectory, bridge crossing trajectory offset, turning behavior characteristics, and ship traffic statistics can be obtained.

[0056] Finally, abnormal navigation behavior is detected based on abnormal navigation behavior detection rules and second navigation status characteristics to obtain a second preliminary detection result. For example, video trajectories can be used to identify whether a vessel has made a U-turn, deviated from its bridge crossing trajectory, illegally moored, or made abnormal berthing or departure; vessel identity can be confirmed based on the vessel nameplate recognition results; real-time vessel traffic can be monitored based on the number of vessels in the video footage; and vessel motion status can be continuously obtained through video recognition when location-related navigation data is missing.

[0057] This embodiment classifies multi-source navigation perception data by type and extracts navigation status features from positioning navigation data and image navigation data respectively. This can make full use of the characteristics of strong continuity and clear trajectory features of positioning data, as well as the strong intuitiveness and identity assistance recognition capabilities of video data. At the same time, multiple preliminary detection results are obtained based on the same abnormal navigation behavior detection rule, providing multi-source evidence for subsequent target association and fusion verification, thereby improving the reliability of abnormal navigation behavior detection.

[0058] In one possible embodiment, associating the same ship target in multi-source navigation perception data may include the following process.

[0059] First, the spatiotemporal state information of the first vessel target is determined based on positioning-type navigation data, resulting in the first target description information. This first target description information may include the first vessel target's timestamp, latitude and longitude position, speed, heading, trajectory, vessel length, vessel width, vessel type, vessel name, or vessel identification. The timestamp and latitude and longitude position can be used for spatiotemporal matching, while the vessel length, width, and type can be used for vessel attribute matching.

[0060] Then, based on the image-based navigation data, the image recognition features of the second vessel target are determined to obtain the second target description information. This second target description information may include the target's position in the video frame, target outline dimensions, target direction of movement, target speed, vessel appearance features, vessel type identification results, vessel nameplate identification results, and the spatial coverage area corresponding to the video acquisition equipment. For video surveillance equipment with camera calibration information, the video frame coordinates can also be converted into actual water space coordinates, thereby improving the accuracy of matching with positioning-based navigation data.

[0061] Next, candidate ship target matching relationships are generated based on the first and second target description information to obtain candidate association results. Specifically, preliminary matching can be performed based on time and spatial information. For example, first and second ship targets that are within the same time window and spatially close can be used as candidate matching objects. In one specific implementation, preliminary screening can be performed using a preset time window and a preset spatial distance threshold. Then, secondary matching can be performed based on ship size, ship type, ship name, video outline size, and ship nameplate recognition results to generate candidate association results.

[0062] Furthermore, a consistency evaluation is performed on the candidate association results to obtain the target association evaluation results. The consistency evaluation may include at least one of the following: temporal consistency evaluation, spatial consistency evaluation, size consistency evaluation, type consistency evaluation, name consistency evaluation, and direction of motion consistency evaluation. Different weights can be assigned to different consistency evaluation results, and the comprehensive association confidence of the candidate matching objects can be calculated. When the comprehensive association confidence reaches a preset threshold, the candidate matching objects can be considered to have a high probability of belonging to the same ship target.

[0063] Finally, the association results of the same ship target in different data sources are determined based on the target association evaluation results. These association results can be used to represent the binding relationship between a first ship target in positioning-based navigation data and a second ship target in image-based navigation data. For scenarios where delays in AIS / BeiDou signals or dense ship traffic can lead to confusion in spatiotemporal matching, the weight of ship feature matching results can be increased, for example, by associating based on ship size, type, name, and video outline, thereby improving the accuracy of target binding.

[0064] This embodiment constructs first target description information and second target description information, and performs candidate matching and consistency evaluation based on spatiotemporal information and target feature information. This can accurately associate the same ship target from different data sources, providing a basis for mapping different preliminary detection results to the same ship target, thereby reducing association errors caused by multiple ships in close proximity, data delays, or unreliable single matching conditions.

