Monitoring system for whole process of offshore project demolition operation

Abstract

The invention provides an offshore project demolition operation whole process monitoring system. The system comprises a data acquisition module, a work ticket generation module, a demolition operation path planning module, a demolition data collection and homing module, an illegal behavior identification module, a risk analysis module and a visualization module. The whole process of dismantling operation is monitored under the condition that a complex marine environment is considered, and a work ticket generation module is used for generating a work operation ticket; generating an operation path planning graph by using a dismantling operation path planning module; the illegal behavior recognition module is used for monitoring illegal behaviors in the demolition operation process for accurate recognition, the risk analysis module is used for recognizing and feeding back risk types possibly corresponding to the illegal behaviors, and meanwhile potential risks in the demolition operation process are analyzed for content recognition; through effective monitoring, the safety and efficiency of operation can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of offshore operation monitoring, and in particular to a full-process monitoring system for offshore project demolition operations. Background Art

[0002] At present, offshore operation platforms are simulated land areas built in seawater for offshore operations, for production operations or other activities, such as oil extraction on the seabed, offshore construction operations, etc. With the combination of 3D visualization and offshore operation platforms, it is more convenient for construction personnel and supervisors to control the production operation process.

[0003] However, in traditional project management, there are problems of information islands and slow response, which will reduce the safety and efficiency of offshore demolition operations. In the process of realizing full-process intelligent monitoring and management of offshore demolition operations, the existing technology still has deficiencies in data acquisition, information processing, violation identification and risk analysis. For this reason, the present application provides a full-process monitoring system for offshore project demolition operations to improve the safety, efficiency and engineering management level of demolition operations. Summary of the invention

[0004] The purpose of the present invention is to provide a whole-process monitoring system for offshore project dismantling operations to solve the problems raised in the above-mentioned background technology.

[0005] The present invention is realized by the following technical scheme: a whole process monitoring system for offshore project demolition operation, the system comprising a data acquisition module, a work ticket generation module, a demolition operation path planning module, a demolition data collection module, a violation identification module, a risk analysis module, and a visualization module;

[0006] The data acquisition module uses drones to perform three-dimensional scanning of the demolition area to obtain high-precision three-dimensional data, uses a combination of laser radar and high-definition cameras to obtain video image data and create a three-dimensional model of the area, and collects information on demolition team personnel, major machinery and ships;

[0007] The work ticket generation module obtains the demolition area, demolition team personnel information, demolition project, demolition objects, demolition sequence, demolition estimated completion time, and generates a work operation ticket;

[0008] The demolition operation path planning module constructs a high-precision three-dimensional visualization model based on the acquired three-dimensional data, allowing the user to intuitively view the structure and potential risks of the demolition area in a virtual environment, and generates an operation path planning map based on the reinforcement learning path planning algorithm while combining the spatial position of the objects and the demolition sequence;

[0009] The demolition data collection module is used to collect all data related to the demolition operation, including the content of the work operation ticket and video images, to form a complete operation file;

[0010] The violation identification module analyzes the real-time video stream of the acquired video image data and automatically identifies the violation of the operator by combining a multi-channel temporal convolutional network with a self-attention mechanism;

[0011] The risk analysis module is used to provide feedback on the risk types that may correspond to the illegal behavior, and to analyze and identify the potential risks in the demolition operation process;

[0012] The visualization module is used to provide a visualization interface, supporting real-time updates on mobile and computer terminals, and real-time monitoring and tracking of the demolition progress, whether objects have been demolished, and whether the demolition sequence is correct, and displaying them on the interface.

[0013] Specifically, high-precision three-dimensional data will be obtained for data cleaning to remove noise and duplicate points. The processed three-dimensional data will be used to generate a three-dimensional mesh model, which will be smoothed using a triangulation algorithm to form a continuous and flat surface. A texture mapping algorithm will be used to add material properties to the model, and structural features and potential hazard locations will be marked on the model.

