A method and system for predicting offshore foundation bearing capacity based on scour pit fractal reconstruction and deep learning, a terminal and a storage medium

Abstract

The application relates to the technical field of data prediction, and discloses a marine foundation bearing capacity prediction method and system based on scour pit fractal reconstruction and deep learning, a terminal and a storage medium.The method comprises the following steps: acquiring measured three-dimensional point cloud data, performing interpolation and reconstruction, and generating a three-dimensional scour pit digital geometric model; based on the three-dimensional scour pit digital geometric model, the foundation limit bearing capacity under different scour pit morphologies is calculated, and a scour pit morphology quantitative descriptor is extracted to form a data pair sample library; the deep learning model is trained by taking the data pair sample library as a training set, and a trained deep learning prediction model is obtained; the newly detected scour pit morphology data is input into the deep learning prediction model after being reconstructed and quantitatively described, and the foundation prediction bearing capacity under the current morphology is obtained. The application realizes end-to-end rapid prediction from detection data to bearing capacity evaluation, and provides an efficient and accurate decision support tool for the safe operation and maintenance of marine structures.

Description

Technical Field

[0001] This invention relates to the field of data prediction technology, and in particular to a method, system, terminal, and computer-readable storage medium for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning. Background Technology

[0002] The foundations of offshore wind power plants, offshore oil platforms, and other marine structures are subjected to the long-term effects of waves and ocean currents, which easily leads to the formation of complex and irregular seabed scour pits around them. The presence of scour pits alters the soil constraints surrounding the foundation, significantly weakening its bearing capacity and stability, and is one of the main causes of instability and failure of marine structures.

[0003] Existing technologies have the following significant drawbacks when assessing the impact of scour on foundation bearing capacity: (1) Simplification of scour pit morphology: Traditional analysis methods usually simplify the actual, irregular scour pit into a regular geometric shape, such as an inverted frustum or an inverted cone. This oversimplification ignores key geometric features of the scour pit, such as local steepness, undulation of the pit bottom, and asymmetry of the shape, resulting in a large deviation between the bearing capacity assessment results and the actual situation, which may seriously underestimate the risk.

[0004] (2) Insufficient application of detection data: Although techniques such as multibeam bathymetry can obtain discrete three-dimensional point cloud data of scour pits, how to transform these sparse and noisy data into a continuous and realistic geometric model of the scour pit that can be used for numerical analysis is a technical challenge. Existing methods cannot make full use of these valuable measured data.

[0005] (3) The contradiction between analysis efficiency and model construction: The bearing capacity analysis of the foundation after scour requires complex three-dimensional numerical simulation (such as finite element method). When it is necessary to analyze a large number of different scour pit morphologies, different soil parameters, and different load conditions, it is impractical to repeatedly perform modeling and time-consuming calculations. Therefore, it is difficult to establish a universal law between scour pit morphology and bearing capacity.

[0006] (4) Lack of rapid prediction tools: During the operation and maintenance phase of a structure, when a new scour pit morphology is detected, management or maintenance personnel urgently need a tool that can quickly assess the current bearing capacity and determine the risk level. However, traditional numerical analysis methods cannot meet this "real-time" or "near real-time" assessment requirement.

[0007] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0008] The main objective of this invention is to provide a method, system, terminal, and computer-readable storage medium for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning, aiming to solve the problems of inaccurate assessment and low efficiency in the prior art when evaluating the impact of irregular scour on foundation bearing capacity.

[0009] To achieve the above objectives, this invention provides a method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning. The method includes the following steps: Measured three-dimensional point cloud data of scour pits around offshore foundations are acquired, and the measured three-dimensional point cloud data are interpolated and reconstructed using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits. Based on the three-dimensional digital geometric model of the scour pit, the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated by a high-precision numerical simulation model, and the scour pit morphology quantification descriptor is extracted to form a data pair sample library consisting of the scour pit morphology quantification descriptor and the corresponding ultimate bearing capacity of the foundation. Using the data pair sample library as the training set, a deep learning model is trained using the training set, so that the deep learning model learns and establishes a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output foundation ultimate bearing capacity, thereby obtaining a trained deep learning prediction model. The newly detected scour pit morphology data is acquired, and after reconstruction and quantification, the scour pit morphology data is input into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology.