[0065] In one possible embodiment, fusing and verifying multiple preliminary detection results based on the correlation results may include the following process.

[0066] First, based on the association results, the first and second preliminary detection results are mapped to the same vessel target, resulting in a target detection result set. This target detection result set may include abnormal behavior judgment results, abnormal behavior types, abnormal occurrence areas, abnormal occurrence times, abnormal confidence levels, and rule hit information for the same vessel target under different data sources. For example, positioning-based navigation data may output results such as vessel speeding, trajectory deviation, or equipment malfunction; image-based navigation data may output results such as vessel turning around, bridge crossing trajectory deviation, vessel name recognition results, or video tracking results.

[0067] Then, a consistency judgment is performed on the abnormal behavior types, abnormal occurrence areas, and abnormal occurrence times in the target detection result set to obtain a consistency judgment result. The consistency judgment may include determining whether different data sources detected the same vessel target, whether they detected the same or mutually supporting abnormal behavior types, whether the abnormal occurrence times were within the same time window, whether the abnormal occurrence areas were located in the same key water area or adjacent areas, and whether the abnormal trajectories had the same movement trend, etc. For example, when both positioning navigation data and image navigation data show that the same vessel deviated from its bridge crossing trajectory in the bridge area, the consistency can be considered high.

[0068] If the consistency judgment result meets the preset confirmation conditions, the existence of corresponding abnormal navigation behavior for the same vessel target is confirmed, and a confirmed abnormal event is obtained. The preset confirmation conditions may include at least two data sources detecting the same abnormal behavior, one data source detecting the abnormal behavior and another data source providing supporting evidence, and the time and area of ​​the abnormal behavior meeting consistency requirements. The confirmed abnormal event can be directly used to generate alarm information or enter the subsequent regulatory handling process.

[0069] If the consistency judgment result does not meet the preset confirmation conditions, a target verification data source is determined from the data sources corresponding to the multi-source navigation perception data. A verification instruction for the target verification data source is generated, and the processing result of the suspected abnormal event is determined based on the verification result returned by the target verification data source. For example, when positioning navigation data detects that a ship is suspected of speeding or veering off course, but the video data has not yet confirmed it, the video data can be triggered to focus on tracking and verifying the ship target; when the video data identifies that a ship is suspected of turning around or anchoring, but the positioning navigation data has not confirmed it, the positioning navigation data can be called for trajectory verification. The processing result of the suspected abnormal event can be to confirm the abnormality, eliminate the abnormality, or keep it in a pending verification state.

[0070] Finally, based on the processing results of confirmed and suspected abnormal events, the abnormal navigation behavior detection results of the vessel under inspection are obtained. These abnormal navigation behavior detection results may include a set of confirmed abnormal events, a set of suspected abnormal events, false alarm filtering results, and verification basis. Through integrated verification, the impact of misjudgments from a single data source on the final regulatory outcome can be reduced.

[0071] This embodiment maps multiple preliminary detection results to the same vessel target and makes a consistency judgment on the type of abnormal behavior, the area where the abnormality occurs, and the time when the abnormality occurs, which can achieve cross-verification between multi-source detection results. For suspected abnormal behavior detected by a single data source, the target can be triggered to verify the data source, which can filter false alarms and supplement evidence, thereby improving the credibility of abnormal navigation behavior detection results and the accuracy of regulatory handling basis.

[0072] In one possible embodiment, if the data availability status of any data source in the multi-source navigation perception data meets a preset abnormal condition, the vessel to be detected can be continuously tracked based on other data sources.

[0073] First, the availability of each data source in the multi-source navigation sensing data is monitored to obtain the data source availability status. For positioning-related navigation data, the on / off status of AIS / BeiDou terminals, data reporting intervals, positioning point continuity, trajectory rationality, equipment online status, and signal quality can be monitored. For image-related navigation data, video stream interruption, image clarity, degree of occlusion, target recognition confidence, camera online status, and nighttime visibility can be monitored. The data source availability status can be used to indicate whether the corresponding data source can continuously provide reliable navigation sensing data.