[0014] Specifically, in the process of generating a three-dimensional mesh model, a triangulation algorithm is used for smoothing to form a continuous and flat surface, as follows:

[0015] For each vertex, identify the neighborhood triangles around it and calculate the average position of the vertex. The neighborhood triangles are set to a fixed radius or number of neighborhood points to ensure that the processed result is smooth without losing structure.

[0016] The coordinate position of the current point is updated by averaging the coordinates of the neighboring points, and the normal vector is kept consistent during the smoothing process to avoid surface inversion.

[0017] Specifically, a texture mapping algorithm is used to add material properties to the 3D network model, and the structural features and potential hazards are marked on the model. The texture selection is as follows:

[0018] Insert the calibrated and processed 3D data into the 3D network model, and use normal maps and displacement maps to enhance the 3D effect and material texture. Normal maps can generate detailed surfaces under lighting conditions.

[0019] Identify and annotate key structural features and potential damage areas in the 3D network model, extract key structures in the model, and ensure that they are linked to the key points of the actual demolition operation;

[0020] The potential hazard areas are divided into colors: red represents the heavy object hanging area to warn of potential safety risks; yellow represents the damaged area to indicate where special attention needs to be paid; blue represents the height of the hazard area, including the gap between the upper and lower layers, so that workers can understand the working space.

[0021] Specifically, the process of generating a work operation ticket by the work ticket generation module specifically includes:

[0022] The operation ticket will list in detail the name, skills, health status, and division of responsibilities of each member in the operation;

[0023] Automatically identify objects to be demolished, including the type of object, material characteristics, potential hazards, and display historical information of the demolition project and accurate marking of the location of the object;

[0024] In the generated operation ticket, a real-time updated 3D model is embedded to clearly mark the location of the demolition object and use color coding to indicate different danger levels;

[0025] Combining the previous data analysis and on-site investigation, the system automatically generates recommended demolition steps and provides operating specifications, including necessary safety measures and emergency plans;

[0026] Based on historical cases and real-time data, predict the duration of each demolition step and generate an estimated completion time in the operation ticket.

[0027] Specifically, the specific process of the path planning algorithm based on reinforcement learning in the demolition operation path planning module to generate an operation path planning diagram by combining the spatial position of the object and the demolition sequence is as follows:

[0028] Update formula for reinforcement learning:

[0029] Q(s t , a t )←Q(s t , a t )+α((R(s t , a t )+γmax Q(s t+1 , a)-Q(s t , a t ))

[0030] Among them, Q is to choose the best action to obtain the maximum total benefit, Q(s t , a t ) means in s t Execute action a in state t The expected long-term cumulative reward, max Q(s t+1 , a) means in state s t+1The maximum value of the Q value of all possible actions a is taken, which means the best expected benefit that can be obtained in the next state, s t is the state at time t, a t is the action at time t, α is the learning rate, R is the reward function, γ is the discount factor, s t+1 is the state at time t+1.

[0031] To maximize the expected reward of the final path P, then:

[0032] P = arg max Q(s t , a t )

[0033] The trust of path selection is introduced, and the reputation value C(P) is defined to evaluate the reliability of different paths P. The formula is as follows:

[0034] C(P)=ω c ·C t (P)+ω r ·R(P)

[0035] Among them, C t (P) is the path feasibility record based on historical data, ω c and ω r They are all reputation evaluation weights, R(P) is the security score of path P, reflecting the potential risk of the path;

[0036] The final optimization objective function is updated as:

[0037]

[0038] Among them, Γ(P) is the path optimization objective function, ω 1 is the time-related weight that affects the speed of path selection, T(P) is the total time of path P, which measures the efficiency of the path, ω 2 is the risk-related weight that affects the speed of path selection, R(P) is the safety score of path P, reflecting the potential risk of the path, ω 3 is the weight associated with the reputation value.

