Control system and control method applied to reclaimed water buoy
By constructing an intelligent control system to monitor and diagnose the status of underwater floating bodies in real time, the problems of high operation and maintenance costs, low response efficiency, and insufficient monitoring and early warning for underwater floating bodies have been solved. This has enabled real-time monitoring and proactive control of underwater floating bodies, improving the operation and maintenance efficiency and safety of deep-sea equipment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-06
- Publication Date
- 2026-04-03
AI Technical Summary
Existing technologies for underwater floating bodies suffer from high maintenance costs, low response efficiency, and insufficient monitoring and early warning capabilities. Traditional underwater floating bodies lack the ability to actively regulate themselves in the deep-sea environment, and the problem of marine organism attachment is difficult to solve effectively.
An intelligent control system is constructed that includes data acquisition, model building, intelligent prediction, decision control, and dynamic optimization. It monitors multi-degree-of-freedom motion, marine organism attachment thickness, and structural mechanical response data in real time through a distributed sensor network. It uses a digital twin model for prediction and anomaly diagnosis, generates control commands, and drives the actuators to adjust the attitude and remove marine organisms.
It enables real-time monitoring and anomaly diagnosis of underwater floating bodies, reduces operation and maintenance costs, improves response efficiency, enhances monitoring and early warning capabilities, can proactively adjust attitude and effectively remove marine organisms, and improves the operation and maintenance efficiency and safety of deep-sea equipment.
Smart Images
Figure CN121785138A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of deep-sea resource development equipment technology, and in particular to a control system and control method for use with mid-water pontoons. Background Technology
[0002] In the process of deep-sea oil and gas development, risers, as a key support structure connecting surface platforms and underwater facilities, are directly related to the safety and reliability of the entire production system in terms of their operational stability.
[0003] Taking flexible riser systems as an example, the mid-water buoy serves as the support structure for plain wave type flexible risers. On the one hand, it provides underwater support to reduce tension and prevent pipe interference; on the other hand, its hydrodynamic characteristics significantly affect the dynamic response and structural integrity of the riser. However, traditional underwater buoys can only passively respond to wave and current environmental loads, lacking active control capabilities. Existing maintenance methods mainly rely on underwater robots for periodic inspections and attitude adjustments, which have the following prominent problems: (1) Poor economic benefits: Underwater robot operations require specialized vessels for support, and the cost of a single operation is high; (2) Low response efficiency: The time between the discovery of an anomaly and the completion of the handling is usually too long, which cannot meet the real-time requirements; (3) Insufficient monitoring capabilities: Traditional monitoring methods are insufficient to achieve all-weather, all-round monitoring of the floating body's operating status; (4) Lack of early warning mechanism: It is impossible to predict and warn of potential faults, and often the problem can only be dealt with passively after the accident occurs.
[0004] Furthermore, the problem of marine organism attachment in the deep-sea environment is becoming increasingly serious. The continuous growth of marine organisms on the surface of floating bodies increases the surface area and roughness of the structure, leading to a significant increase in drag force, which in turn causes attitude deviation and even instability of the floating body. Traditional manual removal methods are not only costly but also difficult to achieve timely and effective removal.
[0005] Therefore, there is an urgent need for an intelligent decision-making and control system that can realize real-time monitoring of the operating status of underwater floating bodies, intelligent diagnosis of abnormal operating conditions, and proactive generation of control commands, in order to overcome the limitations of existing technologies and improve the operation and maintenance efficiency and safety assurance level of deep-sea equipment. Summary of the Invention
[0006] The purpose of this invention is to provide a control system and method for use with medium-water pontoons, in order to solve the problems of high operation and maintenance costs, low response efficiency, and insufficient monitoring and early warning capabilities of existing underwater floating bodies, especially medium-water pontoons.
[0007] To achieve the above objectives, the technical solution provided by the present invention comprises two parts.
[0008] The first part is the control system applied to the reclaimed water pontoon, which includes a data acquisition module, a model building module, an intelligent prediction module, a decision control module, and a dynamic optimization module.
[0009] 1. Data acquisition module.
[0010] The data acquisition module is used to collect and preprocess the real-time operating status data of the buoy in the middle water. This module includes a data acquisition unit, a data preprocessing unit, and a data transmission unit.
[0011] The data acquisition unit collects real-time operational status data of the floating pontoon through a distributed sensor network. The operational status data includes three main categories: multi-degree-of-freedom motion data, marine organism attachment thickness data, and structural mechanical response data.
