Underwater construction visual processing system and processing method for bagged sand

Abstract

The application relates to the field of underwater construction and computer visualization, and discloses a bagged sand underwater construction visualization processing system and a processing method, which comprise a three-dimensional scene model storage for storing a three-dimensional scene model which can be dynamically updated, the three-dimensional scene model representing the physical boundary of an underwater construction area; a data acquisition module configured to acquire motion sensor data of bagged sand to be placed in real time, and to construct a dynamic state vector containing the motion state of the bagged sand based on the motion sensor data; and a parameter identification module connected to the data acquisition module. By arranging the parameter identification module, key physical parameters such as equivalent hydrodynamic damping coefficients and local environmental flow velocities can be identified online by using the moving process of the bagged sand in water before the bagged sand is released, so that a dynamic model of trajectory prediction can dynamically adapt to a real and changeable underwater environment, and errors caused by dependence on fixed or inaccurate empirical parameters can be avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater construction and computer visualization, in particular to a visualization processing system and method for underwater construction of bagged sand. Background Art

[0002] In water conservancy projects, marine engineering projects, and port and waterway construction, operations such as dam reinforcement, underwater pipeline protection, and caisson foundation laying often require the precise placement and deployment of a large number of components underwater. Bag sand is widely used due to its cost-effectiveness, adaptability, and ease of construction. Traditional underwater installation of bag sand relies primarily on lifting equipment (such as cranes or grab buckets) on surface construction platforms (such as ships or barges). The surface platform is positioned using methods such as the Global Positioning System, and the bag sand is then lowered to the desired location for release.

[0003] However, existing technologies suffer from inherent technical flaws. Underwater environments, especially turbid waters, have extremely low visibility, making the entire installation process a "blind operation." Operators on the surface cannot directly observe the actual movement of the bagged sand underwater, its trajectory, or its interaction with existing structures or terrain.

[0004] More importantly, after the bagged sand is released from the water's surface, its trajectory through the water isn't a simple vertical drop; instead, it's subject to complex and dynamically changing hydrodynamic forces. For one thing, the bagged sand's irregular shape and tumbling behavior during its fall make its hydrodynamic drag coefficient difficult to precisely predetermine. Furthermore, localized currents, whose strength and direction vary with depth and time, are prevalent in the construction area. These invisible undercurrents can significantly shift the bagged sand. The combined effects of these uncertainties often result in significant, unpredictable deviations between the bagged sand's final landing point and the theoretical target point estimated based on the water surface. Consequently, operators are largely forced to rely on personal experience through repeated trial and error, which not only leads to low construction efficiency and material waste, but also makes it difficult for the resulting underwater structure to meet precise design requirements.

[0005] In addition, the existing auxiliary installation method generally ignores the dynamic changes of the underwater environment during the construction process even if it uses the initial underwater topographic survey data. Each successful placement of a bagged sand becomes a new physical entity in the underwater scene, thereby changing the local topography. The placement and stability of subsequent bagged sands are directly affected by these previously placed components. The existing technical solutions generally lack an effective feedback and updating mechanism, which cannot dynamically and cumulatively reflect the completed construction results to the environment model on which it depends, leading to a gradual disconnection between its understanding of the environment and the actual construction status, making the subsequent installation guidance information more and more distorted, and unable to meet the needs of continuous, large-scale and high-precision construction. SUMMARY

[0006] In view of the deficiencies of the prior art, the present application provides a bagged sand underwater construction visualization processing system and method, which solves the problem of inaccurate prediction of the landing point and accurate placement of bagged sand components caused by uncertain environmental factors such as water power and dynamic changes of the construction scene in the underwater blind operation environment.

[0007] To achieve the above purpose, the present application is realized by the following technical scheme: a bagged sand underwater construction visualization processing system, comprising:

[0008] a three-dimensional scene model storage for storing a dynamically updateable three-dimensional scene model, the three-dimensional scene model representing the physical boundary of the underwater construction area;

[0009] a data acquisition module configured to acquire motion sensor data of a bagged sand to be placed in real time, and to construct a dynamic state vector containing the motion state of the bagged sand based on the motion sensor data;

[0010] a parameter identification module connected to the data acquisition module and configured to identify physical parameters representing the characteristics of the underwater environment online before the bagged sand is released based on the dynamic state vector;