[0039] Specifically, the process of the violation identification module using a multi-channel temporal convolutional network combined with a self-attention mechanism to automatically identify the violation of the operator specifically includes:

[0040] First, set the time window to T and extract the feature vector x for each video frame. t ∈R d , where t represents the time index and d represents the feature dimension; the video frame sequence is represented as X = [x 1 , x 2 , x3 , …, x T ]∈R T×d

[0041] Second, perform a convolution operation on the input feature sequence, and parameterize the convolution kernel as W∈R K×d , where K is the convolution kernel size, the convolution output is calculated as:

[0042] Z t =σ(W*x t +b)

[0043] Among them, σ is the activation function, b is the bias, and Z t is the output value of the convolutional layer;

[0044] Third, obtain the global feature G through the maximum pooling operation:

[0045] G=MaxPool(Z t )

[0046] Fourth, after obtaining the global features, the self-attention mechanism is introduced to enhance the representation ability of the features. i ∈G, calculate the attention weight:

[0047]

[0048] Among them, A ij is the global feature g i For the global feature g j The attention weight, g i is the ith global feature, g j is the jth global feature, n is the number of global features, g i T Represents the global feature g i The transpose of

[0049] Fifth, the attention weights are applied to the global features to form the weighted features G att :

[0050]

[0051] Sixth, the weighted features are mapped through the fully connected layer to output the probability of violation:

[0052] y=softmax(W fc G att +b fc )

[0053] Among them, W fc and b fc They are all parameters of the fully connected layer, and softmax means normalization.

[0054] Specifically, the risk analysis module specifically includes calculating the risk value R of each violation h h :

[0055] R h =f(P h , S, E)

[0056] Among them, P h is the probability of violation h, S is the environmental factor, and E is the influencing factor of external events.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] The present invention provides a full-process monitoring system for offshore project demolition operations, which provides a solid guarantee for efficient and safe demolition in complex offshore environments, avoids the problems of information islands and slow response in traditional project management, and will greatly improve the safety and efficiency of offshore demolition operations, with broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only preferred embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 A module block diagram of a full-process monitoring system for offshore project dismantling operations provided by the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical scheme and advantages of the present invention more obvious, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described in the present invention, all other embodiments obtained by those skilled in the art without creative work should fall within the protection scope of the present invention.

[0062] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present invention. However, it is apparent to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some technical features well known in the art are not described.

[0063] It should be understood that the present invention can be implemented in different forms and should not be interpreted as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to make the disclosure thorough and complete and to fully convey the scope of the present invention to those skilled in the art.

[0064] The purpose of the terms used herein is only to describe specific embodiments and is not intended to be limiting of the present invention. When used herein, the singular forms "one", "an" and "said / the" are also intended to include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "consisting of" and / or "comprising", when used in this specification, determine the presence of the features, integers, steps, operations, elements and / or parts, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, parts and / or groups. When used herein, the term "and / or" includes any and all combinations of the relevant listed items.

[0065] In order to fully understand the present invention, a detailed structure will be presented in the following description to illustrate the technical solution proposed by the present invention. The optional embodiments of the present invention are described in detail as follows, but in addition to these detailed descriptions, the present invention may also have other implementations.

[0066] See also Figure 1 , a whole process monitoring system for offshore project demolition operations, the system includes a data acquisition module, a work ticket generation module, a demolition operation path planning module, a demolition data collection module, a violation identification module, a risk analysis module, and a visualization module;

[0067] The data acquisition module uses drones to perform three-dimensional scanning of the demolition area to obtain high-precision three-dimensional data, uses a combination of laser radar and high-definition cameras to obtain video image data and create a three-dimensional model of the area, and collects information on demolition team personnel, major machinery and ships;

[0068] The work ticket generation module obtains the demolition area, demolition team personnel information, demolition project, demolition objects, demolition sequence, demolition estimated completion time, and generates a work operation ticket;