[0012] The multi-degree-of-freedom motion data mainly collects various motion states of the buoy in the middle of the water, specifically including three translational degrees of freedom (sway, roll, and heave) and three rotational degrees of freedom (roll, pitch, and yaw), for a total of six degrees of freedom motion data. The data on marine organism attachment thickness mainly obtains the thickness of marine organism attachment on key parts of the buoy, such as the skirt, riser support plate, and outer wall of the buoy. The acquisition methods can be ultrasonic sensors or image recognition technology. The ultrasonic sensor uses the pulse echo ranging principle, while the image recognition system automatically extracts the attachment layer thickness information from the image sequence obtained by the underwater camera. Structural mechanical response data mainly consists of pressure and strain data collected from key components.
[0013] The data preprocessing unit performs data cleaning based on a generative adversarial network (GAN), including missing value imputation, invalid value removal, and duplicate data merging. Specifically, missing value imputation is implemented using the adversarial learning mechanism of the SGAIN framework: the generator actively detects outlier data points using a neural network architecture and performs adaptive imputation, outputting a preliminary imputation matrix; the discriminator rigorously compares the distribution of generated data in non-missing regions with the original dataset; if the verification passes, the final data matrix is directly output; if it fails, a parameter optimization mechanism is triggered, dynamically adjusting the generator weights through gradient backpropagation, iterating until the imputation quality reaches a preset convergence criterion.
[0014] The data transmission unit transmits the cleaned data to the model building module.
[0015] 2. Model building module.
[0016] The model building module is used to construct a digital twin model of the buoy in operation based on operational status data. This module includes a data receiving unit and a digital twin model building unit.
[0017] The digital twin model building unit constructs a digital twin model with capabilities for physical entity mapping, multi-physics coupled simulation, and dynamic parameter calibration based on the following information: First, there are the three-dimensional geometric features, which include key parameters such as the diameter and height of the floating cylinder, the size of the support plate, and the spatial coordinates. Secondly, there are the mechanical properties of the materials, including key parameters such as the yield strength and buckling strength of the structure; Finally, there is historical operational data, which includes six-degree-of-freedom motion data, marine organism attachment thickness evolution data, and pressure and strain data for key components.
[0018] The digital twin model is driven by real-time received data to achieve real-time mapping between physical entities and virtual models.
[0019] 3. Intelligent prediction module.
[0020] The intelligent prediction module is used to predict and analyze the operating status of the reclaimed water pontoons based on a digital twin model, identify abnormal operating conditions, and generate early warning information. This module includes an algorithm prediction unit and an anomaly identification unit.
[0021] The algorithm prediction unit employs a parallel neural network architecture for state prediction. This parallel network includes a first branch (such as a Mamba network) for capturing long-term dependent features and a second branch (such as a Transformer network) for capturing local dynamic features. Through training with a large amount of historical measured data, the algorithm's prediction accuracy is improved. Ultimately, based on data from the current 20 time steps, it predicts the motion response, load distribution, and dynamic damage evolution trend for the next 20 time steps.
[0022] The anomaly identification unit identifies abnormal operating conditions by calculating multi-dimensional mutation coefficients. A dynamic weighting coefficient α is used to couple the current operating data with the prediction results, and a multi-scale convolutional autoencoder is used for deep feature extraction. The key mutation coefficients analyzed include motion mutation coefficient, marine organism thickness mutation coefficient, pressure mutation coefficient, and strain mutation coefficient, among which: The motion mutation coefficient characterizes the degree of abrupt change in motion for each degree of freedom. It is calculated by extracting the peak and trough values of motion using the rainflow counting method. The specific calculation formula and the meaning of each parameter are as follows: ( ) in This represents the peak or trough value of the current cycle. This represents the peak or trough value of the previous cycle of motion. The time interval between two peaks or troughs. It is the average of the peak or trough values of the exercise. These correspond to six degrees of freedom motion.
[0023] The marine organism thickness mutation coefficient characterizes the rate of change in the thickness of marine organism attachment. The specific calculation formula and the meaning of each parameter are as follows: in for The thickness of marine life in key areas of the target object at all times. for The thickness of marine life at any given time It is the product of the time step and the interval time step.
[0024] The pressure mutation coefficient characterizes the degree of pressure mutation in critical components. The specific calculation formula and the meaning of each parameter are as follows: in for Maximum pressure at key locations within a given time period. for Average pressure at key locations over a given time period.
[0025] The strain mutation coefficient characterizes the degree of strain mutation in key components. The specific calculation formula and the meaning of each parameter are as follows: in for Maximum strain at key components within a given time period. for Average strain of key components over time.
[0026] When the mutation coefficient exceeds the preset threshold, the anomaly identification unit determines it as an abnormal operating condition, realizing early diagnosis and fault root cause location of abnormal states such as structural overload and motion instability, and generating a graded early warning report.
[0027] 4. Decision control module.
