Marine soil layering method and apparatus, computer readable storage medium, device
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NORTHWEST ENGINEERING CORPORATION LIMITED
- Filing Date
- 2025-09-11
- Publication Date
- 2026-07-21
AI Technical Summary
Existing technologies have low accuracy and efficiency in seabed soil stratification in marine engineering, making it difficult to meet the needs of high-precision and intelligent exploration. In particular, the reliability of soil identification is poor and the detection depth is insufficient in complex marine environments.
A multi-source feature fusion machine learning model is used to identify soil types by combining soil features and surface laser point cloud data. By using a preset mapping relationship between soil types and stratification methods, candidate stratification methods are determined and stratification processing is performed to generate marine soil stratification results.
It has achieved automation and improved engineering adaptability of marine soil stratification, reduced modeling bias caused by human intervention, and improved the accuracy, consistency and interpretability of stratification results, supporting the autonomous and efficient operation of unmanned systems in complex marine environments.
Smart Images
Figure CN121145039B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a method for marine soil stratification, a device for marine soil stratification, a computer-readable storage medium, and an electronic device. Background Technology
[0002] In the construction of offshore wind power and other marine engineering projects, accurate stratification of seabed soil and identification of the types of each soil layer are key technical aspects of geological exploration, basic design, construction plan formulation, and long-term operation and maintenance.
[0003] Accurate soil stratification and classification provide crucial geological information for determining design parameters, predicting pile driving resistance, assessing foundation bearing capacity, and analyzing long-term settlement and stability of wind turbine monopile or jacket foundations. This directly impacts the safety, economy, and durability of engineering projects. Therefore, developing high-precision and highly adaptable methods for seabed soil stratification and classification has become a core technological requirement in the field of marine geotechnical engineering.
[0004] Therefore, there is an urgent need in this field to develop a new method and apparatus for marine soil stratification.
[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure. Summary of the Invention
[0006] The purpose of this disclosure is to provide a method, apparatus, computer-readable storage medium, and electronic device for marine soil stratification, thereby overcoming, to at least some extent, the technical problems of low stratification accuracy and efficiency caused by limitations in related technologies.
[0007] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.
[0008] According to a first aspect of this disclosure, a method for marine soil stratification is provided, comprising:
[0009] Soil type identification is performed on the marine soil to be identified to obtain the target soil type of the marine soil to be identified;
[0010] Based on the preset mapping relationship between soil type and stratification method, determine the candidate stratification method corresponding to the target soil type;
[0011] The marine soil to be identified is stratified based on the candidate stratification method, the stratification depth of each soil layer interface is determined, and a marine soil stratification result containing soil layer sequence and corresponding depth information is generated.
[0012] In an exemplary embodiment of this disclosure, the step of identifying the soil type of the marine soil to be identified and obtaining the target soil type of the marine soil to be identified includes:
[0013] Extract the soil characteristics of the marine soil to be identified, and extract the point cloud features corresponding to the surface laser point cloud data of the marine soil to be identified;
[0014] The soil features and the point cloud features are input into the trained target machine learning model;
[0015] The target machine learning model is used to identify the soil type of the marine soil to be identified, thereby obtaining the target soil type of the marine soil to be identified.
[0016] In an exemplary embodiment of this disclosure, the target machine learning model is obtained in the following manner:
[0017] Key soil features are selected from a variety of marine soil characteristics; the feature importance score of the key soil features is higher than a preset score threshold;
[0018] Soil survey and measurement data that are compatible with the key soil characteristics are obtained as the first training data, and soil surface laser point cloud data that are spatiotemporally aligned with the survey points are collected as the second training data.
[0019] Based on the first and second training data, a target machine learning model for marine soil category identification is trained and generated.
[0020] In an exemplary embodiment of this disclosure, the marine soil characteristics include: physical properties, mechanical properties, and hydrological characteristics;
[0021] The physical characteristics include at least one of particle size distribution, density, and moisture content;
[0022] The mechanical properties include at least one of the following: tip resistance, sleeve resistance, and friction ratio in a static cone penetration test;
[0023] The hydrological characteristics include at least one of pore water pressure and pore water capacity.