[0069] The demolition operation path planning module constructs a high-precision three-dimensional visualization model based on the acquired three-dimensional data, allowing the user to intuitively view the structure and potential risks of the demolition area in a virtual environment, and generates an operation path planning map based on the reinforcement learning path planning algorithm while combining the spatial position of the objects and the demolition sequence;

[0070] The demolition data collection module is used to collect all data related to the demolition operation, including the content of the work operation ticket and video images, to form a complete operation file;

[0071] The violation identification module analyzes the real-time video stream of the acquired video image data and automatically identifies the violation of the operator by combining a multi-channel temporal convolutional network with a self-attention mechanism;

[0072] The risk analysis module is used to provide feedback on the risk types that may correspond to the illegal behavior, and to analyze and identify the potential risks in the demolition operation process;

[0073] The visualization module is used to provide a visualization interface, supporting real-time updates on mobile and computer terminals, and real-time monitoring and tracking of the demolition progress, whether objects have been demolished, and whether the demolition sequence is correct, and displaying them on the interface.

[0074] Exemplarily, the present invention provides a system for monitoring the whole process of demolition operations of offshore projects, which can monitor the whole process of demolition operations under complex offshore environments. Its working principle is as follows: first, the data acquisition module collects the information of demolition team personnel, main machinery and ship dynamic information, focuses on monitoring the working status and dynamics of the ship, and cooperates with the ship traffic network to check the working status of the ship to determine whether it is working normally; second, the work ticket generation module obtains the demolition area, demolition team personnel information, demolition projects, demolition objects, demolition sequence, and estimated completion time of demolition, and generates a work operation ticket; third, the demolition operation path planning module establishes a three-dimensional view, and generates an operation path plan according to the object position and demolition sequence. Fourth, the demolition data collection module collects all the data related to the demolition operation, including the content of the work operation ticket, video images, etc., to form a complete operation file for subsequent analysis and review. Fifth, the violation identification module monitors the violation during the demolition operation and accurately identifies it. Sixth, the risk analysis module identifies and feedbacks the risk type that the violation may correspond to, and analyzes and identifies the potential risks in the demolition operation process. Seventh, the visualization module provides a visualization interface, supports real-time updates on mobile and computer terminals, monitors and tracks the demolition progress, whether the objects are demolished, and whether the demolition sequence is correct, and displays it on the interface to understand the operation status in real time. Through effective monitoring, the safety and efficiency of the operation can be improved.

[0075] Specifically, high-precision three-dimensional data will be obtained for data cleaning to remove noise and duplicate points. The processed three-dimensional data will be used to generate a three-dimensional mesh model, which will be smoothed using a triangulation algorithm to form a continuous and flat surface. A texture mapping algorithm will be used to add material properties to the model, and structural features and potential hazard locations will be marked on the model.

[0076] Specifically, in the process of generating a three-dimensional mesh model, a triangulation algorithm is used for smoothing to form a continuous and flat surface, as follows:

[0077] For each vertex, identify the neighborhood triangles around it and calculate the average position of the vertex. The neighborhood triangles are set to a fixed radius or number of neighborhood points to ensure that the processed result is smooth without losing structure.

[0078] The coordinate position of the current point is updated by averaging the coordinates of the neighboring points, and the normal vector is kept consistent during the smoothing process to avoid surface inversion.

[0079] Exemplarily, smoothing is performed to ensure that the generated three-dimensional mesh model not only has good visual effects but also effectively retains structural features.

[0080] Specifically, a texture mapping algorithm is used to add material properties to the 3D network model, and the structural features and potential hazards are marked on the model. The texture selection is as follows:

[0081] Insert the calibrated and processed 3D data into the 3D network model, and use normal maps and displacement maps to enhance the 3D effect and material texture. Normal maps can generate detailed surfaces under lighting conditions.

[0082] For example, material mapping and detail enhancement are important steps in 3D modeling, especially when practical applications are required, such as demolition operations, this step is directly related to the authenticity and operability of the model.