[0011] a trajectory solving module connected to the parameter identification module, the data acquisition module and the three-dimensional scene model storage, and configured to:

[0012] solve the predicted trajectory of the bagged sand after it is released based on the physical parameters and the dynamic state vector; and

[0013] determine the predicted landing point of the bagged sand by detecting the collision between the predicted trajectory and the three-dimensional scene model stored in the three-dimensional scene model storage;

[0014] a visualization presentation module connected to the trajectory solving module and configured to display the predicted trajectory and the predicted landing point in an augmented reality in a visualization interface;

[0015] a scene updating module, connected with the three-dimensional scene model storage, and configured to update the three-dimensional scene model stored in the three-dimensional scene model storage according to the detection data of the placed bagged sand obtained by the detection device after the placement of the bagged sand is confirmed.

[0016] Preferably, the parameter identification module is specifically configured to:

[0017] The physical parameters are taken as state variables to be estimated by using an extended Kalman filter, and the physical parameters are optimally estimated on-line by using real-time motion information in the dynamic state vector.

[0018] Preferably, the physical parameters include:

[0019] Equivalent hydrodynamic damping coefficients or local environmental flow velocity vectors.

[0020] Preferably, the trajectory calculation module is specifically configured to:

[0021] The dynamic equation of the bagged sand in water is established, and the physical parameters are solved by using a numerical integration method to obtain the predicted trajectory.

[0022] Preferably, the predicted landing point is a probabilistic placement area on the surface of the three-dimensional scene model.

[0023] Preferably, the probabilistic placement area is generated based on a state covariance matrix output by an extended Kalman filter.

[0024] Preferably, the visual presentation module is specifically configured to:

[0025] The predicted trajectory is displayed in the form of a virtual trajectory line, and the predicted landing point or the corresponding probabilistic placement area is displayed in the form of a highlighted area or a heat map in the visual interface.

[0026] Preferably, the scene updating module is specifically configured to:

[0027] After the placement of the bagged sand is confirmed, the detection device is instructed to scan the newly placed bagged sand to obtain a three-dimensional entity model thereof; and,

[0028] The three-dimensional entity model is incorporated into the three-dimensional scene model by model fusion operation to form an updated three-dimensional scene model.

[0029] Preferably, the dynamic state vector includes at least one kind of information selected from the following group:

[0030] The position, velocity, acceleration, attitude and angular velocity of the bagged sand.

[0031] The application also provides a bagged sand underwater construction visualization processing method, comprising the following steps:

[0032] Step one, a dynamically updated three-dimensional scene model is established to represent the physical boundary of the underwater construction area;

[0033] Step two, real-time motion sensor data of the bagged sand to be placed is obtained, and a dynamic state vector containing the motion state of the bagged sand is constructed;

[0034] Step three, based on the dynamic state vector, the physical parameters representing the characteristics of the underwater environment are identified online before the bagged sand is released;

[0035] Step four, based on the physical parameters and the dynamic state vector, the predicted trajectory of the bagged sand after self-release is calculated, and the predicted landing point of the bagged sand is determined by detecting the collision between the predicted trajectory and the three-dimensional scene model;

[0036] Step five, the predicted trajectory and the predicted landing point are displayed in the visualization interface through augmented reality;

[0037] Step six, after the bagged sand is placed, the three-dimensional scene model is updated based on the detection data of the placed bagged sand obtained by the detection equipment, for subsequent installation operation.

[0038] The application provides a bagged sand underwater construction visualization processing system and method. The application has the following advantages:

[0039] 1. The application sets up a parameter identification module to identify key physical parameters such as equivalent hydrodynamic damping coefficient and local environmental flow velocity online during the movement of the bagged sand in water before the bagged sand is released, and applies these real-time identified parameters to subsequent trajectory calculation, so that the dynamic model of trajectory prediction can dynamically adapt to the real and variable underwater environment, avoiding errors caused by relying on fixed or inaccurate empirical parameters, and significantly improving the accuracy and reliability of the predicted trajectory and the predicted landing point.

[0040] 2. The application sets up a visualization presentation module to present the high-precision predicted trajectory and predicted landing point in the form of virtual trajectory lines and highlighted areas in the visualization interface superimposed with the three-dimensional scene model, converting the blind operation invisible underwater into a clear closed-loop control task based on visual feedback, so that the operator can accurately pre-align, effectively improving the positioning accuracy and operation efficiency of single placement.