[0010] Optionally, the method for predicting the bearing capacity of marine foundations based on fractal reconstruction of scour pits and deep learning includes the following steps: acquiring measured three-dimensional point cloud data of scour pits surrounding the marine foundation, and interpolating and reconstructing the measured three-dimensional point cloud data using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits. Measured three-dimensional point cloud data of scour pits around offshore foundations are acquired, and the measured three-dimensional point cloud data are interpolated and reconstructed using a fractional Brownian motion model. The fractional Brownian motion model controls the irregularity of the reconstructed surface by adjusting the Hurst exponent. By adjusting the Hurst exponent of the fractional Brownian motion model, the generated fractal surface is optimally fitted to the measured three-dimensional point cloud data, thus obtaining a three-dimensional digital geometric model of the scour pit.

[0011] Optionally, the method for predicting the bearing capacity of marine foundations based on fractal reconstruction and deep learning of scour pits, wherein the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated by a high-precision numerical simulation model based on the three-dimensional digital geometric model of the scour pit, and scour pit morphology quantification descriptors are extracted to form a data pair sample library consisting of scour pit morphology quantification descriptors and corresponding ultimate bearing capacities of the foundation, specifically includes: Based on the three-dimensional digital geometric model of the scour pit and various virtual scour pit morphologies generated by transforming the model parameters of the fractional Brownian motion model, a variety of three-dimensional analysis models containing different geometric features of scour pits are established. The finite element-material point method was used to analyze various three-dimensional analysis models using a high-precision numerical simulation model. Loads were applied to each three-dimensional analysis model until failure, and the ultimate bearing capacity of the foundation after scour was calculated. Each set of scour pit morphology quantification descriptors and the calculated basic ultimate bearing capacity are stored as a pair of data in the database to form a data pair sample library.

[0012] Optionally, the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning includes, wherein the scour pit morphology quantification descriptor includes at least one of: maximum scour depth, scour range, scour pit volume, average slope, slope distribution, and fractal dimension.

[0013] Optionally, the method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning, wherein the step of using the data pair sample library as a training set to train a deep learning model, enabling the deep learning model to learn and establish a nonlinear mapping relationship between the input scour pit morphology quantification descriptor and the output foundation ultimate bearing capacity, to obtain a trained deep learning prediction model, specifically includes: Design and build a deep learning model; The data set is used as the training set, the scour pit morphology quantization descriptor is used as the input of the deep learning model, and the corresponding basic ultimate bearing capacity is used as the output of the deep learning model. The deep learning model is trained by optimizing the algorithm, so that the deep learning model learns and discovers the nonlinear mapping relationship between the morphological quantification descriptor of the scour pit and the ultimate bearing capacity of the foundation. Once training is complete, a deep learning prediction model is obtained.

[0014] Optionally, in the method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning, the deep learning model is a multilayer perceptron or a convolutional neural network.

[0015] Optionally, the method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning, wherein acquiring newly detected scour pit morphology data, reconstructing and quantifying the scour pit morphology data, and inputting it into the deep learning prediction model to obtain the predicted bearing capacity of the foundation under the current morphology, specifically includes: When a new scour pit morphology data is obtained through underwater exploration, the scour pit morphology data is reconstructed using fractals, and the current morphology quantization descriptor is extracted. The current morphology quantization descriptor is input into the deep learning prediction model, and the deep learning prediction model outputs the basic predicted carrying capacity under the current morphology.