[0083] Identify and annotate key structural features and potential damage areas in the 3D network model, extract key structures in the model, and ensure that they are linked to the key points of the actual demolition operation;

[0084] The potential hazard areas are divided into colors: red represents the heavy object hanging area to warn of potential safety risks; yellow represents the damaged area to indicate where special attention needs to be paid; blue represents the height of the hazard area, including the gap between the upper and lower layers, so that workers can understand the working space.

[0085] Specifically, the process of generating a work operation ticket by the work ticket generation module specifically includes:

[0086] The operation ticket will list in detail the name, skills, health status, and division of responsibilities of each member in the operation;

[0087] Automatically identify objects to be demolished, including the type of object, material characteristics, potential hazards, and display historical information of the demolition project and accurate marking of the location of the object;

[0088] In the generated operation ticket, a real-time updated 3D model is embedded to clearly mark the location of the demolition object and use color coding to indicate different danger levels;

[0089] Combining the previous data analysis and on-site investigation, the system automatically generates recommended demolition steps and provides operating specifications, including necessary safety measures and emergency plans;

[0090] Based on historical cases and real-time data, predict the duration of each demolition step and generate an estimated completion time in the operation ticket.

[0091] Exemplarily, the content of the work operation ticket is as follows:

[0092] Project Name: Offshore Wind Farm Foundation Demolition

[0093] Creation time: September 19, 2024

[0094] Project number:OCEAN-2024-018

[0095] 1. Removal of team information

[0096] Team Leader:Li Yi

[0097] Members:

[0098] Liu Er-Skills: Offshore Construction, Mechanical Operation; Health: Good Chen San-Skills: Electrician, Equipment Maintenance; Health: Good Zhang Si-Skills: Safety Management; Health: Good

[0099] Zhao Wu - Skills: Diving Operations; Health: Good

[0100] 2. Description of demolition project

[0101] Target object: Offshore wind turbine foundation removal Reason: Equipment update, construction of new type of wind turbine

[0102] Object type: Reinforced concrete foundation

[0103] Address: Beihai Wind Farm

[0104] Coordinates: (23.55°N, 118.92°E)

[0105] 3. Object location

[0106] Mark location:

[0107] 3D views generated using 3D modeling software such as Revit or Navisworks: 3D Model

[0108] The key areas marked in the figure are:

[0109] Area A: Dismantling the foundation of the first wind turbine

[0110] Area B: Dismantle the second wind turbine foundation

[0111] Area C: Cleaning work area

[0112] 4. Dismantling order

[0113] Step breakdown:

[0114] Make sure all safety measures are in place (life jackets, scuba gear, etc.).

[0115] Check offshore weather conditions and tide forecasts to ensure a safe operating window.

[0116] Conduct on-site investigation and confirm the work area.

[0117] Use cutting equipment to dismantle the foundation of the first wind turbine in Area A.

[0118] Use cutting equipment to dismantle the foundation of the second wind turbine in Area B.

[0119] Clean up the debris after demolition and keep records.

[0120] Conduct safety inspections and site assessments.

[0121] 5. Estimated time of completion

[0122] Estimated time for removal:

[0123] Based on historical data analysis and site conditions, the entire demolition process is estimated to take 10 hours.

[0124] Time allocation:

[0125] Preparation: 1 hour

[0126] Demolition work: 7 hours

[0127] Cleaning and safety inspection: 2 hours

[0128] Specifically, the specific process of the path planning algorithm based on reinforcement learning in the demolition operation path planning module to generate an operation path planning diagram by combining the spatial position of the object and the demolition sequence is as follows:

[0129] Update formula for reinforcement learning:

[0130] Q(s t , a t )←Q(s t , a t )+α((R(s t , a t )+γmax Q(s t+1 , a)-Q(s t , a t ))

[0131] Among them, Q is to choose the best action to obtain the maximum total benefit, Q(st , a t ) means in s t Execute action a in state t The expected long-term cumulative reward, max Q(s t+1 , a) means in state s t+1 The maximum value of the Q value of all possible actions a is taken, which means the best expected benefit that can be obtained in the next state, s t is the state at time t, a t is the action at time t, α is the learning rate, R is the reward function, γ is the discount factor, s t+1 is the state at time t+1.

