Visual troubleshooting and treatment system and method for potential safety hazards of marine fishing boat

Abstract

The invention discloses a visual investigation and treatment system and method for potential safety hazards of a marine fishing boat, and aims to improve the intelligent and visual level of safety management of fishing boat operation. The method comprises the following steps: acquiring video monitoring data of different operation areas of a target marine fishing vessel, and generating an operation behavior vector of an operator; constructing a three-dimensional model in combination with ship structure data, and fusing the operation behavior vector and the three-dimensional model to generate an operation scene three-dimensional visual model; carrying out operation potential safety hazard checking based on the model, and obtaining a data updating timestamp in real time to determine a checking delay state; the acquisition frequency of the video monitoring data is adjusted according to the troubleshooting delay state, a data updating strategy is made, and an optimized visualization model is obtained; and performing operation safety assessment on the operator through the optimization model, and forming an operation potential safety hazard treatment strategy according to the operation safety assessment. According to the method, potential safety hazards can be effectively found and treated, and the troubleshooting accuracy and the treatment efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fishing vessel safety management, and in particular to a system and method for visually checking and managing safety hazards of marine fishing vessels. Background Art

[0002] With the expansion of fishing operations and the increasing complexity of operations, marine fishing vessels have become a vital vehicle for both distant-water and near-shore production. However, fishing vessels operate in cramped spaces and harsh environments, requiring personnel to frequently perform high-risk activities such as walking, bending, carrying, and climbing. These activities often occur in continuously adverse sea conditions or at night, making them highly susceptible to accidents such as falls, mechanical collisions, and operational errors. Existing safety management primarily relies on traditional video surveillance and manual inspections. This approach not only results in long inspection cycles and delayed detection of potential hazards, but also suffers from inaccurate location and delayed response.

[0003] In recent years, with the advancement of computer vision, 3D reconstruction, and digital twin technologies, the use of video surveillance to identify and visualize operational behaviors has become a crucial tool for improving vessel safety management. However, existing technologies for 3D modeling of operational scenes often suffer from the following deficiencies: First, the use of only 2D video footage fails to effectively integrate operator actions with the spatial structure of the vessel, lacking intuitive 3D dynamic representation. Second, with the widespread adoption of multi-area high-definition video surveillance on fishing vessels, the amount of video data generated has increased dramatically. The large-scale processing of continuous video frames and the computational demands for action recognition place high demands on computing resources and bandwidth, which can easily lead to data update delays in the 3D visualization model of the operational scene. When data update delays exceed a preset threshold, the operational status reflected in the model deviates from the actual status, significantly reducing the timeliness and accuracy of hazard detection and real-time alerts.

[0004] Furthermore, existing methods lack a dynamic assessment and feedback mechanism for delay status, making it impossible to flexibly adjust the frequency and update strategy of video data acquisition in different operating areas based on the degree of delay. Furthermore, they lack a fully closed-loop risk assessment and hazard management system. To address these issues, a visual inspection and management method for marine fishing vessel safety hazards is urgently needed that integrates operational behavior recognition, three-dimensional scene modeling, delay assessment, and data update optimization to achieve high-precision dynamic monitoring and intelligent safety management in complex operating environments. Summary of the Invention

[0005] In order to solve at least one of the above technical problems, the present invention proposes a system and method for visually checking and managing safety hazards of marine fishing vessels.

[0006] A first aspect of the present invention provides a method for visually checking and managing safety hazards of marine fishing vessels, comprising:

[0007] Obtaining video surveillance data of different operating areas of the target marine fishing vessel, performing operator action recognition and spatial distribution extraction on the video surveillance data, and constructing an operation behavior vector;

[0008] Acquiring ship structure data of a target ocean fishing vessel, constructing a three-dimensional model of the target ocean fishing vessel based on the ship structure data, and constructing a three-dimensional visualization model of an operation scene using the operation behavior vector and the three-dimensional model;

[0009] Performing a visual inspection of the target marine fishing vessel for potential safety hazards in operation based on the three-dimensional visualization model of the operation scene, obtaining a data update timestamp of the three-dimensional visualization model of the operation scene in real time, and determining a model inspection delay state based on the update timestamp;

[0010] Determine the acquisition frequency of video surveillance data of different operation areas according to the troubleshooting delay status, and determine the data update strategy of the three-dimensional visualization model of the operation scene according to the acquisition frequency to obtain an optimized visualization model;

[0011] An operational safety assessment is performed on the operating personnel of the target ocean fishing vessel based on the optimized visualization model, and an operational safety hazard management strategy is constructed based on the operational safety.

[0012] In this solution, the video surveillance data of different operating areas of the target ocean fishing vessel are obtained, the operator action recognition and spatial distribution extraction are performed on the video surveillance data, and the operation behavior vector is constructed, specifically:

[0013] Acquire video surveillance data of different operating areas of the target marine fishing vessel, extract a continuous video frame sequence of the video surveillance data, separate the foreground motion area of ​​each video frame based on a background subtraction algorithm, identify the outline of the operator based on the foreground motion area, calculate the minimum circumscribed matrix box of the operator outline, and obtain operator detection box data;

[0014] Based on the operator detection frame data, a key point detection algorithm is applied to extract human skeleton information, including the coordinates of key points of the head, shoulders, elbows, wrists, hips, knees, and ankles. The angles and displacement changes between key points are calculated to obtain the dynamic feature vectors of the key points.

[0015] Obtain key point change sequence data for different actions, perform action state modeling based on the hidden Markov model, define the action state transition probability matrix and observation probability matrix, and build an action recognition model;

[0016] Importing the key point dynamic feature vector into the action recognition model, calculating the likelihood probability value of the key point dynamic feature and the action state sequence through a forward-backward algorithm, and identifying the action type of the operator, including walking, bending, carrying, climbing and stationary actions, when the likelihood probability value is greater than the action recognition threshold, to obtain action type data;

[0017] Based on action type data and video surveillance data, the operator detection frame data is mapped to the three-dimensional coordinate system of the marine fishing vessel. The position coordinates of the operator relative to the ship structure are calculated through the perspective transformation matrix. Combined with the action type data, the operation behavior vector is constructed. The operation behavior vector includes the action type label, position coordinate vector and timestamp information.