[0024] In an exemplary embodiment of this disclosure, determining the candidate stratification method corresponding to the target soil type based on a preset mapping relationship between soil types and stratification methods includes:
[0025] Obtain a pre-constructed mapping table, which records each marine soil type and at least one corresponding adaptive stratification method;
[0026] The entry matching the target soil type is found in the mapping table, and the associated set of candidate stratification methods is extracted.
[0027] In an exemplary embodiment of this disclosure, the step of performing stratification processing on the marine soil to be identified based on the candidate stratification method to determine the stratification depth of each soil layer interface includes:
[0028] From the set of candidate stratification methods, the target stratification method is determined according to a preset selection strategy;
[0029] The target stratification method is used to stratify the marine soil to be identified, and the stratification depth of each soil layer interface is determined.
[0030] The preset selection strategy includes at least one of cost priority, accuracy priority, and efficiency priority.
[0031] In an exemplary embodiment of this disclosure, the step of performing stratification processing on the marine soil to be identified based on the candidate stratification method to determine the stratification depth of each soil layer interface includes:
[0032] The marine soil to be identified was independently stratified using multiple stratification methods from the candidate stratification method set, resulting in multiple preliminary stratification results;
[0033] Consistency analysis or weighted fusion of the multiple preliminary stratification results are performed to determine the final stratification depth of each soil layer interface.
[0034] According to a second aspect of this disclosure, a marine soil stratification device is provided, comprising:
[0035] The category identification module is used to identify the soil type of the marine soil to be identified, and to obtain the target soil type of the marine soil to be identified.
[0036] The stratification method determination module is used to determine the candidate stratification method corresponding to the target soil type based on the preset mapping relationship between soil type and stratification method;
[0037] The stratification result generation module is used to perform stratification processing on the marine soil to be identified based on the candidate stratification method, determine the stratification depth of each soil layer interface, and generate marine soil stratification results containing soil layer sequence and corresponding depth information.
[0038] According to a third aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the marine soil stratification method described in the first aspect above.
[0039] According to a fourth aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the marine soil stratification method described in the first aspect by executing the executable instructions.
[0040] As can be seen from the above technical solutions, the marine soil stratification method, marine soil stratification apparatus, computer-readable storage medium, and electronic device in the exemplary embodiments of this disclosure have at least the following advantages and positive effects:
[0041] In some embodiments of this disclosure, the technical solutions involve identifying the target soil type of the marine soil to be identified, determining the candidate stratification method corresponding to the target soil type based on a preset mapping relationship between soil type and stratification method, performing stratification processing on the marine soil to be identified based on the candidate stratification method, determining the stratification depth of each soil layer interface, and generating a marine soil stratification result containing soil layer sequence and corresponding depth information. On the one hand, this enables intelligent linkage from soil identification to stratigraphic modeling, improving the automation level and engineering adaptability of marine soil stratification; on the other hand, it reduces modeling bias caused by human intervention through the rule-based matching of soil characteristics and stratification strategies, improving the accuracy, consistency and interpretability of the results, and supporting the autonomous and efficient operation of unmanned systems in complex marine environments.
[0042] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description
[0043] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.
[0044] Figure 1 A flowchart illustrating the marine soil stratification method in an embodiment of this disclosure is shown.
[0045] Figure 2 This diagram illustrates the process of obtaining the target machine learning model as described in the embodiments of this disclosure.
[0046] Figure 3 This illustration shows a flowchart of how to screen key soil features from a variety of marine soil features in an embodiment of this disclosure;
[0047] Figure 4This illustration shows a flowchart of how to determine the candidate stratification method corresponding to the target soil type based on the preset mapping relationship between soil type and stratification method in an embodiment of this disclosure;
[0048] Figure 5 This illustration shows a flowchart of how to perform stratification processing on marine soil to be identified based on a candidate stratification method, and determine the stratification depth of each soil layer interface in an embodiment of the present disclosure.