[0132] To maximize the expected reward of the final path P, then:

[0133] P = arg max Q(s t , a t )

[0134] The trust of path selection is introduced, and the reputation value C(P) is defined to evaluate the reliability of different paths P. The formula is as follows:

[0135] C(P)=ω c ·C t (P)+ω r ·R(P)

[0136] Among them, C t (P) is the path feasibility record based on historical data, ω c and ω r They are all reputation evaluation weights, R(P) is the security score of path P, reflecting the potential risk of the path;

[0137] The final optimization objective function is updated as:

[0138]

[0139] Among them, Γ(P) is the path optimization objective function, ω 1 is the time-related weight that affects the speed of path selection, T(P) is the total time of path P, which measures the efficiency of the path, ω 2 is the risk-related weight that affects the speed of path selection, R(P) is the safety score of path P, reflecting the potential risk of the path, ω 3 is the weight associated with the reputation value.

[0140] Exemplarily, a path planning algorithm based on reinforcement learning is adopted, and the path is optimized with multiple objectives. Finally, the optimized path and related risk markers are displayed on a three-dimensional visualization model, which helps to improve the efficiency and safety of demolition operations.

[0141] Specifically, the process of the violation identification module using a multi-channel temporal convolutional network combined with a self-attention mechanism to automatically identify the violation of the operator specifically includes:

[0142] First, set the time window to T and extract the feature vector x for each video frame. t ∈R d , where t represents the time index and d represents the feature dimension; the video frame sequence is represented as X = [x 1 , x 2 , x 3 , …, x T ]∈R T×d

[0143] Second, perform a convolution operation on the input feature sequence, and parameterize the convolution kernel as W∈R K×d , where K is the convolution kernel size, the convolution output is calculated as:

[0144] Z t =σ(W*x t +b)

[0145] Among them, σ is the activation function, b is the bias, and Z t is the output value of the convolutional layer;

[0146] Third, obtain the global feature G through the maximum pooling operation:

[0147] G=MaxPool(Z t )

[0148] Fourth, after obtaining the global features, the self-attention mechanism is introduced to enhance the representation ability of the features. i ∈G, calculate the attention weight:

[0149]

[0150] Among them, A ij is the global feature g i For the global feature g j The attention weight, g i is the ith global feature, g j is the jth global feature, n is the number of global features, g i T Represents the global feature g i The transpose of

[0151] Fifth, the attention weights are applied to the global features to form the weighted features G att :

[0152]

[0153] Sixth, the weighted features are mapped through the fully connected layer to output the probability of violation:

[0154] y=softmax(W fc G att +b fc )

[0155] Among them, W fc and b fc They are all parameters of the fully connected layer, and softmax means normalization.

[0156] For example, by combining a multi-channel temporal convolutional network with a self-attention mechanism, the accuracy and robustness of violation identification are improved, which not only improves the recognition efficiency, but also enhances the adaptability to complex working environments, providing a strong guarantee for on-site operation safety.

[0157] Specifically, the risk analysis module specifically includes calculating the risk value R of each violation h h :

[0158] R h =f(P h , S, E)

[0159] Among them, P h is the probability of violation h, S is the environmental factor, and E is the influencing factor of external events.

[0160] For example, through the analysis and identification of violations and potential risks, accidents can be effectively prevented, the safety of personnel and equipment can be guaranteed, and the smooth completion of the project can be promoted.