[0018] In this solution, the ship structure data of the target ocean fishing vessel is obtained, a three-dimensional model of the target ocean fishing vessel is constructed based on the ship structure data, and a three-dimensional visualization model of the operation scene is constructed by combining the operation behavior vector and the three-dimensional model. Specifically,

[0019] Acquire ship structure data of the target ocean fishing vessel, the ship structure data including hull geometry, deck layer layout, cabin plan, and structural component position information; extract hull contour lines, cabin boundary information, and structural component contour information based on the ship structure data to construct ship structure contour data;

[0020] Establishing a hull surface mesh model and an internal cabin three-dimensional geometric model based on the ship structure contour data, topologically connecting the hull surface mesh model and the internal cabin three-dimensional geometric model to construct a three-dimensional ship structure, performing ship structure texture mapping on the three-dimensional ship structure, and constructing a three-dimensional mesh model of the target ocean fishing vessel;

[0021] Convert the position coordinate vector in the operation behavior vector to the coordinate system of the three-dimensional grid model, and calculate the mapping coordinates of the operator's position in the three-dimensional grid model;

[0022] Obtain animation data of different actions of the operators, build an action animation model library, match the action type in the operation behavior vector with the action animation model library, and map the matching action animation to the three-dimensional model in real time according to the mapping coordinates and timestamp information of the operators based on the digital twin, and build a three-dimensional visualization model of the target ocean fishing vessel operation scene based on the digital twin.

[0023] In this solution, the target ocean fishing vessel is visually checked for operational safety hazards based on the three-dimensional visualization model of the operation scene, and the data update timestamp of the three-dimensional visualization model of the operation scene is obtained in real time. The model's inspection delay status is determined based on the update timestamp, specifically:

[0024] Perform a visual inspection of safety operation hazards of the target marine fishing vessel based on the three-dimensional visualization model of the operation scene, and obtain in real time the timestamp information of the video frame corresponding to each operation behavior vector input into the three-dimensional visualization model of the operation scene as a data update timestamp;

[0025] Extracting the current clock system time of the three-dimensional visualization model of the operation scene in real time as a reference timestamp, and calculating the time difference between the data update timestamp and the reference timestamp as a single troubleshooting delay;

[0026] Continuously collect multiple single-check delay data within a preset time window to form a time series, and calculate the average delay value of the single-check delay time series within the time window;

[0027] The average delay value is compared with the delay status reference interval set in advance based on the troubleshooting accuracy requirements, the troubleshooting delay status is determined according to the position of the average delay value in the reference interval, and the troubleshooting delay status is marked in real time in the troubleshooting status information of the three-dimensional visualization model of the operation scene.

[0028] In this solution, the acquisition frequency of video surveillance data of different operation areas is determined according to the delay status of the troubleshooting, and the data update strategy of the three-dimensional visualization model of the operation scene is determined according to the acquisition frequency to obtain the optimized visualization model, which is specifically:

[0029] Determining whether the data update of the three-dimensional visualization model of the operation scene meets the preset safety hazard inspection accuracy requirements based on the inspection delay status, and if not, determining the operation behavior activity parameter of each operation area based on the operation behavior vector, wherein the activity parameter is determined based on the number of operation behavior vectors in the operation area per unit time;

[0030] Obtain a time series of troubleshooting delays for each operating area over a continuous time period, establish a confidence interval for the troubleshooting delay based on the average delay value and delay variance in the time series, and compare the confidence interval with a preset troubleshooting delay threshold to obtain a delay sensitivity parameter for each operating area;

[0031] Based on the activity parameter and delay sensitivity parameter of the operation behavior, a scoring function for priority of obtaining data in the operation area is constructed. The scoring function assigns different weight factors to activity and delay sensitivity respectively, and calculates the priority score of obtaining video surveillance data in all operation areas;

[0032] Assign an operation risk level to the action type label corresponding to the operation behavior vector in each operation area to obtain a weighted operation risk score for each operation area;

[0033] Determine the frequency of video surveillance data acquisition for each operation area based on the operation risk weighted score and operation behavior activity parameter of each operation area, and construct a data update strategy for the three-dimensional visualization model of the operation scene based on the video surveillance data acquisition priority score and video surveillance data acquisition frequency of the operation area;

[0034] The three-dimensional visualization model of the operation scene is optimized according to the data update strategy to obtain an optimized visualization model.

[0035] In this solution, the operation safety assessment of the operators of the target ocean fishing vessel is performed based on the optimized visualization model, and an operation safety hazard management strategy is constructed based on the operation safety, specifically:

[0036] Extracting the operator's action type label and position coordinate vector in real time based on the optimized visualization model, and calculating the Euclidean distance between the operator's position and the nearest danger zone boundary based on the danger zone boundary coordinates predefined in the optimized visualization model;

[0037] According to the predefined risk weight values ​​of different action types, the Euclidean distance is weighted to obtain real-time safety assessment parameters, and a safety risk accumulation function is constructed based on the duration data of the operator being in a dangerous warning state continuously;

[0038] When the safety assessment parameter is lower than the preset warning threshold, a real-time alarm level determination is performed according to the safety risk accumulation function value, and the alarm location coordinates are highlighted in the optimized visualization model;

[0039] Obtain the action type label, position coordinate vector, and duration data from the alarm event, analyze the causes of safety hazards through the risk grading logic tree, and determine the type of safety hazard and the category of illegal operation behavior;

[0040] Generate real-time warning information and correction time parameters of different levels according to the severity level of the safety hazard type and the duration data of the illegal operation behavior;

[0041] When multiple people are operating illegally at the same time in the alarm location coordinates, the regional violation density index is calculated and the correction time limit parameter is compressed and adjusted according to the regional violation density index;

[0042] Outputting the real-time alarm information, correction time limit parameters, and adjusted correction time limit parameters to a management terminal of the target operation area, and simultaneously driving the optimized visualization model to record the alarm behavior trajectory and start the correction time limit countdown display;

[0043] The updated safety assessment parameters in the alarm position are continuously monitored within the correction time limit. When the updated safety assessment parameters have not recovered to the safety threshold range, the alarm level is gradually increased according to the timeout duration ratio and the correction time limit is recalculated to obtain an operation safety hazard management strategy.