[0041] 3、The application sets a scene updating module, after confirming the placement of the bagged sand each time, the detection data of the placed bagged sand obtained by the detection device is used to dynamically and cumulatively update the three-dimensional scene model, and a digital twin scene capable of evolving synchronously with the real underwater construction environment is constructed. This mechanism ensures that the subsequent trajectory prediction can fully consider the physical shielding and support effect of the placed components, enhances the adaptive ability of the system to complex and dynamic environments, and ensures the long-term prediction effectiveness in the continuous construction process. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 It is the main framework diagram of the application;

[0043] Figure 2 It is the motion sensor flowchart of the application;

[0044] Figure 3 It is the inertial measurement unit flowchart of the application;

[0045] Figure 4 It is the initialization unit flowchart of the application;

[0046] Figure 5 It is the parameter identification module flowchart of the application. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0048] Please refer to the attached Figure 1 -attached Figure 5 The underwater construction visualization processing system for bagged sand provided by the embodiments of the application sets a parameter identification module, identifies key physical parameters such as equivalent hydrodynamic damping coefficient and local environmental flow velocity online during the movement of bagged sand in water before the bagged sand is released, and applies these real-time identified parameters to subsequent trajectory calculation, so that the dynamic model of trajectory prediction can dynamically adapt to the real and variable underwater environment, avoiding errors caused by relying on fixed or inaccurate empirical parameters, and significantly improving the accuracy and reliability of predicted trajectory and predicted landing point.

[0049] The underwater construction visualization processing system for bagged sand comprises:

[0050] A three-dimensional scene model storage is used to store a three-dimensional scene model capable of dynamic update, and the three-dimensional scene model represents the physical boundary of the underwater construction area.

[0051] a data acquisition module configured to acquire motion sensor data of the bagged sand in real time, and construct a dynamic state vector containing motion state of the bagged sand based on the motion sensor data;

[0052] a parameter identification module connected to the data acquisition module, and configured to identify physical parameters representing characteristics of the underwater environment based on the dynamic state vector, before the bagged sand is released;

[0053] a trajectory calculation module connected to the parameter identification module, the data acquisition module and the three-dimensional scene model storage, and configured to:

[0054] calculate a predicted trajectory of the bagged sand after the bagged sand is released based on the physical parameters and the dynamic state vector; and

[0055] determine a predicted landing point of the bagged sand by detecting collision between the predicted trajectory and a three-dimensional scene model stored in the three-dimensional scene model storage;

[0056] a visualization presentation module connected to the trajectory calculation module, and configured to display the predicted trajectory and the predicted landing point in an augmented reality manner in a visualization interface;

[0057] a scene updating module connected to the three-dimensional scene model storage, and configured to update the three-dimensional scene model stored in the three-dimensional scene model storage based on detection data of the bagged sand obtained by a detection device after the bagged sand is installed.

[0058] First, before the entire underwater installation operation is started, the system needs to initialize the three-dimensional scene model storage to establish an initial three-dimensional scene model representing the original physical boundary.

[0059] Preferably, the initialization process is performed by an external detection device, such as a multi-beam sounding system or a three-dimensional scanning sonar, to conduct a comprehensive topographic survey of the predetermined construction water area, thereby obtaining high spatial resolution three-dimensional point cloud data of the initial riverbed or seabed.

[0060] The obtained original point cloud data will undergo a series of preprocessing operations, including but not limited to coordinate transformation, denoising filtering and outlier removal, and then be constructed by the system into a geometric model that is easy to process in computer graphics, preferably a triangular mesh model or a voxel grid model. This model is the initial three-dimensional scene model, denoted as , which is loaded and stored in the three-dimensional scene model storage as the baseline zero state for all subsequent simulation calculations and visualization presentations.

[0061] In addition, to realize the guiding operation, a CAD design model containing the engineering design profile or the theoretical placement target point is also loaded and accurately spatially registered with the initial three-dimensional scene model in a unified global coordinate system.

[0062] More critically, the three-dimensional scene model stored in the three-dimensional scene model storage is not static but dynamically evolves with the construction process through interaction with the scene updating module, which constitutes the technical basis for solving the problem of cumulative physical impact between components.