[0028] The decision control module is used to generate control commands based on early warning information and drive the actuators to execute them. This module includes an intelligent decision-making unit, a command planning unit, and a command execution unit.
[0029] The intelligent decision-making unit makes decisions based on early warning reports, determining whether operations are needed and which key subsystems need to be activated. These key subsystems include: dynamic positioning system, marine organism removal system, and power generation system.
[0030] The dynamic positioning system includes multiple thrusters. Attitude control is achieved by adjusting the restoring force of the underwater buoy by regulating the thruster's operating state. Specifically, it addresses abnormal attitude deviations of the target object by setting control parameters based on the buoy's motion equations and planning the thruster's magnitude and direction. The buoy's motion equations and the meanings of each parameter are as follows: in, The mass matrix of the pontoons in medium water; The additional mass matrix of the middle-water pontoon; This is the damping coefficient matrix for the mid-water pontoon, including radiation damping and viscous damping; This is the restoring force coefficient matrix of the mid-water pontoon. ； The restoring force provided to the mooring system; The restoring force provided for the dynamic positioning system; , as well as These represent the acceleration, velocity, and displacement of the pontoon in the middle of the water, respectively. The excitation force and torque of the time-wave flow are determined by... The real part gives the complex amplitude of the wave current excitation force and torque components. These represent the six degrees of freedom of motion of the buoy in the middle of the water.
[0031] The marine organism removal system includes multiple linear vibrators with controllable vibration frequency, vibration time, and vibration interval. Vibration is transmitted to the structural surface via a vibration drive shaft to remove attached marine organisms. Specifically, the calculation formula for the drag force on the mid-water float and the meaning of each parameter are as follows: in, This is the drag coefficient of the mid-water pontoon; For fluid density; The projected area of the pontoon in the middle water is perpendicular to the direction of the incoming flow; The velocity is the flow rate. When the thickness of the marine organism attachment is large, it will increase the projected area. On the one hand, it increases surface roughness, leading to more obvious flow separation, which in turn increases the drag coefficient.
[0032] The power generation system supplies power to the dynamic positioning system and the marine organism removal system.
[0033] The command planning unit sets specific control commands based on the decision results, including the thrust magnitude and direction of the thruster, the vibration parameters of the vibrator, the start and stop of the power generation device, etc., and generates an operation instruction framework that includes priority ordering.
[0034] The command execution unit transforms decision instructions into device-level executable commands, formulates detailed control sequences including action timing, execution parameters, and safety thresholds, drives key equipment and systems to complete preset control actions, and synchronously collects equipment response data to verify the execution effect.
[0035] The dynamic positioning system's control is based on a floating body dynamics model. Since the underwater floating body is completely submerged, its hydrostatic stiffness is negligible, and the restoring force is mainly provided by the mooring system and the dynamic positioning system. By adjusting the thruster output of the dynamic positioning system, the restoring force coefficient of the floating body is changed, thus achieving active attitude control.
[0036] The control strategy of the marine organism removal system is determined based on the drag force experienced by the mid-water buoy. As the thickness of the marine organism deposits increases, it leads to a larger projected area facing the current and increased surface roughness, resulting in more pronounced flow separation and thus increasing the drag coefficient. When the deposit thickness reaches a threshold, the vibratory removal system is activated for active removal.
[0037] 5. Dynamic optimization module.
[0038] The dynamic optimization module is used to iteratively optimize the digital twin model and the decision control module based on the execution results. This module establishes a two-way optimization mechanism, including the following:
[0039] The first step is model parameter calibration. This part involves feeding back the real response data of the physical system to the model building module to dynamically calibrate the model parameters and continuously improve the model's prediction accuracy. Secondly, decision rule optimization involves iteratively optimizing the decision rules and control parameters in the decision control module based on multi-dimensional evaluation of execution results, thereby enhancing the overall control performance and adaptability of the system.
[0040] Through a two-way optimization mechanism, a two-way enhancement closed loop is formed from physical execution to virtual model, continuously improving the overall performance of the system.
[0041] The second part of this technical solution is an intelligent decision-making and control method for underwater floating bodies. This control method is based on the aforementioned control system and includes the following steps: S1: A digital twin model is constructed based on the three-dimensional geometric features, material mechanical properties, and historical operation datasets of the underwater floating body; multi-degree-of-freedom motion data of the structure, marine organism attachment thickness, and pressure and strain data of key parts are collected in real time through a distributed sensor network; the model is dynamically updated synchronously.
[0042] S2: Based on the real-time data output by the digital twin model and the historical operation database, a parallel neural network is used to make short-term predictions on the motion trajectory, structural damage evolution trend and load distribution characteristics of the underwater floating body, and generate a state evolution map of the underwater floating body.