[0044] A second aspect of the present invention further provides a system for visually checking and managing safety hazards of marine fishing vessels. The system comprises: a memory and a processor. The memory includes a program for visually checking and managing safety hazards of marine fishing vessels. When the program for visually checking and managing safety hazards of marine fishing vessels is executed by the processor, the following steps are implemented:

[0045] Obtaining video surveillance data of different operating areas of the target marine fishing vessel, performing operator action recognition and spatial distribution extraction on the video surveillance data, and constructing an operation behavior vector;

[0046] Acquiring ship structure data of a target ocean fishing vessel, constructing a three-dimensional model of the target ocean fishing vessel based on the ship structure data, and constructing a three-dimensional visualization model of an operation scene using the operation behavior vector and the three-dimensional model;

[0047] Performing a visual inspection of the target marine fishing vessel for potential safety hazards in operation based on the three-dimensional visualization model of the operation scene, obtaining a data update timestamp of the three-dimensional visualization model of the operation scene in real time, and determining a model inspection delay state based on the update timestamp;

[0048] Determine the acquisition frequency of video surveillance data of different operation areas according to the troubleshooting delay status, and determine the data update strategy of the three-dimensional visualization model of the operation scene according to the acquisition frequency to obtain an optimized visualization model;

[0049] An operational safety assessment is performed on the operating personnel of the target ocean fishing vessel based on the optimized visualization model, and an operational safety hazard management strategy is constructed based on the operational safety.

[0050] This invention discloses a system and method for visually detecting and managing safety hazards on marine fishing vessels, aiming to enhance the intelligence and visualization of fishing vessel operational safety management. The method includes: acquiring video surveillance data from different operating areas of a target marine fishing vessel to generate operational behavior vectors for the operators; constructing a three-dimensional model based on the vessel's structural data, fusing the operational behavior vectors with the three-dimensional model to generate a three-dimensional visualization model of the operational scene; conducting operational safety hazard inspections based on this model, and obtaining data update timestamps in real time to determine the inspection delay status; adjusting the acquisition frequency of video surveillance data based on the inspection delay status, formulating a data update strategy, and obtaining an optimized visualization model; and conducting operational safety assessments on the operators using the optimized model, thereby forming an operational safety hazard management strategy based on the assessment. This method can effectively identify and manage safety hazards, improving inspection accuracy and management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flow chart showing a method for visually checking and managing safety hazards of marine fishing vessels according to the present invention is shown;

[0052] Figure 2 A flowchart of constructing a three-dimensional visualization model of an operation scene according to the present invention is shown;

[0053] Figure 3 A flow chart showing the troubleshooting delay status of a determination model of the present invention is shown;

[0054] Figure 4 A block diagram of a system for visually checking and managing safety hazards of marine fishing vessels according to the present invention is shown. DETAILED DESCRIPTION

[0055] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0056] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0057] Figure 1 A flow chart of a method for visually checking and managing safety hazards of marine fishing vessels according to the present invention is shown.

[0058] like Figure 1 As shown, the first aspect of the present invention provides a method for visually checking and managing safety hazards of marine fishing vessels, comprising:

[0059] S102, acquiring video surveillance data of different operating areas of the target marine fishing vessel, performing operator action recognition and spatial distribution extraction on the video surveillance data, and constructing an operation behavior vector;

[0060] S104, obtaining ship structure data of the target ocean fishing vessel, constructing a three-dimensional model of the target ocean fishing vessel based on the ship structure data, and constructing a three-dimensional visualization model of the operation scene by combining the operation behavior vector and the three-dimensional model;

[0061] S106, performing a visual inspection of the target marine fishing vessel for operational safety hazards based on the three-dimensional visualization model of the operational scene, obtaining a data update timestamp of the three-dimensional visualization model of the operational scene in real time, and determining an inspection delay state of the model based on the update timestamp;

[0062] S108, determining the acquisition frequency of video surveillance data of different operation areas according to the delay status, and determining a data update strategy of the three-dimensional visualization model of the operation scene according to the acquisition frequency to obtain an optimized visualization model;

[0063] S110, conducting an operation safety assessment on the operators of the target ocean fishing vessel according to the optimized visualization model, and constructing an operation safety hazard management strategy based on the operation safety.

[0064] It should be noted that by obtaining video surveillance data of different operating areas of the target ocean fishing vessel and performing operator action recognition and spatial distribution extraction on the video data, it is possible to accurately extract the key action characteristics and position distribution information of the operators in a complex environment, and realize high-precision modeling of the operating behavior; by obtaining ship structure data to construct a three-dimensional model and integrating the operating behavior vector with the three-dimensional model, it is possible to intuitively present the relationship between the personnel action and the spatial structure of the ship, and improve the authenticity and intuitiveness of the three-dimensional visualization of the operating scene; by checking the three-dimensional visualization model for operational safety hazards and obtaining data update timestamps in real time, the synchronization between the model and the real-time operating status can be accurately monitored, thereby dynamically evaluating the inspection delay status; determining the video surveillance data acquisition frequency of different operating areas according to the inspection delay status, and then formulating a data update strategy based on the acquisition frequency, can effectively optimize the resource allocation of video data acquisition and processing, reduce the model update delay caused by large-scale video streams, and ensure the real-time performance of high-frequency change areas; based on the optimized visualization model, the operator safety is assessed and a safety hazard management strategy is generated, which can realize hierarchical identification, dynamic warning and efficient management of operational risks.

[0065] According to an embodiment of the present invention, the acquisition of video surveillance data of different operating areas of a target marine fishing vessel, the identification of operator actions and the extraction of spatial distribution of the video surveillance data, and the construction of an operating behavior vector are specifically as follows:

[0066] Acquire video surveillance data of different operating areas of the target marine fishing vessel, extract a continuous video frame sequence of the video surveillance data, separate the foreground motion area of ​​each video frame based on a background subtraction algorithm, identify the outline of the operator based on the foreground motion area, calculate the minimum circumscribed matrix box of the operator outline, and obtain operator detection box data;

[0067] Based on the operator detection frame data, a key point detection algorithm is applied to extract human skeleton information, including the coordinates of key points of the head, shoulders, elbows, wrists, hips, knees, and ankles. The angles and displacement changes between key points are calculated to obtain the dynamic feature vectors of the key points.