[0049] Figure 6 This illustration shows another flowchart of how to perform stratification of marine soil to be identified based on a candidate stratification method in an embodiment of this disclosure, and determine the stratification depth of each soil layer interface.
[0050] Figure 7 This diagram illustrates the structure of a marine soil stratification device in an exemplary embodiment of this disclosure.
[0051] Figure 8 A schematic diagram of the structure of an electronic device in an exemplary embodiment of this disclosure is shown. Detailed Implementation
[0052] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this disclosure more comprehensive and complete, and to fully convey the concept of the example embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, apparatus, steps, etc., can be employed. In other instances, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of this disclosure.
[0053] The terms “a,” “an,” “the,” and “the” are used in this specification to indicate the presence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended inclusion and to mean that there may be other elements / components / etc. in addition to the listed elements / components / etc.; the terms “first” and “second” are used only as markings and are not a limitation on the number of objects.
[0054] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities.
[0055] In offshore wind power projects, the stratification of seabed soil and the determination of soil type for each layer are key foundational tasks in the exploration, design, construction, and operation and maintenance phases. These tasks directly relate to the selection of wind turbine foundations, bearing capacity assessment, settlement prediction, and long-term structural safety, providing important scientific basis for the safety and economy of the project.
[0056] In existing technologies, geophysical methods such as resistivity analysis or ground penetrating radar (GPR) are commonly used to classify seabed soil layers. Resistivity analysis measures the resistivity differences between different soil layers to create resistivity profiles and delineate interfaces. Ground penetrating radar, on the other hand, emits electromagnetic waves and utilizes the changes in dielectric constant caused by differences in water content in different soil types such as sand and clay. It then receives the reflected signals to generate stratigraphic profile images, thereby identifying the soil layer structure.
[0057] Meanwhile, some solutions combine machine learning methods to identify soil types within soil layers, attempting to achieve integrated analysis of "layering + classification." However, these methods have significant limitations:
[0058] The resistivity method is easily affected by factors such as groundwater salinity and temperature. Different soil types may have overlapping resistivity values, which leads to blurred stratigraphic boundaries and reduced accuracy.
[0059] Ground-penetrating radar (GPR) suffers severe electromagnetic wave attenuation in highly conductive soils (such as saline soil and saturated clay), limiting its detection depth and making it difficult to meet the deep exploration requirements of offshore wind power.
[0060] Existing machine learning methods for soil identification mostly rely on single-type features (such as using only mechanical parameters like penetration resistance from static cone penetration tests), failing to fully integrate soil physical characteristics (such as particle size distribution and water content), hydrological characteristics (such as permeability and void ratio), and microscopic morphology information. The single feature dimension results in weak generalization ability and low recognition accuracy of classification models.
[0061] In summary, existing technologies suffer from problems such as low stratification accuracy, insufficient detection depth, and poor reliability of soil identification in complex marine environments, making it difficult to meet the development needs of high-precision and intelligent marine engineering exploration.
[0062] In the embodiments of this disclosure, a method for stratifying marine soil is first provided, which at least to some extent overcomes the shortcomings of low stratification accuracy and efficiency in related technologies.
[0063] Figure 1 The diagram shows a flowchart of a marine soil stratification method in an embodiment of this disclosure. The subject executing the marine soil stratification method may be a server that stratifies marine soil.
[0064] refer to Figure 1 A marine soil stratification method according to an embodiment of the present disclosure includes the following steps:
[0065] Step S110: Identify the soil type of the marine soil to be identified to obtain the target soil type of the marine soil to be identified;
[0066] Step S120: Based on the preset mapping relationship between soil type and stratification method, determine the candidate stratification method corresponding to the target soil type;
[0067] Step S130: Based on the candidate stratification method, perform stratification processing on the marine soil to be identified, determine the stratification depth of each soil layer interface, and generate marine soil stratification results containing soil layer sequence and corresponding depth information.