[0063] The dynamic updating process is triggered after a bagged sand placement operation is confirmed. The scene updating module instructs the detection device, preferably a forward-looking high-frequency imaging sonar installed on an underwater robot or a grab bucket, to perform a close-range, high-resolution fine scan of the bagged sand just placed on the bottom.

[0064] The purpose of this scan is to obtain the final, real, three-dimensional point cloud data of the newly placed component, including its actual deformation and posture due to water flow impact and interaction with the environment.

[0065] Subsequently, the system processes the point cloud data obtained from this scan and generates an independent, closed three-dimensional solid model that accurately describes the geometric shape of the single bagged sand, denoted as , where is the current installation number.

[0066] Next, the system performs a model fusion operation to merge the newly generated three-dimensional solid model with the three-dimensional scene model currently stored in the three-dimensional scene model storage. Preferably, the model fusion operation is a Boolean union operation of three-dimensional models, which can be mathematically expressed as: Through this operation, the solid model of the new component is seamlessly and permanently integrated into the scene model, forming an updated three-dimensional scene model that better reflects the real physical boundaries after completion, with version .This updated model will replace the old version and become the latest data in the storage to serve the next operation. This historical state accumulation mechanism makes the three-dimensional scene model storage play a role far beyond a simple data warehouse.

[0067] It becomes a self-evolving digital twin world. When performing subsequent installation operations, the trajectory calculation module will real-time retrieve the current latest three-dimensional scene model , as the physical boundary for collision detection. This ensures that the predicted trajectory of the subsequent bagged sand can physically interact with all previously placed components correctly, rather than "penetrating" them, thus significantly improving the accuracy of trajectory prediction in continuous construction. Meanwhile, the visualization module also retrieves this latest model from memory as the background scene for augmented reality overlay rendering, so that the operator sees a three-dimensional environment that is perfectly synchronized with the underwater construction progress.

[0068] In particular, the data acquisition module runs throughout the entire process from when the bagged sand is grabbed by the grab bucket, moves in the water, and is finally released. This process provides a crucial data basis for the subsequent online parameter identification of the invention.

[0069] To achieve a comprehensive perception of the motion state of the bagged sand, the data acquisition module is configured to obtain data from a set of preferred multi-source heterogeneous sensor systems.

[0070] Preferably, the sensor system includes a differential global positioning system (RTK / DGPS) installed on the surface work mother ship or platform to provide a high-precision global geographic position reference; an ultra-short baseline (USBL) acoustic positioning beacon installed on the underwater grab bucket or adjacent position to obtain the three-dimensional position of the grab bucket in real time; and an inertial measurement unit (IMU) integrated in the grab bucket, which has built-in accelerometers and gyroscopes to directly measure the linear acceleration and angular velocity of the grab bucket.

[0071] At the data acquisition level, the data acquisition module continuously performs the following operations:

[0072] It obtains the position information of the bagged sand from the ultra-short baseline system ;

[0073] It obtains three-axis linear acceleration information , three-axis angular velocity information , and attitude information usually expressed in quaternion form .

[0074] After obtaining the raw sensor data, the core task of the data acquisition module is to build and update the dynamic state vector in real time. This vector is a mathematical expression that describes the instantaneous physical profile of the bagged sand in the invention, which not only contains directly measured kinematic quantities, but also reserves data structure positions for unknown parameters to be identified.

[0075] The complete form of the dynamic state vector can be expressed as:

[0076] ;

[0077] In this vector, the origin and role of each component are as follows:

[0078] position vector with attitude quaternion , angular velocity vector and acceleration vector , are directly derived from the real-time measurements of the aforementioned sensors.

[0079] velocity vector , preferably derived by the data acquisition module through time-differentiation or derivation operation on the consecutive position vectors , reflects the instantaneous movement rate and direction of the bagged sand.

[0080] physical parameters, i.e. equivalent hydrodynamic damping coefficient and local environmental flow velocity vector , whose values are unknown in the initial stage of the vector construction. They are included in the vector structure as placeholders, whose real values will be estimated and filled in by the subsequent parameter identification module online. This structure design ensures the uniformity and integrity of the data stream.

[0081] effective mass , representing the mass corresponding to the apparent weight of the bagged sand in water, is usually included in the vector as a constant preset according to the specifications of the bagged sand for subsequent dynamic calculation.