[0068] Obtain key point change sequence data for different actions, perform action state modeling based on the hidden Markov model, define the action state transition probability matrix and observation probability matrix, and build an action recognition model;

[0069] Importing the key point dynamic feature vector into the action recognition model, calculating the likelihood probability value of the key point dynamic feature and the action state sequence through a forward-backward algorithm, and identifying the action type of the operator, including walking, bending, carrying, climbing and stationary actions, when the likelihood probability value is greater than the action recognition threshold, to obtain action type data;

[0070] Based on action type data and video surveillance data, the operator detection frame data is mapped to the three-dimensional coordinate system of the marine fishing vessel. The position coordinates of the operator relative to the ship structure are calculated through the perspective transformation matrix. Combined with the action type data, the operation behavior vector is constructed. The operation behavior vector includes the action type label, position coordinate vector and timestamp information.

[0071] It should be noted that by obtaining video surveillance data from different operating areas of fishing vessels to identify operators, and determining the dynamic feature vectors of key points, and then performing action recognition through the hidden Markov model, HMM can effectively characterize the dynamic change law of the human body key point sequence in the time dimension during the operator's action process, and regard human action as an observation sequence generated by a series of hidden states according to a certain probability transfer, where each hidden state corresponds to a specific action stage, and the observation sequence is composed of the angle and displacement changes between key points. By pre-training based on a large number of labeled action samples to obtain the state transfer probability matrix and observation probability matrix, the time dimension of different action types can be accurately modeled. After extracting the dynamic feature vector of key points in the real-time video frame, the forward-backward algorithm of HMM is used to calculate the likelihood probability value of the feature vector and the modeled action state sequence, so that the system can accurately determine the action type according to the probability distribution, thereby realizing stable recognition of actions such as walking, bending, carrying, climbing and standing still in a variety of working scenarios; then, combining the action recognition results with the three-dimensional coordinate information obtained by the perspective transformation matrix, the action type label, position coordinate vector and timestamp are encapsulated together into an operation behavior vector, which can not only reflect the action category of the operator, but also describe its position and timing attributes in the three-dimensional space.

[0072] Figure 2 A flow chart of the present invention for constructing a three-dimensional visualization model of an operation scene is shown.

[0073] According to an embodiment of the present invention, the acquisition of ship structure data of the target ocean fishing vessel, the construction of a three-dimensional model of the target ocean fishing vessel based on the ship structure data, and the construction of a three-dimensional visualization model of the operation scene using the operation behavior vector and the three-dimensional model are specifically as follows:

[0074] S202, obtaining ship structure data of the target ocean fishing vessel, wherein the ship structure data includes hull geometry, deck layer layout, cabin plan, and structural component position information; extracting hull contour lines, cabin boundary information, and structural component contour information based on the ship structure data, and constructing ship structure contour data;

[0075] S204: establishing a hull surface mesh model and an internal cabin three-dimensional geometric model based on the ship structure contour data, topologically connecting the hull surface mesh model and the internal cabin three-dimensional geometric model to construct a three-dimensional ship structure, performing ship structure texture mapping on the three-dimensional ship structure, and constructing a three-dimensional mesh model of the target ocean fishing vessel;

[0076] S206, converting the position coordinate vector in the work behavior vector into the coordinate system of the three-dimensional grid model, and calculating the mapping coordinates of the worker's position in the three-dimensional grid model;

[0077] S208, obtain animation data of different actions of the operating personnel, build an action animation model library, match the action type in the operation behavior vector with the action animation model library, and map the matching action animation to the three-dimensional model in real time according to the mapping coordinates and timestamp information of the operating personnel based on the digital twin, and build a three-dimensional visualization model of the target ocean fishing vessel operation scene based on the digital twin.

[0078] It should be noted that through the topological connection of the hull surface mesh model and the internal cabin three-dimensional geometric model and the ship structure texture mapping, the three-dimensional spatial environment of the marine fishing vessel, including the deck, cabin, and structural components, is completely reconstructed; the three-dimensional coordinates in the operation behavior vector and the action type are matched with the timestamp through perspective transformation, and the animation model corresponding to the action type is driven to be presented continuously in chronological order in the three-dimensional scene, so that the action status of the operator has realism and time sequence in the digital twin model, which can not only intuitively display the real-time position and action of the operator in each structural area, but also can trace and monitor the operation behavior in different time periods and different spatial positions; it effectively improves the intuitiveness, accuracy and efficiency of the operation safety hazard inspection, so that managers can quickly identify potential risk areas and illegal operations.

[0079] Figure 3 A flow chart of the present invention is shown for determining the troubleshooting delay status of the model.

[0080] According to an embodiment of the present invention, the target marine fishing vessel is visually checked for operational safety hazards based on the three-dimensional visualization model of the operation scene, a data update timestamp of the three-dimensional visualization model of the operation scene is obtained in real time, and the model's inspection delay state is determined based on the update timestamp, specifically:

[0081] S302, performing a visual inspection of safety operation hazards of the target marine fishing vessel based on the three-dimensional visualization model of the operation scene, and obtaining in real time the timestamp information of the video frame corresponding to each operation behavior vector input into the three-dimensional visualization model of the operation scene as a data update timestamp;

[0082] S304, extracting the current clock system time of the three-dimensional visualization model of the operation scene in real time as a reference timestamp, and calculating the time difference between the data update timestamp and the reference timestamp as a single troubleshooting delay;

[0083] S306, continuously collecting multiple single-check delay data within a preset time window to form a time series, and calculating the average delay value of the single-check delay time series within the time window;

[0084] S308, comparing the average delay value with a delay status reference interval pre-set based on the troubleshooting accuracy requirement, determining the troubleshooting delay status according to the position of the average delay value in the reference interval, and marking the troubleshooting delay status in real time in the troubleshooting status information of the three-dimensional visualization model of the operation scene.