[0016] Furthermore, to achieve the above objectives, the present invention also provides a marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning, wherein the marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning includes: The point cloud data processing module is used to acquire measured three-dimensional point cloud data of the scour pits around the offshore foundation, and to interpolate and reconstruct the measured three-dimensional point cloud data using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits. The data pair sample library formation module is used to calculate the ultimate bearing capacity of the foundation under different scour pit morphologies based on the three-dimensional scour pit digital geometric model through a high-precision numerical simulation model, and extract the scour pit morphology quantification descriptor to form a data pair sample library composed of the scour pit morphology quantification descriptor and the corresponding ultimate bearing capacity of the foundation. The deep learning model training module is used to train a deep learning model using the data pair sample library as the training set, so that the deep learning model learns and establishes a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output basic ultimate bearing capacity, thereby obtaining a trained deep learning prediction model. The bearing capacity prediction module is used to acquire newly detected scour pit morphology data, and after the scour pit morphology data is reconstructed and quantified, it is input into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology.

[0017] Furthermore, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning stored in the memory and executable on the processor. When the marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning is executed by the processor, it implements the steps of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning as described above.

[0018] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning, and when the marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning is executed by a processor, it implements the steps of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning as described above.

[0019] In this invention, measured 3D point cloud data of scour pits surrounding marine foundations are acquired, and a fractional Brownian motion model is used to interpolate and reconstruct the measured 3D point cloud data to generate a continuous 3D digital geometric model of the scour pits. Based on the 3D digital geometric model of the scour pits, the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated using a high-precision numerical simulation model, and scour pit morphology quantification descriptors are extracted to form a data pair sample library consisting of scour pit morphology quantification descriptors and corresponding ultimate bearing capacities of the foundation. Using the data pair sample library as a training set, a deep learning model is trained, enabling the deep learning model to learn and establish a nonlinear mapping relationship between the input scour pit morphology quantification descriptor and the output ultimate bearing capacity of the foundation, resulting in a trained deep learning prediction model. Newly detected scour pit morphology data is acquired, reconstructed, and quantified, and then input into the deep learning prediction model to obtain the predicted bearing capacity of the foundation under the current morphology. This invention achieves end-to-end rapid prediction from detection data to bearing capacity assessment, providing an efficient and accurate decision support tool for the safe operation and maintenance of marine structures. Attached Figure Description

[0020] Figure 1 This is a flowchart of a preferred embodiment of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning of the present invention; Figure 2 This is a schematic diagram illustrating the entire process of bearing capacity prediction in a preferred embodiment of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning of the present invention. Figure 3 This is a schematic diagram of fractional Brownian motion reconstructing the morphology of a scour pit in a preferred embodiment of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning of the present invention. Figure 4 This is a schematic diagram of the input and output of the deep learning prediction model in a preferred embodiment of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning of the present invention. Figure 5 This is a structural diagram of a preferred embodiment of the marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning of the present invention. Figure 6This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0022] The preferred embodiment of the present invention describes a method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning. Figure 1 and Figure 2 As shown, the method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning includes the following steps: Step S10: Obtain measured three-dimensional point cloud data of the scour pits around the offshore foundation, and use a fractional Brownian motion model to interpolate and reconstruct the measured three-dimensional point cloud data to generate a continuous three-dimensional digital geometric model of the scour pits.

[0023] Specifically, measured 3D point cloud data of scour pits surrounding marine foundations is acquired using a multibeam echo sounder. A fractional Brownian motion (FBM) model is then used to interpolate and reconstruct this data. The FBM model controls the irregularity of the reconstructed surface by adjusting the Hurst exponent; that is, the FBM model can describe the self-affineness and irregularity of natural surfaces through the Hurst exponent, making it very suitable for simulating fractal surfaces like seabed scour pits. For example, by setting a suitable Hurst exponent (e.g., H=0.8), the fractional Brownian motion algorithm is used to interpolate and refine the measured 3D point cloud data, generating a smooth 3D surface of the scour pit with natural random undulations. Fractional Brownian motion is a stochastic process with self-similar characteristics. The specific implementation process in scour pit reconstruction is typically as follows: Input: Discrete and sparse measured points (e.g., multibeam bathymetry data).

[0024] Process: Use measured points as "hard constraints" or benchmark skeletons; use random displacement method or successive random addition method to inject random fluctuations that conform to statistical characteristics between sparse points.

[0025] For example, if the measured point only shows a 5-meter-deep depression at the bottom of the pit, FBM will generate a fine, coarse texture on the slope, similar to real seabed sand waves, based on the Hearst exponent, instead of a smooth geometric surface, so that the numerical simulation is closer to the natural reality.