[0082] Preferably, considering the possible differences in data update frequency of different sensors, the data acquisition module further includes a data synchronization unit. This unit is responsible for time alignment and synchronization processing of data streams from different sensors to ensure that each frame of dynamic state vector generated accurately reflects the physical state at the same time, providing a guarantee for the accuracy of subsequent algorithms.

[0083] Finally, the dynamic state vector constructed and output in real time by the data acquisition module will be the most direct and core data input, which will be passed to the parameter identification module and the trajectory solving module. The parameter identification module will use the time series of this vector to identify unknown physical parameters, while the trajectory solving module will use this vector at the moment of bagged sand release as the initial condition for its physical simulation.

[0084] In this embodiment, the parameter identification module is the core technical unit that realizes the adaptability and high-precision prediction capability of the bagged sand underwater construction visualization processing system of the present application. Its fundamental task is to solve the technical problem of unknown and dynamically changing key physical parameters in underwater environment, and to provide accurate model input for subsequent trajectory solving through online identification, which conforms to the current real environmental conditions.

[0085] In underwater environment, the key disturbance factor affecting the falling trajectory of bagged sand is fluid dynamics, the size and direction of which not only depends on the motion state of bagged sand itself, but also strongly relies on two physical parameters which are difficult to be set in advance: one is the equivalent hydrodynamic damping coefficient representing the interaction between bagged sand and water body; the other is the local environmental water flow velocity vector representing the motion of water body itself. These parameters show high nonlinearity and time-varying due to the irregular shape of bagged sand, the variable falling posture, and the existence of undercurrent and eddy current in water area. If fixed empirical values are used for trajectory prediction, large errors will be inevitably introduced.

[0086] To solve this problem, the parameter identification module starts working in the stage when bagged sand is moving in water with the lifting grab before it is officially released. It connects with the data acquisition module to continuously obtain the time series of dynamic state vector generated by the latter, and based on this information, it uses advanced filtering estimation algorithm for online parameter identification.

[0087] Preferably, in the present embodiment, the parameter identification module uses the extended Kalman filter (EKF) as the core algorithm to realize this function. This choice is because EKF is particularly suitable for handling state estimation problems of nonlinear systems. The implementation principle is as follows:

[0088] Firstly, the module takes the physical parameters to be identified, i.e. the equivalent hydrodynamic damping coefficient and the local environmental water flow velocity vector, as the state variables to be estimated, and together with the core kinematic state (position and velocity) of bagged sand, they form an augmented EKF state vector, which can be in the form of: By embedding the physical parameters in the state vector, the filter can use the dynamic response of the system to inversely deduce and estimate the values of these parameters in the iteration process.

[0089] Secondly, the module establishes the state transition equation that describes how the state vector evolves over time, i.e. the process model. The function here is a discrete-time representation based on Newton's law of motion. For the physical parameters and, a random walk model is preferred to describe their time-varying characteristics, i.e. they are assumed to be approximately constant over a short time but allowed to change slowly over time, which is consistent with physical reality. Thirdly, the module establishes the measurement equation that relates the state vector to the actual sensor measurements. In the present invention, the measurements are mainly the real-time position of the sandbag provided by the ultra-short baseline system, so the function's role is mainly to extract the position component from the state vector. With the above models, the parameter identification module runs in a continuous prediction-update cycle: in the prediction step, the module uses the state transition equation to predict the prior estimate of the state vector and its covariance at the current time based on the optimal estimate at the last time. In the update step, when the module receives the new sensor measurements from the data acquisition module, it calculates the residual between the measurements and the predicted measurements based on the prior estimate. According to this residual and combined with the system's uncertainty (characterized by the covariance matrix), the module calculates the Kalman gain and uses it to correct the prior estimate, thereby obtaining the posterior optimal estimate of the state at the current time that has fused the latest measurement information.

[0090] By continuously repeating this prediction-update cycle during the movement of the sandbag before release, the physical parameter components (and) in the state vector will start from unknown initial values (which can be set to a reasonable guess) and quickly converge to stable values that can best explain the observed behavior of the sandbag's movement.

[0091] Finally, when preparing to predict the trajectory, the parameter identification module provides the converged optimal estimated physical parameter values output by it to the trajectory solving module. In addition, the final state covariance matrix output by the module quantifies the degree of uncertainty in the estimates of all state variables (including the identified parameters), and this covariance matrix can also be used by the subsequent module to generate a probabilistic landing zone to provide the operator with an intuitive basis for judging the reliability of the prediction.