[0043] S3: Achieve coupled analysis of current operating data and prediction results through dynamic weight coefficient α; use multi-scale convolutional autoencoder for deep feature extraction; calculate motion mutation coefficient, marine organism thickness mutation coefficient, pressure mutation coefficient and strain mutation coefficient; determine anomaly when the mutation coefficient exceeds the preset threshold; realize early diagnosis and fault root cause location of abnormal working conditions such as structural overload and motion instability; generate graded early warning reports.
[0044] S4: Make decisions based on early warning reports to determine whether operations are needed and which key subsystems need to be activated, including the dynamic positioning system, marine organism removal system, and power generation system.
[0045] S5: Set the specific commands for the devices or systems to be run, and generate an operation instruction framework that includes priority ordering.
[0046] S6: Transform decision instructions into device-level executable commands, and formulate detailed control sequences including action timing, execution parameters, and safety thresholds. Specific control parameters include: thrust magnitude and direction of the thruster, vibration time and frequency of the vibrator, and start and stop of the power generation unit.
[0047] S7: Drives key equipment and systems to complete preset control actions, and synchronously collects equipment response data to verify the execution effect.
[0048] S8: Feeds back the execution results of the device to the digital twin model building module to drive adaptive calibration of model parameters; at the same time, it optimizes the decision rule base based on the execution performance evaluation; forming a two-way enhancement mechanism from physical execution to virtual model, continuously improving the overall performance of the system.
[0049] Based on the above complete technical solution, compared with the prior art, the present invention has the following significant advantages: 1. By constructing a high-precision digital twin model and integrating multi-dimensional dynamic data such as multi-degree-of-freedom motion, pressure distribution, strain, and evolution of marine organism attachment thickness, a closed-loop management mechanism covering monitoring, diagnosis, decision-making, control, and optimization is established to reduce the operation and maintenance risks and manpower costs in the harsh deep-sea environment.
[0050] 2. By relying on the dynamic simulation of the hydrodynamic characteristics of underwater floating bodies using digital twins, attitude deviations caused by ocean current impacts and marine organism attachments are accurately captured. Through intelligent decision-making, attitude adjustments are automatically made, breaking through the limitation of traditional underwater floating bodies that can only passively respond to environmental loads.
[0051] 3. Based on in-depth analysis of real-time and historical data, potential faults can be predicted and warned, providing a scientific basis for preventive maintenance and comprehensively improving the efficiency of equipment life cycle management.
[0052] 4. By establishing a two-way optimization feedback mechanism, the system can continuously calibrate model parameters and optimize decision rules based on actual execution results, thereby achieving continuous improvement in control performance.
[0053] 5. By establishing a digital twin monitoring and operation platform for underwater floating bodies, a four-in-one technical paradigm of 'physical entity-virtual model-data analysis-intelligent control' will be formed, thereby strengthening innovation capabilities in the field of marine engineering equipment.
[0054] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0055] Figure 1 This is a flowchart of the intelligent decision-making control system for the buoy in the middle water of this invention; Figure 2 This is a technology roadmap for the intelligent decision-making control system for medium-water pontoons; Figure 3 This is a schematic diagram of the middle-water pontoon structure in this invention. Detailed Implementation
[0056] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.
[0057] Example 1 like Figures 1 to 3 As shown, this embodiment provides a control system for mid-water pontoons. This system is used for intelligent monitoring and autonomous control of mid-water pontoons in Plaint Wave type flexible riser systems during deep-sea oil and gas development. The control system includes a data acquisition module, a model building module, a prediction module, a decision control module, and a dynamic optimization module. These modules work together to form a closed-loop management mechanism covering monitoring, diagnosis, decision-making, control, and optimization.
[0058] In this embodiment, the pontoon is a cylindrical structure, mainly including the pontoon body, skirts, riser support plates and other components. Multiple skirts are evenly arranged around the outer periphery of the pontoon body to enhance hydrodynamic stability. The riser support plate is located on the upper part of the pontoon to support and constrain the flexible riser. Ballast chambers are provided inside the pontoon, and the buoyancy characteristics can be adjusted as needed.
[0059] The data acquisition module is used to collect and preprocess the operating status data of the floating pontoon in real time, and includes a data acquisition unit, a data preprocessing unit, and a data transmission unit.
[0060] The data acquisition unit collects real-time operational status data of the floating buoys through a distributed sensor network, specifically including the following three categories: Multi-degree-of-freedom motion data: The main data collected includes six degrees of freedom motion data of the buoy in mid-water, including three translational degrees of freedom (swell, roll, and heave) and three rotational degrees of freedom (roll, pitch, and yaw). Measurements can be taken using an inertial measurement unit (IMU) and a differential GPS combined navigation system, which is a mature technology solution for marine engineering.