[0074] Then, if any data source meets preset anomaly conditions in the data source availability status characterization, a data source anomaly event is generated. The preset anomaly conditions may include: abnormal shutdown of AIS / BeiDou terminal, continuous loss of positioning data, positioning data delay exceeding a preset threshold, abnormal jumps in positioning trajectory, video stream interruption, video recognition confidence level below a preset threshold, and severe video obstruction. The data source anomaly event may include information such as anomaly data source identifier, anomaly type, anomaly occurrence time, anomaly duration, and impact on ship targets.

[0075] Next, based on the abnormal events of the data source, the target data source that can be used for continuous tracking is determined, and the tracking data source selection result is obtained. For example, when the AIS / BeiDou terminal is abnormally shut down or the positioning signal is missing, CCTV video surveillance data can be selected as the target data source, and the vessel under inspection can be continuously monitored through video target recognition, ship nameplate recognition, and video target tracking; when the video surveillance is affected by rain, fog, nighttime lighting, or obstruction, resulting in a decrease in recognition confidence, AIS / BeiDou positioning data can be selected as the target data source, and the vessel under inspection can be continuously monitored through continuous trajectory analysis.

[0076] Furthermore, based on the results of the tracking data source selection, the identity of the vessel under investigation is confirmed and its navigation status is continuously analyzed to obtain continuous tracking results. When location-based navigation data is abnormal but image-based navigation data is available, the vessel's identity can be confirmed through nameplate recognition, vessel outline, vessel direction of movement, and continuous video frame tracking results, and the presence of abnormal navigation behavior can be continuously analyzed. When image-based navigation data is abnormal but location-based navigation data is available, the vessel's identity can be confirmed through vessel identification, continuous positioning points, and trajectory trends, and its speed, heading, and trajectory changes can be continuously analyzed.

[0077] Finally, the abnormal navigation behavior detection results are updated based on continuous tracking, and tiered alarm information is generated based on the updated results. The tiered alarm information can be determined according to the type of abnormal behavior, its impact range, severity, duration, and risk level of the target key waterway. For example, alarms can be divided into Level 1, Level 2, and Level 3, with different levels corresponding to different push channels and recipients. Level 1 alarms can be pushed to on-duty supervisors and emergency response personnel; Level 2 alarms can be pushed to the monitoring platform and relevant management personnel; and Level 3 alarms can be used for platform recording, statistical analysis, or general alerts. The tiered alarm information can also be used for visualization, historical data review, and export of detection reports.

[0078] This embodiment monitors the availability of various data sources and selects other available data sources for identity verification and continuous analysis of navigation status when any data source is abnormal. This allows for continuous monitoring of vessel dynamics even when positioning terminals are off, signals are blocked, or video is affected by the environment. By updating the detection results of abnormal navigation behavior based on continuous tracking results and generating tiered alarm information, the timeliness of handling abnormal events and the continuity of monitoring in key waters can be improved.

[0079] like Figure 2As shown in the illustration, this embodiment also provides an intelligent detection system for abnormal navigation behavior of vessels in key waterways. This system can be deployed in an inland waterway smart water transport supervision platform, a key waterway supervision center, a bridge area navigation safety management system, or a vessel dynamic monitoring system. The system includes a data acquisition module, a rule determination module, a preliminary detection module, a target association module, a fusion verification module, a continuous tracking module, and a result output module.

[0080] The data acquisition module is used to acquire basic control data for key target waterways and multi-source navigation perception data of vessels to be monitored. Specifically, the data acquisition module can interface with basic service interfaces of electronic navigable charts, GIS geographic information systems, vessel static file systems, AIS / BeiDou terminal systems, CCTV video surveillance systems, and inland waterway smart transportation supervision systems to obtain basic spatial data of waterways, navigation control data of waterways, vessel static file information, positioning-based navigation data, and image-based navigation data. Through the data acquisition module, a unified data entry point can be provided for subsequent rule determination and anomaly detection.