[0026] like Figure 3 The diagram shown illustrates how fractional Brownian motion reconstructs the morphology of scour pits. Figure 3 In the diagram, (a) represents the measured discrete point cloud data. Figure 3 (b) in the diagram represents the FBM surface interpolation generation process. Figure 3 (c) in the figure represents the reconstructed realistic 3D scour pit surface.

[0027] By adjusting the Hurst exponent of the fractional Brownian motion model, the generated fractal surface is optimally fitted to the measured three-dimensional point cloud data, thereby obtaining a continuous, realistic, and richly detailed three-dimensional scour pit digital geometric model (i.e., a smooth scour pit three-dimensional surface with natural random undulations).

[0028] This step aims to transform discrete, sparse measured point cloud data into a continuous, realistic three-dimensional digital geometric model of scour pits.

[0029] Step S20: Based on the three-dimensional scour pit digital geometric model, the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated by a high-precision numerical simulation model, and the scour pit morphology quantification descriptor is extracted to form a data pair sample library composed of the scour pit morphology quantification descriptor and the corresponding ultimate bearing capacity of the foundation.

[0030] Specifically, the morphological descriptor of the scour pit includes at least one of the following: maximum scour depth, scour range, scour pit volume, average slope, slope distribution, and fractal dimension. The purpose of extracting the fractal dimension is to quantify the spatial complexity of the surface, while H is a control parameter for the reconstruction process. Based on the three-dimensional digital geometric model of the scour pit and various virtual scour pit morphologies generated by transforming the model parameters of the fractional Brownian motion model, multiple three-dimensional analysis models containing different geometric features of the scour pit (including pile foundation geometry, reconstructed scour pit surface, and surrounding soil domain, etc.) are established. For example, by changing the FBM model parameters (changing the depth, range, and irregularity of the scour pit) and soil parameters (changing cohesion and friction angle), 200 different analysis models are generated.

[0031] Advanced numerical methods capable of accurately simulating large soil deformations and solid-liquid coupling effects are employed, such as using the Finite Element Method-Material Point Method (FEM-MPM) to analyze various three-dimensional analysis models through a high-precision numerical simulation model. Furthermore, existing monitoring data related to foundation bearing capacity can be used to calibrate the geotechnical parameters in the high-precision numerical simulation model.

[0032] Submarine soil exhibits strong nonlinearity, dilatation, and state dependence under stress. Simple linear elastic models cannot accurately simulate the yielding and failure process of soil under the "ultimate bearing capacity" state. In order to describe mechanical behavior, appropriate constitutive models (such as modified Cambridge models, boundary surface models, etc.) and parameters are assigned to the soil.

[0033] If monitoring data related to foundation bearing capacity (such as foundation tilt, strain, etc.) is available, it can be used to calibrate the geotechnical parameters of high-precision numerical simulation models to improve model fidelity. Among them, the monitoring data is "feedback" data, which usually comes from sensors (such as inclinometers, strain gauges, accelerometers) installed on the wind turbine tower or foundation, and reflects the dynamic response of the structure under actual wind and wave loads.

[0034] Apply loads to each of the three-dimensional analysis models until failure, and calculate the ultimate bearing capacity of the foundation after scour (vertical, horizontal, or overturning resistance). Store each set of scour pit morphology quantification descriptors (such as maximum scour depth, scour range, slope distribution, fractal dimension, etc.) and the calculated ultimate bearing capacity of the foundation as a pair of data in the database to form a data pair sample library (i.e., a high-fidelity numerical analysis sample library).

[0035] For example, by applying a horizontal load to each three-dimensional analysis model, the ultimate horizontal bearing capacity can be obtained. H ult For each 3D analysis model, the maximum scour depth of the scour pit is extracted. S max scour pit volume V Average slope α avg fractal dimension D f Features such as X , and the calculated H ult As output Y , forming data pairs ( X , Y ).

[0036] This step aims to obtain baseline data on "scour pit morphology-bearing capacity" through high-precision numerical simulation.