[0061] Marine organism attachment thickness data: Ultrasonic thickness sensors and underwater camera systems are deployed at key locations such as the skirts, riser support plates, and outer walls of the mid-water pontoon. Measurements can be taken using ultrasonic sensors or image recognition. Ultrasonic sensors utilize the pulse-echo ranging principle, resulting in high measurement accuracy. The underwater camera system, combined with deep learning image recognition algorithms, can automatically identify and measure the thickness distribution of the marine organism attachment layer. In practical applications, a measurement cycle can be set, with a comprehensive scan performed after each measurement cycle.
[0062] (3) Structural mechanical response data: Pressure sensors and strain gauges are arranged at key structural nodes of the mid-water pontoon. The strain gauges adopt the form of triaxial strain rosettes, which can measure the strain state of key parts.
[0063] The data preprocessing unit employs an adversarial learning mechanism based on the SGAIN (Self-attention Generative Adversarial Imputation Networks) framework for data cleaning. Specific cleaning tasks include missing value imputation, invalid value removal, and duplicate data merging. The generator uses a self-attention neural network architecture to actively detect data gaps caused by sensor malfunctions or communication interruptions, and adaptively imputes them based on temporal and spatial correlations. The discriminator compares and verifies the generated imputed data with the original complete data, continuously optimizing the imputation quality through adversarial training until a preset convergence criterion is met.
[0064] The data transmission unit can adopt a hybrid transmission architecture combining underwater acoustic communication and satellite communication. Short-range data transmission uses an underwater acoustic modem, while long-range data backhaul uses Iridium or Inmarsat systems to ensure that data can be transmitted to the shore-based control center in real time. This part is also a mature communication method in the maritime field.
[0065] The model building module constructs a digital twin model of the buoy based on the operational status data, including a data receiving unit and a digital twin model building unit.
[0066] The digital twin model building unit integrates the following information to create a high-fidelity digital twin: (1) Three-dimensional geometric features: In this embodiment, based on the diameter, height and support plate size of the pontoon, a parametric modeling method is used to accurately describe the geometry of the pontoon and establish a finite element mesh model containing a large number of key nodes.
[0067] (2) Material mechanical properties: Based on the materials and specifications used in the pontoon body, support plate, etc., obtain parameters such as yield strength and elastic modulus, and accurately define the material property parameters of each component in the model.
[0068] (3) Historical operation data: The model continuously receives and stores the six-degree-of-freedom motion data of the floating pontoon, the evolution data of marine organism attachment thickness, and the pressure and strain data of key parts, and the corresponding storage period can be set.
[0069] The digital twin model employs a multiphysics coupled simulation framework, integrating fluid dynamics, structural mechanics, and control systems modules. The fluid dynamics module uses RANS equations to solve for the interaction between the floating body and the flow field; the structural mechanics module uses the finite element method to analyze the structural response; and the control systems module simulates the control logic of the thruster and vibrator. The model undergoes dynamic parameter calibration based on real-time acquired data to ensure a high degree of consistency between the virtual model and the physical entity.
[0070] The prediction module uses a digital twin model to predict and analyze the operating status of the pontoon in the middle water, identify abnormal operating conditions and generate early warning information. It includes an algorithm prediction unit and an anomaly identification unit.
[0071] The algorithm prediction unit employs a parallel neural network architecture for state prediction. This parallel network consists of two branches: the first branch uses a Mamba network architecture, leveraging a selective state-space model to capture long-term dependent features and effectively model the low-frequency trend changes in the motion of the buoy; the second branch uses a Transformer network architecture, employing a multi-head self-attention mechanism to capture local dynamic features and accurately predict the high-frequency response under wave-current excitation. The outputs of the two branches are weighted and fused through a feature fusion layer to generate the final prediction result.
[0072] In this embodiment, the prediction network predicts the motion response, load distribution, and dynamic damage evolution trend for the next 20 time steps based on input data over 20 time steps. Through training with a large amount of historical measured data, the root mean square error of the six-degree-of-freedom motion prediction can be maintained within a small range.
[0073] The anomaly detection unit identifies abnormal operating conditions by calculating multi-dimensional mutation coefficients. It employs dynamic weighting coefficients α to couple the analysis of current operating data and prediction results, and utilizes a multi-scale convolutional autoencoder for deep feature extraction. The key mutation coefficients analyzed include: (1) Motion mutation coefficient This characterizes the degree of abrupt change in motion for each degree of freedom, calculated by extracting the peak and trough values of the motion using the rainflow counting method. When the motion abrupt change coefficient of any degree of freedom exceeds a preset threshold, it is determined to be a motion anomaly.