[0092] The operation of the trajectory solving module is usually triggered by the operator's instruction or automatically activated when the system determines that the sandbag has entered the appropriate release area. Once activated, the module will immediately obtain all the input information it needs to perform accurate calculation from other related modules.

[0093] Specifically, it obtains from the parameter identification module the physical parameters that characterize the current underwater environment, i.e. the equivalent hydrodynamic damping coefficient and the local environmental flow velocity vector . At the same time, it obtains from the data acquisition module the dynamic state vector of the sandbag at the instant of release and extracts the position vector as the initial condition for simulation and velocity vector . In addition, it retrieves the latest version of the 3D scene model from the 3D scene model storage , which has already contained all the previously placed components, as the geometric boundary for the subsequent collision detection. After obtaining all the above inputs, the trajectory solver module first establishes the dynamic equation describing the free-fall process of the bagged sand in water after it leaves the grab. This equation is based on Newton's second law, whose vector form can be expressed as: In this equation, is the mass of the bagged sand, is its acceleration vector.

[0094] is the net gravity of the bagged sand in water, i.e. the vector sum of its own gravity and the buoyancy force it experiences, which can be generally considered as a constant driving force.

[0095] is the core, nonlinear hydrodynamic drag term, which is the key to the prediction accuracy. In this embodiment, the calculation of this force makes full use of the online identification results of the present invention, and its expression is preferably:

[0096] ;

[0097] where, is the density of water, and A is the equivalent cross-sectional area of the bagged sand. It is crucial that the relative velocity vector here takes into account the influence of the local environmental flow velocity vector provided by the parameter identification module. Moreover, the equivalent hydrodynamic damping coefficient in the formula also directly adopts the latest value identified online by the parameter identification module. This approach enables the dynamic model to dynamically adapt to the real environment, rather than relying on fixed, possibly inaccurate empirical parameters. Since the above dynamic equation is a nonlinear ordinary differential equation, the trajectory solver module solves it using a high-precision numerical integration method. Preferably, the fourth-order Runge-Kutta method (RK4) is used, starting with the initial condition and sending it forward step by step at a small time step to calculate the position and velocity of the bagged sand at each future time. All these consecutive position points are connected in three-dimensional space to form the predicted trajectory of the bagged sand.

[0098] At each time step of numerical integration, the trajectory solver module must perform a collision detection. The module compares the latest calculated position of the bagged sand with its geometric envelope, and with the 3D scene model The intersection operation is performed. This process continues until the first collision is detected. This first collision point is the predicted landing point of the bagged sand determined by the trajectory solver module.

[0099] To provide the operator with quantitative information about the reliability of the prediction, the trajectory solver module in this embodiment can further generate a probabilistic placement zone (PPZ). This is achieved by utilizing the final state covariance matrix output by the extended Kalman filter of the parameter identification module. This matrix characterizes the uncertainty of the estimates of the physical parameters and the initial state. By means of error propagation theory or by performing a small number of Monte Carlo simulations (i.e. randomly sampling the input parameters within their uncertainty ranges and repeating the trajectory solving), the uncertainty at the input can be mapped to a two-dimensional probability distribution area of the predicted landing point on the surface of the three-dimensional scene model. This area visually demonstrates the most likely landing range of the bagged sand.

[0100] In particular, the visualization presentation module operates in real-time and continuously with the support of a three-dimensional graphics rendering engine. It serves as a terminal of information fusion and presentation, receiving data streams from multiple upstream modules and unifying them in a visual interface. First, it retrieves the latest version of the three-dimensional scene model from the three-dimensional scene model storage and renders it as the basic background of the three-dimensional virtual world. This provides the operator with a complete and accurate context of the underwater construction area.

[0101] Secondly, it connects to the data acquisition module to obtain the current position and attitude of the bagged sand in real time and renders a virtual model corresponding to it in the three-dimensional scene, enabling the operator to keep track of the real state of the bagged sand at all times. In addition, the pre-set engineering goals, such as the theoretical placement position or the design contour line, are also rendered as reference elements in the scene.