[0037] Step S30: Using the data pair sample library as the training set, train a deep learning model using the training set, so that the deep learning model learns and establishes a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output basic ultimate bearing capacity, and obtains a trained deep learning prediction model.

[0038] Specifically, a deep learning model (DNN, Deep Neural Network) is designed and constructed. The deep learning model is either a multilayer perceptron (MLP) or a convolutional neural network (CNN). The data pair sample library is used as the training set, the quantized descriptor of the scour pit morphology (or a two-dimensional depth map of the scour pit morphology) is used as the input of the deep learning model, and the corresponding ultimate bearing capacity of the foundation is used as the output of the deep learning model. The deep learning model is trained through an optimization algorithm so that the deep learning model learns and discovers the nonlinear mapping relationship between the quantized descriptor of the scour pit morphology and the ultimate bearing capacity of the foundation. After training is completed, a deep learning prediction model is obtained. That is, after training, a deep learning prediction model that can quickly predict the bearing capacity of the foundation in milliseconds based on the input scour pit morphology information is obtained.

[0039] For example, such as Figure 4 As shown, an input layer with 4 neurons is constructed (corresponding to...) S max , V , α avg , D f It contains 3 hidden layers, and the output layer has 1 neuron (corresponding to...). H ult The MLP network is used for training. Training: The network is trained using 200 data pairs until the loss function converges, resulting in the final deep learning prediction model. f DL .

[0040] Step S40: Obtain the morphological data of the newly detected scour pits, and after reconstructing and quantifying the scour pit morphological data, input it into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology.

[0041] Specifically, in practical engineering applications, when a new scour pit morphology data is obtained through underwater exploration, the scour pit morphology data is reconstructed using fractals to extract the current morphology quantization descriptor. The current morphology quantization descriptor is then input into the deep learning prediction model, which can instantly output the predicted bearing capacity of the foundation under the current morphology. Based on this, managers can conduct risk assessments and formulate maintenance or reinforcement plans.

[0042] For example, during a routine inspection, new scouring was discovered on the wind turbine foundation. Its shape was reconstructed, and its characteristics were extracted: X new = {S max =5m, V=300m³, ...} , X new This represents the quantization descriptor for the erosion pit morphology. S max Indicates the maximum scour depth of the scour pit. V Indicates the volume of the scour pit, X new Input deep learning prediction model f DL The model output is: H ult _ predicted = 8.5 MN , H ult _predicted This indicates the predicted ultimate bearing capacity of the foundation. H ult _predicted If the load-bearing capacity is lower than the design limit, the reinforcement plan should be activated immediately.

[0043] This invention discloses a method for predicting the bearing capacity of marine foundations based on fractal reconstruction of scour pits and deep learning, aiming to solve the problems of inaccurate and inefficient assessment of the bearing capacity of foundations after irregular scour. The method first innovatively employs a fractional Brownian motion (FBM) model to interpolate and reconstruct the measured sparse scour pit point cloud data, generating a realistic three-dimensional digital geometric model of the scour pit. Then, based on the reconstructed and virtual scour pit morphologies, a high-fidelity sample library containing the relationship between "scour pit morphology and ultimate bearing capacity" is established through high-precision finite element-material point method (FEM-MPM) numerical simulation. Finally, a surrogate model is trained using deep learning technology, which can learn and mine the nonlinear mapping relationship between the complex morphological features of scour pits and the bearing capacity of the foundation. This invention achieves rapid end-to-end prediction from detection data to bearing capacity assessment, providing an efficient and accurate decision support tool for the safe operation and maintenance of marine structures.