[0074] Specifically, the calculation formula and the meaning of each parameter are as follows: ( ) in This represents the peak or trough value of the current cycle. This represents the peak or trough value of the previous cycle of motion. The time interval between two peaks or troughs. It is the average of the peak or trough values of the exercise. Corresponding to six degrees of freedom motion (2) Marine organism thickness mutation coefficient This characterizes the rate of change in the thickness of marine organism attachment. When the coefficient of change in marine organism thickness exceeds a preset threshold, it is considered an attachment anomaly.
[0075] Specifically, the calculation formula and the meaning of each parameter are as follows: in for The thickness of marine life in key areas of the target object at all times. for The thickness of marine life at any given time It is the product of the time step and the interval time step.
[0076] (3) Pressure mutation coefficient This characterizes the degree of pressure abrupt change in key components. When the pressure abrupt change coefficient exceeds a preset threshold, it is determined to be a pressure anomaly.
[0077] Specifically, the calculation formula and the meaning of each parameter are as follows: in for Maximum pressure at key locations within a given time period. for Average pressure at key locations over a given time period.
[0078] (4) Strain mutation coefficient This characterizes the degree of abrupt change in strain at key components. When the strain abrupt change coefficient exceeds a preset threshold, it is determined to be a strain anomaly.
[0079] Specifically, the calculation formula and the meaning of each parameter are as follows: in for Maximum strain at key components within a given time period. for Average strain of key components over time.
[0080] When any of the above mutation coefficients exceeds the preset threshold, the anomaly identification unit determines it as an abnormal operating condition, realizes early diagnosis and fault root cause location of abnormal states such as structural overload and motion instability, and determines four different levels of warning degree based on the number of mutation coefficients exceeding the threshold, and generates four different levels of warning reports.
[0081] The decision control module generates control commands based on early warning information and drives the actuators to execute them. It includes an intelligent decision-making unit, a command planning unit, and a command execution unit.
[0082] The intelligent decision-making unit makes decisions based on early warning reports, determining whether action is needed and which key subsystems need to be activated. These key subsystems include a dynamic positioning system, a marine organism removal system, and a power generation system.
[0083] For example, the dynamic positioning system includes multiple azimuth-rotating electric thrusters evenly distributed at the four corners of the bottom of the buoy. Attitude control is achieved by adjusting the restoring force coefficient of the buoy in mid-water by regulating the thruster output. The control strategy is based on the equation of motion of the floating body: in, The mass matrix of the pontoons in medium water; The additional mass matrix of the middle-water pontoon; This is the damping coefficient matrix for the mid-water pontoon, including radiation damping and viscous damping; This is the restoring force coefficient matrix of the mid-water pontoon. ； The restoring force provided to the mooring system; The restoring force provided for the dynamic positioning system; , as well as These represent the acceleration, velocity, and displacement of the pontoon in the middle of the water, respectively. The excitation force and torque of the time-wave flow are determined by... The real part gives the complex amplitude of the wave current excitation force and torque components. These represent the six degrees of freedom of motion of the buoy in the middle of the water.
[0084] The marine organism removal system comprises multiple linear vibrators distributed in key areas of the outer wall of the buoy. The frequency range of the vibrators is adjustable, and the vibration time and interval can be flexibly set according to the type and thickness of the marine organisms. High-frequency vibrations are transmitted to the structural surface via a vibration drive shaft, and the shear force generated by the vibrations causes the attached marine organisms to detach.
[0085] The marine organism removal system is activated based on changes in the drag force acting on the floating body. The drag force is calculated using the following formula: in, This is the drag coefficient of the mid-water pontoon; For fluid density; The projected area of the pontoon in the middle water is perpendicular to the direction of the incoming flow; The velocity is the flow rate. When the thickness of the marine organism attachment is large, it will increase the projected area. On the one hand, it increases surface roughness, leading to more obvious flow separation, which in turn increases the drag coefficient.
[0086] The power generation system can adopt a power supply scheme that combines ocean current energy generation with battery energy storage. The ocean current energy generation device can make full use of deep ocean current energy; the battery capacity can meet the continuous power supply needs under extreme operating conditions.
[0087] The command planning unit sets specific control commands based on the decision results, including thruster thrust magnitude and direction, vibrator vibration parameters, generator start / stop, etc., and generates an operation instruction framework with priority ordering. The command execution unit transforms the decision instructions into equipment-level executable commands, formulates detailed control sequences including action timing, execution parameters, and safety thresholds, drives key equipment and systems to complete preset control actions, and simultaneously collects equipment response data to verify the execution effect.
[0088] The dynamic optimization module is used to iteratively optimize the digital twin model and the decision control module based on the execution results. This module establishes a two-way optimization mechanism: (1) Model parameter calibration: The actual response data of the physical system is fed back to the model building module, and the model parameters, including the added mass coefficient, damping coefficient and restoring force coefficient, are dynamically calibrated using the Kalman filter algorithm to continuously improve the prediction accuracy of the model. In this embodiment, the model parameters are automatically calibrated every 1 hour.