[0081] The rule determination module is used to determine abnormal navigation behavior detection rules corresponding to the target key waterways based on basic control data. Specifically, the rule determination module can determine the waterway type of the target key waterway based on waterway spatial control data, match candidate detection rules from a preset rule base, and then configure rule parameters by combining navigation restrictions, passage rules, and risk area information to obtain abnormal navigation behavior detection rules adapted to the target key waterway. The rule determination module also supports customized configuration and dynamic updates of control scope, detection rules, alarm thresholds, and push methods.

[0082] The preliminary detection module is used to perform ship target identification and abnormal navigation behavior detection on multi-source navigation perception data, obtaining multiple preliminary detection results. The preliminary detection module can include a positioning detection unit and an image detection unit. The positioning detection unit can process AIS / BeiDou data, extracting information such as ship position, speed, heading, trajectory, and equipment status, and outputting a first preliminary detection result according to abnormal navigation behavior detection rules. The image detection unit can process CCTV video surveillance data, using methods such as target detection, character recognition, and target tracking to identify ship targets, nameplate information, and motion status, and output a second preliminary detection result according to abnormal navigation behavior detection rules.

[0083] The target association module is used to associate the same ship target in multi-source navigation perception data to obtain association results. Specifically, the target association module can match and bind ship targets in positioning navigation data and image navigation data based on features such as timestamp, latitude and longitude, ship size, ship type, ship outline, ship name recognition results, and direction of movement, to obtain association results of the same ship target in different data sources.

[0084] The fusion verification module is used to fuse and verify multiple preliminary detection results based on correlation results to obtain the detection results of abnormal navigation behavior of the vessel under test. Specifically, the fusion verification module can map multiple preliminary detection results to the same vessel target and make consistency judgments on the type of abnormal behavior, the area where the abnormality occurs, and the time of the abnormality. When multiple data sources detect the same abnormal behavior of the same vessel, the abnormal event can be confirmed; when a single data source detects suspected abnormal behavior, it can trigger other data sources to perform secondary verification, thereby filtering false alarm data and improving the credibility of the detection results.

[0085] The continuous tracking module is used to continuously track the vessel under test based on other data sources when the data availability status of any data source in the multi-source navigation perception data meets preset abnormal conditions, and updates the abnormal navigation behavior detection results based on the continuous tracking results. Specifically, when an AIS / BeiDou terminal is detected to be abnormally shut down, the positioning signal is missing, or the device is offline, the continuous tracking module can trigger the image detection unit to continuously track and detect the corresponding vessel, and identify the vessel by recognizing the vessel nameplate, the vessel's appearance features, and the video trajectory. When video data is unavailable, the continuous tracking module can also perform continuous trajectory analysis using positioning-based navigation data.

[0086] The results output module is used to output the detection results of abnormal navigation behavior or the updated abnormal navigation behavior detection results. Specifically, the results output module can output the detection results to the monitoring system in real time and generate level 1, level 2, and level 3 alarm information according to the type, scope of impact, and severity of abnormal behavior. The results output module can also provide functions such as visualization, historical data backtracking, detection report export, alarm push, and rule iteration feedback.

[0087] This embodiment, through the collaborative work of the data acquisition module, rule determination module, preliminary detection module, target association module, fusion verification module, continuous tracking module, and result output module, can form a complete regulatory link from the construction of basic control data, preliminary detection of multi-source data, association of the same vessel target, fusion verification of detection results, blind tracking of abnormal data sources, to hierarchical alarm output. The system can take into account the differentiated regulatory needs of different key waterways, improve the accuracy of identifying complex navigation behaviors, and maintain the ability to continuously track vessel dynamics when some data sources are unavailable, thereby improving the reliability and practicality of intelligent detection of abnormal vessel navigation behaviors in key inland waterways.

[0088] The specific implementation details of each module can be found in the above description of the limitations of the intelligent detection method for abnormal navigation behavior of ships in key waterways, and will not be repeated here.