[0044] Furthermore, such as Figure 5 As shown, based on the above-mentioned method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning, this invention also provides a marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning, wherein the marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning includes: The point cloud data processing module 51 is used to acquire measured three-dimensional point cloud data of the scour pits around the offshore foundation, and to interpolate and reconstruct the measured three-dimensional point cloud data using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits. The data pair sample library forming module 52 is used to calculate the ultimate bearing capacity of the foundation under different scour pit morphologies based on the three-dimensional scour pit digital geometric model through a high-precision numerical simulation model, and extract the scour pit morphology quantification descriptor to form a data pair sample library composed of the scour pit morphology quantification descriptor and the corresponding ultimate bearing capacity of the foundation. The deep learning model training module 53 is used to train a deep learning model using the data pair sample library as the training set, so that the deep learning model learns and establishes a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output basic ultimate bearing capacity, and obtains a trained deep learning prediction model. The bearing capacity prediction module 54 is used to acquire newly detected scour pit morphology data, and after the scour pit morphology data is reconstructed and quantified, it is input into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology.

[0045] Furthermore, such as Figure 6 As shown, based on the above-mentioned method and system for predicting marine foundation bearing capacity based on scour pit fractal reconstruction and deep learning, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 6 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0046] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 may also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a marine foundation bearing capacity prediction program 40 based on scour pit fractal reconstruction and deep learning. This marine foundation bearing capacity prediction program 40 based on scour pit fractal reconstruction and deep learning can be executed by the processor 10, thereby realizing the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning in this application.

[0047] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning.

[0048] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface. The terminal's processor 10, memory 20, and display 30 communicate with each other via a system bus.

[0049] In one embodiment, when the processor 10 executes the marine foundation bearing capacity prediction program 40 based on scour pit fractal reconstruction and deep learning in the memory 20, the following steps are performed: Measured three-dimensional point cloud data of scour pits around offshore foundations are acquired, and the measured three-dimensional point cloud data are interpolated and reconstructed using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits. Based on the three-dimensional digital geometric model of the scour pit, the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated by a high-precision numerical simulation model, and the scour pit morphology quantification descriptor is extracted to form a data pair sample library consisting of the scour pit morphology quantification descriptor and the corresponding ultimate bearing capacity of the foundation. Using the data pair sample library as the training set, a deep learning model is trained using the training set, so that the deep learning model learns and establishes a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output foundation ultimate bearing capacity, thereby obtaining a trained deep learning prediction model. The newly detected scour pit morphology data is acquired, and after reconstruction and quantification, the scour pit morphology data is input into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology.

[0050] The process of acquiring measured 3D point cloud data of scour pits surrounding the offshore foundation, and then interpolating and reconstructing the measured 3D point cloud data using a fractional Brownian motion model to generate a continuous 3D digital geometric model of the scour pits, specifically includes: Measured three-dimensional point cloud data of scour pits around offshore foundations are acquired, and the measured three-dimensional point cloud data are interpolated and reconstructed using a fractional Brownian motion model. The fractional Brownian motion model controls the irregularity of the reconstructed surface by adjusting the Hurst exponent. By adjusting the Hurst exponent of the fractional Brownian motion model, the generated fractal surface is optimally fitted to the measured three-dimensional point cloud data, thus obtaining a three-dimensional digital geometric model of the scour pit.

[0051] Specifically, based on the three-dimensional digital geometric model of the scour pit, the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated through a high-precision numerical simulation model, and scour pit morphology quantification descriptors are extracted to form a data pair sample library consisting of scour pit morphology quantification descriptors and corresponding ultimate bearing capacities of the foundation, including: Based on the three-dimensional digital geometric model of the scour pit and various virtual scour pit morphologies generated by transforming the model parameters of the fractional Brownian motion model, a variety of three-dimensional analysis models containing different geometric features of scour pits are established. The finite element-material point method was used to analyze various three-dimensional analysis models using a high-precision numerical simulation model. Loads were applied to each three-dimensional analysis model until failure, and the ultimate bearing capacity of the foundation after scour was calculated. Each set of scour pit morphology quantification descriptors and the calculated basic ultimate bearing capacity are stored as a pair of data in the database to form a data pair sample library.

[0052] The morphological quantification descriptor of the scour pit includes at least one of the following: maximum scour depth, scour range, scour pit volume, average slope, slope distribution, and fractal dimension.