[0161] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A monitoring system for the whole process of offshore project dismantling operations, characterized in that: The system includes a data acquisition module, a work ticket generation module, a demolition operation path planning module, a demolition data collection module, a violation identification module, a risk analysis module, and a visualization module; The data acquisition module uses drones to perform three-dimensional scanning of the demolition area to obtain high-precision three-dimensional data, uses a combination of laser radar and high-definition cameras to obtain video image data and create a three-dimensional model of the area, and collects information on demolition team personnel, major machinery and ships; The work ticket generation module obtains the demolition area, demolition team personnel information, demolition project, demolition objects, demolition sequence, demolition estimated completion time, and generates a work operation ticket; The demolition operation path planning module constructs a high-precision three-dimensional visualization model based on the acquired three-dimensional data, allowing the user to intuitively view the structure and potential risks of the demolition area in a virtual environment, and generates an operation path planning map based on the reinforcement learning path planning algorithm while combining the spatial position of the objects and the demolition sequence; The demolition data collection module is used to collect all data related to the demolition operation, including the content of the work operation ticket and video images, to form a complete operation file; The violation identification module analyzes the real-time video stream of the acquired video image data and automatically identifies the violation of the operator by combining a multi-channel temporal convolutional network with a self-attention mechanism; The risk analysis module is used to provide feedback on the risk types that may correspond to the illegal behavior, and to analyze and identify the potential risks in the demolition operation process; The visualization module is used to provide a visualization interface, supporting real-time updates on mobile and computer terminals, and real-time monitoring and tracking of the demolition progress, whether objects have been demolished, and whether the demolition sequence is correct, and displaying them on the interface.

2. A monitoring system for the whole process of offshore project dismantling operations according to claim 1, characterized in that: The high-precision three-dimensional data is cleaned to remove noise and duplicate points. The processed three-dimensional data is used to generate a three-dimensional mesh model, which is smoothed using a triangulation algorithm to form a continuous and flat surface. The texture mapping algorithm is used to add material properties to the model, and the structural features and locations of potential hazards are marked on the model.

3. A monitoring system for the whole process of offshore project dismantling operations according to claim 2, characterized in that: In the process of generating a 3D mesh model, a triangulation algorithm is used for smoothing to form a continuous and flat surface, as follows: For each vertex, identify the neighborhood triangles around it and calculate the average position of the vertex. The neighborhood triangles are set to a fixed radius or number of neighborhood points to ensure that the processed result is smooth without losing structure. The coordinate position of the current point is updated by averaging the coordinates of the neighboring points, and the normal vector is kept consistent during the smoothing process to avoid surface inversion.

4. A monitoring system for the whole process of offshore project dismantling operations according to claim 3, characterized in that: Use a texture mapping algorithm to add material properties to the 3D network model and annotate the structure features and potential hazards on the model. Texture selection is as follows: Insert the calibrated and processed 3D data into the 3D network model, and use normal maps and displacement maps to enhance the 3D effect and material texture. Normal maps can generate detailed surfaces under lighting conditions. Identify and annotate key structural features and potential damage areas in the 3D network model, extract key structures in the model, and ensure that they are linked to the key points of the actual demolition operation; The potential hazard areas are divided into colors: red represents the heavy object hanging area to warn of potential safety risks; yellow represents the damaged area to indicate where special attention needs to be paid; blue represents the height of the hazard area, including the gap between the upper and lower layers, so that workers can understand the working space.

5. A monitoring system for the whole process of offshore project dismantling operations according to claim 4, characterized in that: The process of generating a work operation ticket by the work ticket generation module specifically includes: The operation ticket will list in detail the name, skills, health status, and division of responsibilities of each member in the operation; Automatically identify objects to be demolished, including the type of object, material characteristics, potential hazards, and display historical information of the demolition project and accurate marking of the location of the object; In the generated operation ticket, a real-time updated 3D model is embedded to clearly mark the location of the demolition object and use color coding to indicate different danger levels; Combining the previous data analysis and on-site investigation, the system automatically generates recommended demolition steps and provides operating specifications, including necessary safety measures and emergency plans; Based on historical cases and real-time data, predict the duration of each demolition step and generate an estimated completion time in the operation ticket.