[0085] It should be noted that with the increase in the scale of ship operations and the expansion of the coverage of video surveillance systems, the amount of video surveillance data generated in different operating areas is enormous and updated frequently. The system needs to perform high-frequency decoding, action recognition, three-dimensional position mapping, and digital twin model rendering on multiple video streams in a short period of time, resulting in a significant increase in the data processing pressure of the three-dimensional visualization model of the operating scene. In this case, problems such as limited data transmission bandwidth, video analysis processing delays, and delayed three-dimensional model updates are prone to occur, directly causing time synchronization deviations between operational behaviors and three-dimensional models. This deviation can cause delays in the investigation of operational safety hazards, thereby affecting the real-time and accuracy of hazard identification, reducing the credibility and timeliness of the investigation results, and making it difficult to promptly detect and intervene in risky operational conditions.

[0086] According to an embodiment of the present invention, the acquisition frequency of video surveillance data of different operation areas is determined based on the delay status of the troubleshooting, and the data update strategy of the three-dimensional visualization model of the operation scene is determined based on the acquisition frequency to obtain an optimized visualization model, specifically:

[0087] Determining whether the data update of the three-dimensional visualization model of the operation scene meets the preset safety hazard inspection accuracy requirements based on the inspection delay status, and if not, determining the operation behavior activity parameter of each operation area based on the operation behavior vector, wherein the activity parameter is determined based on the number of operation behavior vectors in the operation area per unit time;

[0088] Obtain a time series of troubleshooting delays for each operating area over a continuous time period, establish a confidence interval for the troubleshooting delay based on the average delay value and delay variance in the time series, and compare the confidence interval with a preset troubleshooting delay threshold to obtain a delay sensitivity parameter for each operating area;

[0089] Based on the activity parameter and delay sensitivity parameter of the operation behavior, a scoring function for priority of obtaining data in the operation area is constructed. The scoring function assigns different weight factors to activity and delay sensitivity respectively, and calculates the priority score of obtaining video surveillance data in all operation areas;

[0090] Assign an operation risk level to the action type label corresponding to the operation behavior vector in each operation area to obtain a weighted operation risk score for each operation area;

[0091] Determine the frequency of video surveillance data acquisition for each operation area based on the operation risk weighted score and operation behavior activity parameter of each operation area, and construct a data update strategy for the three-dimensional visualization model of the operation scene based on the video surveillance data acquisition priority score and video surveillance data acquisition frequency of the operation area;

[0092] The three-dimensional visualization model of the operation scene is optimized according to the data update strategy to obtain an optimized visualization model.

[0093] It should be noted that by dynamically adjusting the acquisition frequency of video surveillance data in different operating areas according to the inspection delay status, and comprehensively constructing a data update strategy based on the operation behavior activity parameter, delay sensitivity parameter and operation risk weighted score, this method can realize the intelligent allocation and dynamic optimization of the update priority of the three-dimensional visualization model of the operation scene under the pressure of processing massive multi-channel video surveillance data and the constraints of limited computing resources, thereby significantly reducing the probability of data delay in high-activity and high-risk areas, improving the real-time visualization accuracy and inspection timeliness of operation behavior in key areas; at the same time, for areas with low delay sensitivity or low risk, reducing the frequency of data acquisition and update, effectively alleviating the global data bandwidth and computing load pressure, and optimizing system resource allocation; ultimately ensuring that in the complex and changeable marine fishing vessel operation environment, the three-dimensional visualization model always maintains priority perception and accurate synchronization of high-risk operation status, improving the response speed of hidden danger inspection and the reliability of safety management. The inspection delay confidence interval refers to the statistical interval composed of the mean value and fluctuation range (variance) of the inspection delay time series within a set time window, which is used to measure the uncertainty and confidence level of delay fluctuation.

[0094] According to an embodiment of the present invention, the operation safety assessment of the operating personnel of the target ocean fishing vessel is performed based on the optimized visualization model, and an operation safety hazard management strategy is constructed based on the operation safety, specifically:

[0095] Extracting the operator's action type label and position coordinate vector in real time based on the optimized visualization model, and calculating the Euclidean distance between the operator's position and the nearest danger zone boundary based on the danger zone boundary coordinates predefined in the optimized visualization model;

[0096] According to the predefined risk weight values ​​of different action types, the Euclidean distance is weighted to obtain real-time safety assessment parameters, and a safety risk accumulation function is constructed based on the duration data of the operator being in a dangerous warning state continuously;

[0097] When the safety assessment parameter is lower than the preset warning threshold, a real-time alarm level determination is performed according to the safety risk accumulation function value, and the alarm location coordinates are highlighted in the optimized visualization model;

[0098] Obtain the action type label, position coordinate vector, and duration data from the alarm event, analyze the causes of safety hazards through the risk grading logic tree, and determine the type of safety hazard and the category of illegal operation behavior;

[0099] Generate real-time warning information and correction time parameters of different levels according to the severity level of the safety hazard type and the duration data of the illegal operation behavior;

[0100] When multiple people are operating illegally at the same time in the alarm location coordinates, the regional violation density index is calculated and the correction time limit parameter is compressed and adjusted according to the regional violation density index;

[0101] Outputting the real-time alarm information, correction time limit parameters, and adjusted correction time limit parameters to a management terminal of the target operation area, and simultaneously driving the optimized visualization model to record the alarm behavior trajectory and start the correction time limit countdown display;

[0102] The updated safety assessment parameters in the alarm position are continuously monitored within the correction time limit. When the updated safety assessment parameters have not recovered to the safety threshold range, the alarm level is gradually increased according to the timeout duration ratio and the correction time limit is recalculated to obtain an operation safety hazard management strategy.