[0089] (2) Decision rule optimization: Based on multi-dimensional evaluation of execution performance, reinforcement learning algorithms are used to iteratively optimize the decision rules and control parameters in the decision control module. Evaluation indicators include attitude recovery time, energy consumption, and execution success rate. The system continuously learns the optimal control strategy to enhance overall control performance and adaptability.
[0090] Through a two-way optimization mechanism, a two-way enhancement closed loop is formed from physical execution to virtual model, continuously improving the overall performance of the system.
[0091] Example 2 This embodiment provides a control method for a reclaimed water pontoon, which is based on the control system described in Embodiment 1 and includes the following steps: Step S1: Construct a digital twin model of the mid-water pontoon. A digital twin model is constructed based on the three-dimensional geometric features of the mid-water pontoon (diameter, height, support plate dimensions, etc.), material mechanical properties (yield strength, elastic modulus, etc.), and historical operational datasets. A distributed sensor network is used to collect real-time multi-degree-of-freedom motion data of the structure, marine organism attachment thickness, and pressure and strain data of key components, dynamically driving synchronous model updates and achieving real-time mapping between the physical entity and the virtual model.
[0092] Step S2: Based on the real-time data output by the digital twin model and the historical operation database, the Mamba-Transformer parallel neural network architecture is used to make short-term predictions (prediction time is 20 time steps) on the motion trajectory, structural damage evolution trend and load distribution characteristics of the pontoon in the middle water, and generate the state evolution map of the pontoon in the middle water.
[0093] Step S3: Couple the analysis of current running data and prediction results through dynamic weight coefficient α; use a multi-scale convolutional autoencoder to extract deep features; calculate the motion mutation coefficient, marine organism thickness mutation coefficient, pressure mutation coefficient and strain mutation coefficient; when the mutation coefficient exceeds the preset threshold, it is judged as abnormal; realize early diagnosis and fault root cause location of abnormal working conditions such as structural overload and motion instability; generate a graded early warning report.
[0094] Step S4: Make decisions based on the early warning report to determine whether operation is required and which key subsystems need to be activated. For abnormal motion, prioritize activating the dynamic positioning system for attitude adjustment; for abnormal marine organisms, activate the marine organism removal system for active removal; for structural overload, issue an alarm and adjust operating parameters to reduce the load.
[0095] Step S5: Set the specific commands for the equipment or system to be run, including the thrust magnitude and direction of the thruster (calculated according to the equation of motion of the floating body), the vibration time and frequency of the vibrator (determined according to the type and thickness of marine organisms), the start and stop of the power generation device, and generate an operation instruction framework containing priority order.
[0096] Step S6: Transform the decision instructions into device-level executable commands, and formulate a detailed control sequence that includes action timing, execution parameters, and safety thresholds. For example, the thruster control sequence of a dynamic positioning system includes steps such as start-up delay (0.5s), thrust climb rate (10kN / s), target thrust hold time, thrust descent rate, and shutdown.
[0097] Step S7: Drive key equipment and systems to complete preset control actions, and synchronously collect equipment response data to verify the execution effect. Verification includes: whether the actual output of the thruster reaches the target value, whether the buoy attitude returns to the set range, and whether marine organisms are effectively detached, etc.
[0098] Step S8: Feed back the device execution results to the digital twin model building module to drive adaptive calibration of model parameters; at the same time, optimize the decision rule base based on execution performance evaluation; form a two-way enhancement mechanism from physical execution to virtual model to continuously improve the overall system performance.
[0099] Example 3 This embodiment provides a typical control scenario application.
[0100] In the operating area of a deep-water oil and gas field in the South China Sea, a mid-water buoy is deployed at a depth of 1500 meters to support the flexible riser connecting the surface FPSO platform and the subsea production tree. One day, the system detected an anomaly: (1) The data acquisition module detected a sudden increase in the roll angle, from ±2° in normal working condition to ±5°, which exceeds the normal fluctuation range; at the same time, it detected that the thickness of marine organisms attached to the outer wall of the north side of the buoy reached 35mm, which exceeded the removal threshold.
[0101] (2) The prediction module calculated the roll motion mutation coefficient to be 2.3, exceeding the threshold of 2.0; the marine organism thickness mutation coefficient showed an abnormally rapid attachment rate in the past 3 days. Based on the current state, it is predicted that without intervention, the roll angle may further increase to ±8° in the next 2 hours, affecting the safety of the riser. The system generates an early warning report.