[0102] On this basis, the core function of the visualization presentation module is the augmented reality display of the prediction information. It receives the core output of the trajectory solver module, i.e. the predicted trajectory and the predicted landing point, and presents them using specific visual coding strategies.

[0103] For the predicted trajectory, it is preferred in this embodiment to render it as a clear virtual trajectory line extending from the virtual model position of the current bagged sand to the predicted landing point.

[0104] This trajectory line can be given unique visual properties, such as using a translucent material to avoid blocking important background information, or using a gradient color to represent the speed or time at different points on the trajectory. This dynamically updated trajectory line, like a "ghost guide", intuitively reveals to the operator the most likely falling path that the bagged sand will follow if it is released at this moment.

[0105] For the visualization of the predicted drop point, the specific form determines the way it is visualized.

[0106] If the predicted drop point is a certain 3D point, the module can render a prominent marker at that point, such as a 3D crosshair or a flashing indicator.

[0107] More preferably, when the predicted drop point is represented as a probabilistic placement zone (PPZ), the visualization rendering module renders it as a highlighted area or heat map on the surface of the 3D scene model. The extent of the area intuitively represents the uncertainty of the prediction, while the color or brightness distribution of the heat map can further represent the probability density - for example, the central area is warmer or brighter in color, representing the highest probability of landing.

[0108] By superimposing the real-time state information, prediction information and target information on the same screen, the application transforms the originally complex and operator experience-dependent blind underwater operation into a clear and visual feedback-based closed-loop control task.

[0109] The operator's task is no longer to guess, but to continuously adjust the pre-release position of the bagged sand by manipulating the lifting equipment, observing the changes in the virtual trajectory line and highlighted area in real time, and using them as a guide until the center of the rendered predicted drop point or probabilistic placement zone is accurately aligned with the pre-set engineering target placement point on the screen.

[0110] The operation of the scene update module is not continuous, but event-triggered. Specifically, when a bagged sand placement operation is completed and its state is confirmed by the operator or the automated system as stable and settled, the module is activated to perform a cumulative update of the scene.

[0111] After activation, the scene update module first performs a post-placement detection data acquisition process.

[0112] To this end, the module issues instructions to a specific detection device. Preferably, the detection device is a forward-looking high-frequency imaging sonar or a three-dimensional scanning sonar installed on an underwater robot or a lifting grab bucket, as it can provide high-resolution three-dimensional point cloud data at close range.

[0113] The goal of this scan is to accurately capture the final three-dimensional form of the bagged sand that has just been placed and settled, including its actual attitude due to collision with the riverbed or other components, as well as its deformation due to water flow impact or its own gravity. This step aims to obtain "as-built" real data, rather than relying on prediction models.

[0114] After obtaining the raw point cloud data containing information about the newly placed bagged sand, the scene update module further performs an individual model generation process.

[0115] The process includes a series of processing of the original point cloud data, such as filtering, denoising and surface reconstruction, and finally generates an independent, closed, and accurate three-dimensional solid model that describes the geometric shape of the single newly placed bag of sand. The model generated by the installation operation is recorded as The significance of this step is to separate the new information introduced by this installation operation from the complex environment and encapsulate it into an independent digital object.

[0116] Next, the scene update module performs its most core model fusion operation. This operation connects to the 3D scene model memory and adds the newly generated 3D solid model to the 3D scene model memory. , and the version currently stored in the memory is 3D scene model Merge. In 3D computer graphics, this model fusion operation is preferably implemented as a Boolean union operation of the 3D models. Its mathematical expression can be summarized as:

[0117] ;

[0118] The physical meaning of this formula is that the entity model of the new component is seamlessly and incrementally incorporated into the existing scene geometry, forming a version number of The updated 3D scene model The updated model fully incorporates the All physical components after the installation is completed. Finally, the scene update module will update the updated 3D scene model Write back to the 3D scene model memory to overwrite the old version.

[0119] Through this cycle of “operation-detection-fusion-update”, the present invention builds a digital twin environment that can evolve dynamically. Its key role is: in the next (i.e. During the installation operation, the trajectory solver module will perform collision detection based on this data which contains the actual historical installation results. This ensures that subsequent predictions correctly account for the physical occlusion and support effects of placed components. Furthermore, the visualization module renders based on this updated model, providing operators with a visual environment that is fully synchronized with the actual underwater construction progress. This ensures that the entire system's prediction and guidance capabilities remain accurate as construction progresses.