[0053] Specifically, the step of using the data pair sample library as a training set to train a deep learning model, enabling the deep learning model to learn and establish a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output foundation ultimate bearing capacity, thereby obtaining a trained deep learning prediction model, includes: Design and build a deep learning model; The data set is used as the training set, the scour pit morphology quantization descriptor is used as the input of the deep learning model, and the corresponding basic ultimate bearing capacity is used as the output of the deep learning model. The deep learning model is trained by optimizing the algorithm, so that the deep learning model learns and discovers the nonlinear mapping relationship between the morphological quantification descriptor of the scour pit and the ultimate bearing capacity of the foundation. Once training is complete, a deep learning prediction model is obtained.

[0054] The deep learning model is either a multilayer perceptron or a convolutional neural network.

[0055] Specifically, acquiring newly detected scour pit morphology data, reconstructing and quantifying the scour pit morphology data, and inputting it into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology includes: When a new scour pit morphology data is obtained through underwater exploration, the scour pit morphology data is reconstructed using fractals, and the current morphology quantization descriptor is extracted. The current morphology quantization descriptor is input into the deep learning prediction model, and the deep learning prediction model outputs the basic predicted carrying capacity under the current morphology.

[0056] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning, and the marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning, when executed by a processor, implements the steps of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning as described above.

[0057] In summary, this invention provides a method, system, terminal, and computer-readable storage medium for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning. The method includes: acquiring measured three-dimensional point cloud data of scour pits surrounding a marine foundation, and interpolating and reconstructing the measured three-dimensional point cloud data using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits; calculating the ultimate bearing capacity of the foundation under different scour pit morphologies using a high-precision numerical simulation model based on the three-dimensional digital geometric model of the scour pits, and extracting scour pit morphology quantification descriptors to form a data pair sample library composed of scour pit morphology quantification descriptors and corresponding ultimate bearing capacities of the foundations; using the data pair sample library as a training set, training a deep learning model using the training set, enabling the deep learning model to learn and establish a nonlinear mapping relationship between the input scour pit morphology quantification descriptors and the output ultimate bearing capacity of the foundations, thereby obtaining a trained deep learning prediction model; acquiring newly detected scour pit morphology data, and inputting the reconstructed and quantified scour pit morphology data into the deep learning prediction model to obtain the predicted bearing capacity of the foundation under the current morphology. This invention enables rapid end-to-end prediction from detection data to bearing capacity assessment, providing an efficient and accurate decision support tool for the safe operation and maintenance of marine structures.

[0058] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0059] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.

[0060] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning, characterized in that, The method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning includes: Measured three-dimensional point cloud data of scour pits around offshore foundations are acquired, and the measured three-dimensional point cloud data are interpolated and reconstructed using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits. Based on the three-dimensional digital geometric model of the scour pit, the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated by a high-precision numerical simulation model, and the scour pit morphology quantification descriptor is extracted to form a data pair sample library consisting of the scour pit morphology quantification descriptor and the corresponding ultimate bearing capacity of the foundation. Using the data pair sample library as the training set, a deep learning model is trained using the training set, so that the deep learning model learns and establishes a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output foundation ultimate bearing capacity, thereby obtaining a trained deep learning prediction model. The newly detected scour pit morphology data is acquired, and after reconstruction and quantification, the scour pit morphology data is input into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology. The process of acquiring measured 3D point cloud data of scour pits surrounding offshore foundations and then interpolating and reconstructing the measured 3D point cloud data using a fractional Brownian motion model to generate a continuous 3D digital geometric model of the scour pits specifically includes: Measured three-dimensional point cloud data of scour pits around offshore foundations are acquired, and the measured three-dimensional point cloud data are interpolated and reconstructed using a fractional Brownian motion model. The fractional Brownian motion model controls the irregularity of the reconstructed surface by adjusting the Hurst exponent. By adjusting the Hurst exponent of the fractional Brownian motion model, the generated fractal surface is optimally fitted to the measured three-dimensional point cloud data, thus obtaining a three-dimensional digital geometric model of the scour pit. Based on the three-dimensional digital geometric model of the scour pit, the ultimate bearing capacity of the foundation under different scour pit morphologies is calculated through a high-precision numerical simulation model. A quantification descriptor for the scour pit morphology is extracted, forming a data pair sample library consisting of the quantification descriptor for the scour pit morphology and the corresponding ultimate bearing capacity of the foundation. Specifically, this includes: Based on the three-dimensional digital geometric model of the scour pit and various virtual scour pit morphologies generated by transforming the model parameters of the fractional Brownian motion model, a variety of three-dimensional analysis models containing different geometric features of scour pits are established. The finite element-material point method was used to analyze various three-dimensional analysis models using a high-precision numerical simulation model. Loads were applied to each three-dimensional analysis model until failure, and the ultimate bearing capacity of the foundation after scour was calculated. Each set of scour pit morphology quantification descriptors and the calculated basic ultimate bearing capacity are stored as a pair of data in the database to form a data pair sample library. The process of acquiring newly detected scour pit morphology data, reconstructing and quantifying the scour pit morphology data, and inputting it into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology specifically includes: When a new scour pit morphology data is obtained through underwater exploration, the scour pit morphology data is reconstructed using fractals, and the current morphology quantization descriptor is extracted. The current morphology quantization descriptor is input into the deep learning prediction model, and the deep learning prediction model outputs the basic predicted carrying capacity under the current morphology.