[0102] (3) The decision control module determines that the dynamic positioning system and the marine organism removal system need to be activated simultaneously. The command planning unit calculates that two thrusters need to be activated with output thrusts of 25kN and 20kN respectively, in the direction of 15° east of southeast; at the same time, four vibrators on the north side are activated with a vibration frequency of 100Hz for a duration of 30 minutes.
[0103] (4) The command execution unit drives the equipment to run according to the control sequence. After 30 minutes of execution, the system collects status data again: the roll angle has recovered to ±2.5°, and the thickness of marine organisms has decreased to 15mm. The execution effect has been verified.
[0104] (5) The dynamic optimization module records the current control process, updates the control effect evaluation database, calibrates the damping coefficient parameters in the digital twin model (damping changes caused by marine organisms), and optimizes the vibration parameter strategy for marine organism removal.
[0105] This embodiment demonstrates the complete workflow of the reclaimed water pontoon control system in a practical engineering application, showcasing the system's fully closed-loop intelligent control capabilities.
[0106] In summary, this invention, through the construction of five functional modules—data acquisition, model building, prediction, decision control, and dynamic optimization—achieves all-weather real-time monitoring of the operational status of the pontoon in deep water, intelligent diagnosis and early warning of abnormal operating conditions, autonomous generation of control commands, and continuous iterative optimization of system performance. Compared with traditional maintenance methods that rely on periodic inspections by underwater robots, this invention significantly reduces operation and maintenance costs, improves response efficiency, and enhances monitoring and early warning capabilities, providing an effective technical solution for the intelligent operation and maintenance of deep-sea resource development equipment.
[0107] It should be noted that the described embodiments are merely some, not all, of the embodiments of the present invention. Those skilled in the art will understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents; that is, all other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
Claims
1. A control system for a reclaimed water pontoon, characterized in that, include: The data acquisition module is used to collect and preprocess the real-time operating status data of the floating pontoons. A model building module, which constructs a digital twin model of the buoy based on the operational status data provided by the data acquisition module; The prediction module predicts and analyzes the operating status of the pontoon based on the digital twin model, identifies abnormal operating conditions, and generates early warning information. The decision control module generates control commands based on the early warning information and drives the actuator to execute them. The dynamic optimization module is used to iteratively optimize the digital twin model and the decision control module based on the execution results.
2. The control system according to claim 1, characterized in that, The data acquisition module collects operational status data including multi-degree-of-freedom motion data of the mid-water pontoon, marine organism attachment thickness data, and structural mechanical response data.
3. The control system according to claim 1, characterized in that, The preprocessing of the data acquisition module includes missing value imputation based on a generative adversarial network, which includes a generator for data imputation and a discriminator for verifying the quality of the imputation.
4. The control system according to claim 1, characterized in that, The prediction module uses a parallel neural network architecture for state prediction. The parallel neural network includes a first branch for capturing long-term dependent features and a second branch for capturing local dynamic features.
5. The control system according to claim 1, characterized in that, The prediction module identifies abnormal operating conditions by calculating multi-dimensional mutation coefficients, which include one or more of the following: motion mutation coefficient, marine organism thickness mutation coefficient, pressure mutation coefficient, and strain mutation coefficient.
6. The control system according to claim 1, characterized in that, The actuator includes a dynamic positioning system and / or a marine organism removal system. The dynamic positioning system achieves attitude control by adjusting the working state of the thruster to change the restoring force of the mid-water buoy.
7. The control system according to claim 1, characterized in that, The dynamic optimization module establishes a two-way optimization mechanism. On the one hand, it feeds back the execution results to the model building module to calibrate the model parameters. On the other hand, it optimizes the decision rules of the decision control module based on the evaluation of the execution effect.
8. A control method for a reclaimed water float, characterized in that, Includes the following steps: Construct a digital twin model of the pontoon in the middle water, and drive the model to be updated synchronously through real-time collected operational data; Based on the real-time and historical data output by the digital twin model, the state evolution of the pontoon in the middle water is predicted. The system performs coupled analysis of current operating data and prediction results to identify abnormal operating conditions and generate early warning information. Decisions are made based on the warning information, control commands are generated, and the actuators are driven to execute them. The execution results are fed back to the digital twin model and decision-making module for iterative optimization.
9. The control method according to claim 8, characterized in that, The identification of abnormal operating conditions includes: calculating a mutation coefficient that characterizes the degree of drastic change in operating status, and determining an abnormality when the mutation coefficient exceeds a preset threshold.
10. The control method according to claim 8, characterized in that, The drive actuator performs the following actions: calculating control parameters based on the dynamic model of the buoy, and adjusting the working state of the propeller of the dynamic positioning system and / or activating the vibrator of the marine organism removal system to achieve state regulation.