6. A monitoring system for the whole process of offshore project dismantling operations according to claim 5, characterized in that: The specific process of the path planning algorithm based on reinforcement learning in the demolition operation path planning module to generate the operation path planning diagram by combining the spatial position of the object and the demolition sequence is as follows: Update formula for reinforcement learning: Q(s t ,a t )←Q(s t ,a t )+α((R(s t ,a t )+γmax Q(s t+1 ,a)-Q(s t ,a t )) Among them, Q is to choose the best action to obtain the maximum total benefit, Q(s t , a t ) means in s t Execute action a in state t The expected long-term cumulative reward, max Q(s t+1 , a) means in state s t+1 The maximum value of the Q value of all possible actions a is taken, which means the best expected benefit that can be obtained in the next state, s t is the state at time t, a t is the action at time t, α is the learning rate, R is the reward function, γ is the discount factor, s t+1 is the state at time t+1. To maximize the expected reward of the final path P, then: P=arg max Q(s t ,a t ) The trust of path selection is introduced, and the reputation value C(P) is defined to evaluate the reliability of different paths P. The formula is as follows: C(P)=ω c ·C t (P)+ω r ·R(P) Among them, C t (P) is the path feasibility record based on historical data, ω c and ω r They are all reputation evaluation weights, R(P) is the security score of path P, reflecting the potential risk of the path; The final optimization objective function is updated as follows: Among them, Γ(P) is the path optimization objective function, ω1 is the time-related weight that affects the speed of path selection, T(P) is the total time of path P, which measures the efficiency of the path, ω2 is the risk-related weight that affects the speed of path selection, R(P) is the security score of path P, reflecting the potential risk of the path, and ω3 is the weight related to the reputation value.

7. A monitoring system for the whole process of offshore project dismantling operations according to claim 6, characterized in that: The process of the violation identification module using a multi-channel temporal convolutional network combined with a self-attention mechanism to automatically identify the violation of the operator specifically includes: First, set the time window to T and extract the feature vector x for each video frame. t ∈R d , where t represents the time index and d represents the feature dimension; the video frame sequence is represented as X = [x1, x2, x3, ..., x T ]∈R T×d Second, perform a convolution operation on the input feature sequence, and parameterize the convolution kernel as W∈R K×d , where K is the convolution kernel size, the convolution output is calculated as: Z t =σ(W*x t +b) Among them, σ is the activation function, b is the bias, and Z t is the output value of the convolutional layer; Third, obtain the global feature G through the maximum pooling operation: G=MaxPool(Z t ) Fourth, after obtaining the global features, the self-attention mechanism is introduced to enhance the representation ability of the features. i ∈G, calculate the attention weight: Among them, A ij is the global feature g i For the global feature g j The attention weight, g i is the ith global feature, g j is the jth global feature, n is the number of global features, g i T Represents the global feature g i The transpose of Fifth, the attention weights are applied to the global features to form the weighted features G att : Sixth, the weighted features are mapped through the fully connected layer to output the probability of violation: <h2 style=";text-align:left;direction:ltr">y = softmax(W<h2 style=";text-align:left;direction:ltr"> fc <h2 style=";text-align:left;direction:ltr"> G<h2 style=";text-align:left;direction:ltr"> att <h2 style=";text-align:left;direction:ltr"> +b<h2 style=";text-align:left;direction:ltr"> fc <h2 style=";text-align:left;direction:ltr"> ) Among them, W fc and b fc They are all parameters of the fully connected layer, and softmax means normalization.

8. The whole process monitoring system for offshore project dismantling operation according to claim 7 is characterized in that: The risk analysis module specifically includes calculating the risk value R of each violation h h : R h =f(P h ,S,E) Among them, P h is the probability of violation h, S is the environmental factor, and E is the influencing factor of external events.

Images