[0103] It should be noted that the described operational safety hazard management strategy establishes a cumulative safety risk function to continuously track the length of time workers spend in a danger warning state, achieving a time-series cumulative assessment of safety risks. When safety assessment parameters fall below a preset warning threshold, the system instantly issues a multi-level alert and highlights the alert location in a 3D visualization model, assisting managers in quickly locating the safety hazard. Combining the action type, location, and duration data from the alert event, a risk grading logic tree enables precise analysis of the cause of the safety hazard, identifying the hazard type and the category of the illegal operational behavior. The management strategy also generates corresponding alert information and correction deadlines based on the severity of the hazard and the duration of the violation. In scenarios involving multiple violations, the correction deadline is dynamically adjusted by calculating a regional violation density index, enabling prioritized control of high-risk areas. Alert information and correction parameters are pushed to the management terminal in real time, enabling on-site managers to respond promptly. The alert trajectory is also recorded in the 3D model, with a countdown display. This strategy continuously monitors the recovery of safety assessment parameters within the corrective action timeframe. If safety standards are not met, the alarm level is automatically upgraded based on the timeout ratio and the corrective action timeframe is recalculated, achieving dynamic closed-loop management of alarm handling. The overall technical effect is to significantly enhance the real-time identification and accurate early warning capabilities of operational safety hazards, strengthen the quantitative management and dynamic correction of violations, ensure a safe and stable operating environment for fishing vessels, and minimize accident risks.

[0104] According to an embodiment of the present invention, the further embodiment includes:

[0105] Based on the real-time alarm information in the optimized visualization model, the position coordinate vectors and action type labels of all alarm events are extracted. Based on the position coordinate vectors, spatial clustering analysis is performed on the alarm points in the three-dimensional space to generate an alarm spatial aggregation heat map and mark the alarm density core area. The alarm aggregation area range is divided according to the boundary of the alarm density core area;

[0106] Obtain the action type label set of the operators in each alarm aggregation area, classify the action type label set into the operation risk, count the number of alarm events with the same risk level in the area, and calculate the comprehensive risk index of the alarm aggregation area based on the safety risk accumulation function value corresponding to the alarm event;

[0107] Based on the comparison results of the comprehensive risk index and the preset risk level threshold, a behavioral logic association analysis is performed on multiple alarm events in the same alarm aggregation area. When consecutive alarm events have the same action type label sequence, the alarm events are merged into a continuous violation behavior chain, and the cumulative risk value of the behavior chain is calculated as a replacement alarm event;

[0108] Determine a real-time response priority ranking of the alarm aggregation area according to the number of the alternative alarm events and the comprehensive risk index of the alarm aggregation area, and establish an alarm response suppression coefficient, wherein the alarm response suppression coefficient decays nonlinearly with an increase in the number of alternative alarm events in the alarm aggregation area;

[0109] Dynamically filtering the original alarm information generated by the optimized visualization model according to the alarm response suppression coefficient, retaining only the alternative alarm events whose comprehensive risk index is higher than a preset filtering threshold to generate a compressed alarm list, and outputting the mapping relationship between the compressed alarm list and the spatial position of the alarm aggregation area to the management terminal;

[0110] In the optimized visualization model, the regional violation density index is marked for the alarm aggregation area corresponding to the compressed alarm list, and the correction time limit parameter is dynamically corrected according to the product of the regional violation density index and the alarm response suppression coefficient, and the corrected correction time limit parameter and the alarm position coordinates are output to the operation terminal of the target area.

[0111] According to an embodiment of the present invention, calculating the cumulative risk value of the behavior chain specifically includes:

[0112] Extract the security assessment parameters of the first alarm event in the chain of continuous violation behavior as the base value, and add the incremental part of the security risk accumulation function value of subsequent alarm events;

[0113] A time decay factor is introduced to perform weighted correction on the security risk accumulation function value of historical alarm events, and the time decay factor is negatively correlated with the time interval of the alarm event;

[0114] Update the position coordinate vector of the replacement alarm event to the geometric center point of the coordinates of all alarm points in the behavior chain.

[0115] It is important to note that in the safety monitoring scenario of marine fishing vessels, the alarm system faces the challenge of information overload in the context of multiple, continuous violations. Discrete alarm events lack behavioral continuity correlation and spatial and temporal integration. On the one hand, multiple alarms generated by the same individual's continuous violations are processed independently, resulting in high-risk persistent behaviors being obscured by a flood of instantaneous alarms. On the other hand, historical alarm data does not dynamically quantify the cumulative risk effect according to the time decay principle, making it difficult for managers to identify cumulative high-risk states. Furthermore, the fragmented alarm location information further exacerbates the difficulty of focusing emergency response. By designing a mechanism for calculating the cumulative risk value of a behavioral chain, this approach extracts the safety assessment parameters of the first alarm event in a chain of continuous violations as the risk base value, superimposes the incremental values ​​of the safety risk accumulation function corresponding to subsequent alarm events, and introduces a time decay factor negatively correlated with the alarm interval for dynamic weighting correction. This accurately constructs a cumulative risk signature reflecting the persistent harm of violations. This mechanism also drives an intelligent upgrade of alarm spatial positioning, integrating the location coordinates of alternative alarm events into the geometric center of the original alarm point, thus enabling the spatial aggregation of behavioral trajectories. This has achieved three breakthroughs in technical effectiveness: Consecutive alerts with the same action type label are integrated into a single behavioral logic chain, significantly reducing redundant alert interference; the risk quantification model is deeply coupled with temporal characteristics, dynamically presenting the exponential risk escalation patterns of persistent violations; and discrete alert spatial information is reconstructed into aggregated risk hotspots, enabling managers to quickly identify core risk sources within a 3D visualization interface. Ultimately, this creates a closed-loop optimization process from identification, quantification, to location of violations, ensuring a fundamental improvement in the efficiency of responding to critical, high-risk incidents.

[0116] Figure 4 A block diagram of a system for visually checking and managing safety hazards of marine fishing vessels according to the present invention is shown.

[0117] A second aspect of the present invention further provides a system 4 for visually checking and managing safety hazards of marine fishing vessels. The system comprises: a memory 41 and a processor 42. The memory comprises a program for visually checking and managing safety hazards of marine fishing vessels. When the program is executed by the processor, the following steps are implemented:

[0118] Obtaining video surveillance data of different operating areas of the target marine fishing vessel, performing operator action recognition and spatial distribution extraction on the video surveillance data, and constructing an operation behavior vector;

[0119] Acquiring ship structure data of a target ocean fishing vessel, constructing a three-dimensional model of the target ocean fishing vessel based on the ship structure data, and constructing a three-dimensional visualization model of an operation scene using the operation behavior vector and the three-dimensional model;

[0120] Performing a visual inspection of the target marine fishing vessel for potential safety hazards in operation based on the three-dimensional visualization model of the operation scene, obtaining a data update timestamp of the three-dimensional visualization model of the operation scene in real time, and determining a model inspection delay state based on the update timestamp;

[0121] Determine the acquisition frequency of video surveillance data of different operation areas according to the troubleshooting delay status, and determine the data update strategy of the three-dimensional visualization model of the operation scene according to the acquisition frequency to obtain an optimized visualization model;

[0122] An operational safety assessment is performed on the operating personnel of the target ocean fishing vessel based on the optimized visualization model, and an operational safety hazard management strategy is constructed based on the operational safety.