[0120] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. Visual processing system for underwater construction of bagged sand, characterized by: include: A three-dimensional scene model memory, used to store a dynamically updateable three-dimensional scene model, wherein the three-dimensional scene model represents the physical boundary of the underwater construction area; a data acquisition module configured to acquire in real time motion sensor data of the bagged sand to be placed, and construct a dynamic state vector containing the motion state of the bagged sand based on the motion sensor data; a parameter identification module, connected to the data acquisition module, and configured to identify physical parameters characterizing underwater environmental characteristics online based on the dynamic state vector before the bagged sand is released; The trajectory calculation module is connected to the parameter identification module, the data acquisition module and the three-dimensional scene model memory, and is configured as follows: Calculating a predicted trajectory of the bagged sand after self-release based on the physical parameters and the dynamic state vector; and Determining a predicted landing point of the bagged sand by detecting a collision between the predicted trajectory and the three-dimensional scene model stored in the three-dimensional scene model memory; a visualization presentation module, connected to the trajectory calculation module, configured to display the predicted trajectory and the predicted landing point in an augmented reality display in a visualization interface; a scene update module connected to the three-dimensional scene model memory and configured to: after the bagged sand is placed, update the three-dimensional scene model stored in the three-dimensional scene model memory according to the detection data of the placed bagged sand obtained by the detection device; The parameter identification module is specifically configured as follows: An extended Kalman filter is used to take the physical parameter as a state variable to be estimated, and real-time motion information in the dynamic state vector is used to perform online optimal estimation of the physical parameter; The scene update module is specifically configured as follows: After the bagged sand is confirmed to be placed, instructing the detection device to scan the newly placed bagged sand to obtain a three-dimensional solid model thereof; and Through a model fusion operation, the three-dimensional entity model is incorporated into the three-dimensional scene model to form an updated three-dimensional scene model.

2. The bagged sand underwater construction visualization processing system according to claim 1 is characterized in that: The physical parameters include: Equivalent hydrodynamic damping coefficient or local ambient water velocity vector.

3. The bagged sand underwater construction visualization processing system according to claim 1 is characterized in that: The trajectory solution module is specifically configured as follows: A dynamic equation of the bagged sand in water is established, and a numerical integration method is used to solve the equation in combination with the physical parameters to obtain the predicted trajectory.

4. The bagged sand underwater construction visualization processing system according to claim 1 is characterized in that: The predicted landing point is expressed as a probabilistic placement area on the surface of the three-dimensional scene model.

5. The bagged sand underwater construction visualization processing system according to claim 4 is characterized in that: The probabilistic placement region is generated based on a state covariance matrix output by an extended Kalman filter.

6. The bagged sand underwater construction visualization processing system according to claim 1 is characterized in that: The visualization presentation module is specifically configured as follows: The predicted trajectory is displayed in the form of a virtual trajectory line, and the predicted landing point or its corresponding probabilistic placement area is displayed in the form of a highlighted area or a heat map in a superimposed manner in the visualization interface.

7. The bagged sand underwater construction visualization processing system according to claim 1 is characterized in that: The dynamic state vector includes at least one information selected from the following group: The position, speed, acceleration, attitude and angular velocity of the bagged sand.

8. A visual processing method for underwater installation of bagged sand, according to the visual processing system for underwater construction of bagged sand according to any one of claims 1 to 7, characterized in that: The following steps are involved: Step 1: Establish a dynamically updateable 3D scene model to represent the physical boundaries of the underwater construction area; Step 2: acquiring motion sensor data of the bagged sand to be placed in real time, and constructing a dynamic state vector containing the motion state of the bagged sand; Step 3: Based on the dynamic state vector, before the bagged sand is released, online identification of physical parameters characterizing underwater environmental characteristics; Step 4: Based on the physical parameters and the dynamic state vector, a predicted trajectory of the bagged sand after self-release is calculated, and a predicted landing point of the bagged sand is determined by detecting a collision between the predicted trajectory and the three-dimensional scene model; Step 5: Display the predicted trajectory and the predicted landing point in an augmented reality display in a visualization interface; Step 6: After the bagged sand is placed, the three-dimensional scene model is updated based on the detection data of the bagged sand obtained by the detection equipment for subsequent installation operations.

Images