[0089] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for intelligent detection of abnormal navigation behavior of vessels in key waterways, characterized in that, The method includes: Acquire basic control data for key target waters and multi-source navigation perception data of vessels to be inspected; Based on the aforementioned basic control data, abnormal navigation behavior detection rules corresponding to the target key waterways are determined; The multi-source navigation perception data were used to perform ship target identification and abnormal navigation behavior detection, resulting in several preliminary detection results. The same ship target in the multi-source navigation perception data is associated, and the multiple preliminary detection results are fused and verified based on the association results to obtain the abnormal navigation behavior detection results of the ship to be detected; If the data availability status of any data source in the multi-source navigation perception data meets the preset abnormal conditions, the vessel to be detected is continuously tracked based on other data sources, and the abnormal navigation behavior detection result is updated according to the continuous tracking result. Output the abnormal navigation behavior detection result or the updated abnormal navigation behavior detection result.

2. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 1, characterized in that, Obtain basic control data for key target water areas, including: Electronic navigation map data and geographic information data of the target key water area are obtained to obtain basic spatial data of the water area; Based on the aforementioned basic spatial data of the water area, the spatial control range of the target key water area is determined, and the spatial control data of the water area is obtained. Obtain navigation restriction information, passage rule information, and risk area information corresponding to the target key waterway to obtain waterway navigation control data; Obtain the static file information of the vessel, and associate the static file information of the vessel with the water space control data and the water navigation control data to obtain the basic control data.

3. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 2, characterized in that, Based on the aforementioned basic control data, abnormal navigation behavior detection rules corresponding to the target key waterways are determined, including: Based on the water space control data, the water type of the target key water area is determined, and the water type identification result is obtained; Based on the water area type identification results, the corresponding candidate detection rules are matched from the preset rule base to obtain a candidate rule set; The rule parameters in the candidate rule set are configured based on the waterway navigation control data to obtain the rule parameter configuration result; Based on the rule parameter configuration results, abnormal navigation behavior detection rules corresponding to the target key water area are determined.

4. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 3, characterized in that, When the water type identification result indicates that the target key water area is a bridge area, the abnormal navigation behavior detection rules corresponding to the target key water area are determined, including: Obtain navigation channel information, bridge area boundary information, and bridge area traffic restriction information corresponding to the water area of ​​the bridge area to obtain bridge area control and constraint data; Based on the bridge area control and constraint data, the standard passage status of vessels in the bridge area waters is determined, and the standard passage status information of the bridge area is obtained. Based on the bridge area standard traffic status information, configure at least one detection rule among anchoring, U-turn, overtaking, speeding, wrong navigation channel, reverse navigation, sudden navigation, and encounter risk to obtain the bridge area rule configuration result; Based on the bridge area rule configuration results, abnormal navigation behavior detection rules for the bridge area are generated, and the navigation status of ships in the bridge area is detected in parallel according to the abnormal navigation behavior detection rules for the bridge area to obtain abnormal navigation behavior detection results for the bridge area.

5. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 3, characterized in that, When the water type identification result indicates that the target key water area is a ferry crossing, a reservoir, or a shallow water area, the abnormal navigation behavior detection rules corresponding to the target key water area are determined, including: Based on the water area type identification results, the corresponding scene control object is determined, and scene control object information is obtained; Based on the information of the scene control object, the corresponding scene abnormal behavior type is determined, and a set of scene abnormal behaviors is obtained; Based on the waterway navigation control data, corresponding judgment conditions are configured for the set of abnormal behaviors in the scenario to obtain scenario-based detection rules; Based on the aforementioned scenario-based detection rules, the navigation status of ships in the target key waters is detected, and the abnormal navigation behavior detection results for different waters are obtained.

6. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 1, characterized in that, Ship target identification and abnormal navigation behavior detection were performed on the multi-source navigation perception data, respectively, to obtain several preliminary detection results, including: The multi-source navigation perception data is classified into data source types to obtain positioning navigation data and image navigation data. Based on the positioning-type navigation data, at least one navigation state feature among the ship's position, speed, heading, and trajectory is extracted to obtain the first navigation state feature; Based on the abnormal navigation behavior detection rules and the first navigation state features, abnormal navigation behavior is detected to obtain a first preliminary detection result. Based on the image-type navigation data, ship target identification and motion state analysis are performed to obtain the second navigation state characteristics; Based on the abnormal navigation behavior detection rules and the second navigation state characteristics, abnormal navigation behavior is detected to obtain a second preliminary detection result.

7. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 6, characterized in that, Associating the same ship target in the multi-source navigation perception data includes: Based on the positioning navigation data, the spatiotemporal state information of the first vessel target is determined, and the description information of the first target is obtained; Based on the image-type navigation data, the image recognition features of the second ship target are determined, and the description information of the second target is obtained; Based on the first target description information and the second target description information, candidate ship target matching relationships are generated to obtain candidate association results; The consistency evaluation of the candidate association results is performed to obtain the target association evaluation result; The association results of the same ship target in different data sources are determined based on the target association evaluation results.

8. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 7, characterized in that, Based on the correlation results, the multiple preliminary detection results are fused and verified to obtain the abnormal navigation behavior detection results of the vessel under test, including: Based on the association results, the first preliminary detection result and the second preliminary detection result are mapped to the same ship target to obtain a target detection result set; A consistency judgment is made on the abnormal behavior type, abnormal occurrence area, and abnormal occurrence time in the target detection result set to obtain a consistency judgment result; If the consistency judgment result meets the preset confirmation conditions, it is confirmed that the same ship target has corresponding abnormal navigation behavior, and an abnormal event is confirmed. If the consistency judgment result does not meet the preset confirmation condition, a target verification data source is determined from the data source corresponding to the multi-source navigation perception data, a verification instruction is generated for the target verification data source, and the processing result of the suspected abnormal event is determined according to the verification result returned by the target verification data source. Based on the processing results of the confirmed abnormal events and the suspected abnormal events, the detection results of the abnormal navigation behavior of the vessel to be detected are obtained.

9. The intelligent detection method for abnormal navigation behavior of vessels in key waterways according to claim 1, characterized in that, If the data availability status of any data source in the multi-source navigation perception data meets a preset anomaly condition, the vessel to be detected is continuously tracked based on other data sources, and the abnormal navigation behavior detection result is updated according to the continuous tracking result, including: Availability monitoring is performed on each data source in the multi-source navigation perception data to obtain the data source availability status; If any data source meets a preset abnormal condition as indicated by the data source availability status, a data source abnormal event is generated. Based on the abnormal events of the data source, a target data source that can be used for continuous tracking is determined, and the tracking data source selection result is obtained; Based on the tracking data source selection results, the vessel to be detected is identified and its navigation status is continuously analyzed to obtain continuous tracking results. The abnormal navigation behavior detection results are updated based on the continuous tracking results, and graded alarm information is generated based on the updated abnormal navigation behavior detection results.

10. An intelligent detection system for abnormal navigation behavior of vessels in key waterways, characterized in that, The system includes: The data acquisition module is used to acquire basic control data for key target waters and multi-source navigation perception data of vessels to be monitored. The rule determination module is used to determine the abnormal navigation behavior detection rules corresponding to the target key water area based on the control basic data; The preliminary detection module is used to perform ship target identification and abnormal navigation behavior detection on the multi-source navigation perception data, and obtain multiple preliminary detection results; The target association module is used to associate the same ship target in the multi-source navigation perception data to obtain the association result; The fusion verification module is used to perform fusion verification on the multiple preliminary detection results based on the correlation results to obtain the abnormal navigation behavior detection results of the vessel to be detected; The continuous tracking module is used to continuously track the vessel to be detected based on other data sources when the data availability status of any data source in the multi-source navigation perception data meets the preset abnormal conditions, and update the abnormal navigation behavior detection results according to the continuous tracking results. The result output module is used to output the abnormal navigation behavior detection result or the updated abnormal navigation behavior detection result.

Images