[0123] This invention discloses a system and method for visually detecting and managing safety hazards on marine fishing vessels, aiming to enhance the intelligence and visualization of fishing vessel operational safety management. The method includes: acquiring video surveillance data from different operating areas of a target marine fishing vessel to generate operational behavior vectors for the operators; constructing a three-dimensional model based on the vessel's structural data, fusing the operational behavior vectors with the three-dimensional model to generate a three-dimensional visualization model of the operational scene; conducting operational safety hazard inspections based on this model, and obtaining data update timestamps in real time to determine the inspection delay status; adjusting the acquisition frequency of video surveillance data based on the inspection delay status, formulating a data update strategy, and obtaining an optimized visualization model; and conducting operational safety assessments on the operators using the optimized model, thereby forming an operational safety hazard management strategy based on the assessment. This method can effectively identify and manage safety hazards, improving inspection accuracy and management efficiency.

[0124] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0125] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0126] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0127] Those skilled in the art will appreciate that all or part of the steps of the above-mentioned method embodiments may be implemented by hardware associated with program instructions, and the aforementioned program may be stored in a computer-readable storage medium. When the program is executed, the program executes the steps of the above-mentioned method embodiments. The aforementioned storage medium includes various media that can store program codes, such as mobile storage devices, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0128] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0129] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for visually checking and managing safety hazards of marine fishing vessels, characterized in that: The following steps are involved: Obtaining video surveillance data of different operating areas of the target marine fishing vessel, performing operator action recognition and spatial distribution extraction on the video surveillance data, and constructing an operation behavior vector; Acquiring ship structure data of a target ocean fishing vessel, constructing a three-dimensional model of the target ocean fishing vessel based on the ship structure data, and constructing a three-dimensional visualization model of an operation scene using the operation behavior vector and the three-dimensional model; Performing a visual inspection of the target marine fishing vessel for potential safety hazards in operation based on the three-dimensional visualization model of the operation scene, obtaining a data update timestamp of the three-dimensional visualization model of the operation scene in real time, and determining a model inspection delay state based on the update timestamp; Determine the acquisition frequency of video surveillance data of different operation areas according to the troubleshooting delay status, and determine the data update strategy of the three-dimensional visualization model of the operation scene according to the acquisition frequency to obtain an optimized visualization model; An operational safety assessment is performed on the operating personnel of the target ocean fishing vessel based on the optimized visualization model, and an operational safety hazard management strategy is constructed based on the operational safety.

2. A method for visually checking and managing safety hazards of marine fishing vessels according to claim 1, characterized in that: The acquisition of video surveillance data of different operating areas of the target ocean fishing vessel, the recognition of operator actions and extraction of spatial distribution of the video surveillance data, and the construction of an operating behavior vector are specifically as follows: Acquire video surveillance data of different operating areas of the target marine fishing vessel, extract a continuous video frame sequence of the video surveillance data, separate the foreground motion area of ​​each video frame based on a background subtraction algorithm, identify the outline of the operator based on the foreground motion area, calculate the minimum circumscribed matrix box of the operator outline, and obtain operator detection box data; Based on the operator detection frame data, a key point detection algorithm is applied to extract human skeleton information, including the coordinates of key points of the head, shoulders, elbows, wrists, hips, knees, and ankles. The angles and displacement changes between key points are calculated to obtain the dynamic feature vectors of the key points. Obtain key point change sequence data for different actions, perform action state modeling based on the hidden Markov model, define the action state transition probability matrix and observation probability matrix, and build an action recognition model; Importing the key point dynamic feature vector into the action recognition model, calculating the likelihood probability value of the key point dynamic feature and the action state sequence through a forward-backward algorithm, and identifying the action type of the operator, including walking, bending, carrying, climbing and stationary actions, when the likelihood probability value is greater than the action recognition threshold, to obtain action type data; Based on action type data and video surveillance data, the operator detection frame data is mapped to the three-dimensional coordinate system of the marine fishing vessel. The position coordinates of the operator relative to the ship structure are calculated through the perspective transformation matrix. Combined with the action type data, the operation behavior vector is constructed. The operation behavior vector includes the action type label, position coordinate vector and timestamp information.

3. The method for visually checking and managing safety hazards of marine fishing vessels according to claim 1 is characterized in that: The acquiring of the ship structure data of the target ocean fishing vessel, constructing a three-dimensional model of the target ocean fishing vessel according to the ship structure data, and constructing a three-dimensional visualization model of the operation scene by combining the operation behavior vector and the three-dimensional model, specifically: Acquire ship structure data of the target ocean fishing vessel, the ship structure data including hull geometry, deck layer layout, cabin plan, and structural component position information; extract hull contour lines, cabin boundary information, and structural component contour information based on the ship structure data to construct ship structure contour data; Establishing a hull surface mesh model and an internal cabin three-dimensional geometric model based on the ship structure contour data, topologically connecting the hull surface mesh model and the internal cabin three-dimensional geometric model to construct a three-dimensional ship structure, performing ship structure texture mapping on the three-dimensional ship structure, and constructing a three-dimensional mesh model of the target ocean fishing vessel; Convert the position coordinate vector in the operation behavior vector to the coordinate system of the three-dimensional grid model, and calculate the mapping coordinates of the operator's position in the three-dimensional grid model; Obtain animation data of different actions of the operators, build an action animation model library, match the action type in the operation behavior vector with the action animation model library, and map the matching action animation to the three-dimensional model in real time according to the mapping coordinates and timestamp information of the operators based on the digital twin, and build a three-dimensional visualization model of the target ocean fishing vessel operation scene based on the digital twin.