2. The method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning according to claim 1, characterized in that, The morphological quantification descriptor of the scour pit includes at least one of the following: maximum scour depth, scour range, scour pit volume, average slope, slope distribution, and fractal dimension.

3. The method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning according to claim 1, characterized in that, The step of using the data pair sample library as a training set to train a deep learning model, enabling the deep learning model to learn and establish a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output foundation ultimate bearing capacity, thereby obtaining a trained deep learning prediction model, specifically includes: Design and build a deep learning model; The data set is used as the training set, the scour pit morphology quantization descriptor is used as the input of the deep learning model, and the corresponding basic ultimate bearing capacity is used as the output of the deep learning model. The deep learning model is trained by optimizing the algorithm, so that the deep learning model learns and discovers the nonlinear mapping relationship between the morphological quantification descriptor of the scour pit and the ultimate bearing capacity of the foundation. Once training is complete, a deep learning prediction model is obtained.

4. The method for predicting the bearing capacity of marine foundations based on scour pit fractal reconstruction and deep learning according to claim 1 or 3, characterized in that, The deep learning model is a multilayer perceptron or a convolutional neural network.

5. A marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning, characterized in that, The marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning is used to implement the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning as described in any one of claims 1-4. The marine foundation bearing capacity prediction system based on scour pit fractal reconstruction and deep learning includes: The point cloud data processing module is used to acquire measured three-dimensional point cloud data of the scour pits around the offshore foundation, and to interpolate and reconstruct the measured three-dimensional point cloud data using a fractional Brownian motion model to generate a continuous three-dimensional digital geometric model of the scour pits. The data pair sample library formation module is used to calculate the ultimate bearing capacity of the foundation under different scour pit morphologies based on the three-dimensional scour pit digital geometric model through a high-precision numerical simulation model, and extract the scour pit morphology quantification descriptor to form a data pair sample library composed of the scour pit morphology quantification descriptor and the corresponding ultimate bearing capacity of the foundation. The deep learning model training module is used to train a deep learning model using the data pair sample library as the training set, so that the deep learning model learns and establishes a nonlinear mapping relationship between the input scour pit morphology quantization descriptor and the output basic ultimate bearing capacity, thereby obtaining a trained deep learning prediction model. The bearing capacity prediction module is used to acquire newly detected scour pit morphology data, and after the scour pit morphology data is reconstructed and quantified, it is input into the deep learning prediction model to obtain the basic predicted bearing capacity under the current morphology.

6. A terminal, characterized in that, The terminal includes: a memory, a processor, and a marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning, stored in the memory and executable on the processor. When the marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning is executed by the processor, it implements the steps of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning as described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning. When the marine foundation bearing capacity prediction program based on scour pit fractal reconstruction and deep learning is executed by a processor, it implements the steps of the marine foundation bearing capacity prediction method based on scour pit fractal reconstruction and deep learning as described in any one of claims 1-4.

Images