4. A method for visually checking and managing safety hazards of marine fishing vessels according to claim 1, characterized in that: The method of visually checking the target marine fishing vessel for potential safety hazards in operation based on the three-dimensional visualization model of the operation scene, obtaining a data update timestamp of the three-dimensional visualization model of the operation scene in real time, and determining the model's inspection delay state based on the update timestamp is specifically as follows: Perform a visual inspection of safety operation hazards of the target marine fishing vessel based on the three-dimensional visualization model of the operation scene, and obtain in real time the timestamp information of the video frame corresponding to each operation behavior vector input into the three-dimensional visualization model of the operation scene as a data update timestamp; Extracting the current clock system time of the three-dimensional visualization model of the operation scene in real time as a reference timestamp, and calculating the time difference between the data update timestamp and the reference timestamp as a single troubleshooting delay; Continuously collect multiple single-check delay data within a preset time window to form a time series, and calculate the average delay value of the single-check delay time series within the time window; The average delay value is compared with the delay status reference interval set in advance based on the troubleshooting accuracy requirements, the troubleshooting delay status is determined according to the position of the average delay value in the reference interval, and the troubleshooting delay status is marked in real time in the troubleshooting status information of the three-dimensional visualization model of the operation scene.

5. The method for visually checking and managing safety hazards of marine fishing vessels according to claim 1 is characterized in that: The acquisition frequency of video surveillance data of different operation areas is determined according to the delay state of the troubleshooting, and the data update strategy of the three-dimensional visualization model of the operation scene is determined according to the acquisition frequency to obtain the optimized visualization model, specifically: Determining whether the data update of the three-dimensional visualization model of the operation scene meets the preset safety hazard inspection accuracy requirements based on the inspection delay status, and if not, determining the operation behavior activity parameter of each operation area based on the operation behavior vector, wherein the activity parameter is determined based on the number of operation behavior vectors in the operation area per unit time; Obtain a time series of troubleshooting delays for each operating area over a continuous time period, establish a confidence interval for the troubleshooting delay based on the average delay value and delay variance in the time series, and compare the confidence interval with a preset troubleshooting delay threshold to obtain a delay sensitivity parameter for each operating area; Based on the activity parameter and delay sensitivity parameter of the operation behavior, a scoring function for priority of obtaining data in the operation area is constructed. The scoring function assigns different weight factors to activity and delay sensitivity respectively, and calculates the priority score of obtaining video surveillance data in all operation areas; Assign an operation risk level to the action type label corresponding to the operation behavior vector in each operation area to obtain a weighted operation risk score for each operation area; Determine the frequency of video surveillance data acquisition for each operation area based on the operation risk weighted score and operation behavior activity parameter of each operation area, and construct a data update strategy for the three-dimensional visualization model of the operation scene based on the video surveillance data acquisition priority score and video surveillance data acquisition frequency of the operation area; The three-dimensional visualization model of the operation scene is optimized according to the data update strategy to obtain an optimized visualization model.

6. The method for visually checking and managing safety hazards of marine fishing vessels according to claim 1 is characterized in that: The operation safety assessment of the operating personnel of the target ocean fishing vessel is performed according to the optimized visualization model, and an operation safety hazard management strategy is constructed according to the operation safety, specifically: Extracting the operator's action type label and position coordinate vector in real time based on the optimized visualization model, and calculating the Euclidean distance between the operator's position and the nearest danger zone boundary based on the danger zone boundary coordinates predefined in the optimized visualization model; According to the predefined risk weight values ​​of different action types, the Euclidean distance is weighted to obtain real-time safety assessment parameters, and a safety risk accumulation function is constructed based on the duration data of the operator being in a dangerous warning state continuously; When the safety assessment parameter is lower than the preset warning threshold, a real-time alarm level determination is performed according to the safety risk accumulation function value, and the alarm location coordinates are highlighted in the optimized visualization model; Obtain the action type label, position coordinate vector, and duration data from the alarm event, analyze the causes of safety hazards through the risk grading logic tree, and determine the type of safety hazard and the category of illegal operation behavior; Generate real-time warning information and correction time parameters of different levels according to the severity level of the safety hazard type and the duration data of the illegal operation behavior; When multiple people are operating illegally at the same time in the alarm location coordinates, the regional violation density index is calculated and the correction time limit parameter is compressed and adjusted according to the regional violation density index; Outputting the real-time alarm information, correction time limit parameters, and adjusted correction time limit parameters to a management terminal of the target operation area, and simultaneously driving the optimized visualization model to record the alarm behavior trajectory and start the correction time limit countdown display; The updated safety assessment parameters in the alarm position are continuously monitored within the correction time limit. When the updated safety assessment parameters have not recovered to the safety threshold range, the alarm level is gradually increased according to the timeout duration ratio and the correction time limit is recalculated to obtain an operation safety hazard management strategy.

7. A visual inspection and management system for potential safety hazards of marine fishing vessels, characterized by: The system for visually checking and managing hidden dangers of marine fishing vessels includes a memory and a processor. The memory includes a program for visually checking and managing hidden dangers of marine fishing vessels. When the program is executed by the processor, the following steps are implemented: Obtaining video surveillance data of different operating areas of the target marine fishing vessel, performing operator action recognition and spatial distribution extraction on the video surveillance data, and constructing an operation behavior vector; Acquiring ship structure data of a target ocean fishing vessel, constructing a three-dimensional model of the target ocean fishing vessel based on the ship structure data, and constructing a three-dimensional visualization model of an operation scene using the operation behavior vector and the three-dimensional model; Performing a visual inspection of the target marine fishing vessel for potential safety hazards in operation based on the three-dimensional visualization model of the operation scene, obtaining a data update timestamp of the three-dimensional visualization model of the operation scene in real time, and determining a model inspection delay state based on the update timestamp; Determine the acquisition frequency of video surveillance data of different operation areas according to the troubleshooting delay status, and determine the data update strategy of the three-dimensional visualization model of the operation scene according to the acquisition frequency to obtain an optimized visualization model; An operational safety assessment is performed on the operating personnel of the target ocean fishing vessel based on the optimized visualization model, and an operational safety hazard management strategy is constructed based on the operational safety.

Images