[0068] exist Figure 1 The technical solution provided in the illustrated embodiment involves identifying the target soil type of the marine soil to be identified, determining the candidate stratification method corresponding to the target soil type based on a preset mapping relationship between soil types and stratification methods, and then performing stratification processing on the marine soil to be identified based on the candidate stratification methods to determine the stratification depth of each soil layer interface and generate a marine soil stratification result containing soil layer sequence and corresponding depth information. On the one hand, this enables intelligent linkage from soil identification to stratigraphic modeling, improving the automation level and engineering adaptability of marine soil stratification; on the other hand, it reduces modeling bias caused by human intervention through the rule-based matching of soil characteristics and stratification strategies, improving the accuracy, consistency and interpretability of the results, and supporting the autonomous and efficient operation of unmanned systems in complex marine environments.
[0069] The following are Figure 1 The specific implementation process of each step in the process will be explained in detail:
[0070] In step S110, the marine soil to be identified is subjected to soil type identification to obtain the target soil type of the marine soil to be identified.
[0071] In this step, the trained target machine learning can be used to identify the soil type of the marine soil to be identified, and obtain the target soil type of the marine soil to be identified.
[0072] refer to Figure 2 , Figure 2 This diagram illustrates the process of obtaining the target machine learning model as described in this embodiment, including steps S201-S203:
[0073] In step S201, key soil features are selected from a variety of marine soil features; the feature importance score of the key soil features is higher than the preset score threshold.
[0074] In this step, the aforementioned marine soil characteristics may include, but are not limited to: physical characteristics: used to characterize the soil's particle composition, compaction state, and water content, including at least one of particle size distribution (such as effective particle size, uniformity coefficient, curvature coefficient), compaction (such as void ratio, relative density), water content, and dry density; mechanical characteristics: reflecting the soil's mechanical response under external loads, preferably obtained through a cone penetration test (CPT), including at least one of tip resistance, sleeve resistance, and friction ratio; in other embodiments, in-situ or laboratory test parameters such as shear wave velocity and undrained shear strength may also be included; hydrological characteristics: describing the soil's water distribution, permeability, and pore pressure evolution characteristics, including at least one of pore water pressure, porosity, saturation, permeability coefficient, or pore water capacity (which can be defined as the free water content per unit volume of soil, related to water holding capacity).
[0075] refer to Figure 3 , Figure 3 This invention illustrates a flowchart of how to screen key soil features from a variety of marine soil characteristics in an embodiment of the present disclosure, including steps S301-S302:
[0076] In step S301, an unsupervised feature analysis mechanism is used to identify candidate soil features that have high distinguishability among different marine soils, forming a candidate feature set.
[0077] In this step, an unsupervised feature analysis mechanism can be used to identify candidate features with high discriminative power against different soil types from the original marine soil characteristics, and construct an initial candidate feature set. This mechanism does not rely on soil category labeling information, but rather evaluates the potential discriminative power of features in distinguishing different soil types based on their own statistical properties and multidimensional distribution structure, achieving preliminary dimensionality reduction and feature focusing.
[0078] Specifically, the unsupervised feature analysis includes one or more of the following methods:
[0079] Principal Component Analysis (PCA) maps a high-dimensional soil feature space to a low-dimensional principal component space through orthogonal transformation. The explanatory power of each principal component for overall data variation is determined based on its variance contribution rate. Further analysis of the loading values of the original features on high-contribution principal components reveals features with high loadings in the dominant components as candidates. For example, if "friction ratio" and "pore water pressure" have significant loadings on the first principal component, it indicates that they play a crucial role in characterizing the structural differences between different soil types.
[0080] Feature correlation analysis: Calculate the Pearson or Spearman rank correlation coefficients between all pairs of features to construct a correlation matrix and identify highly redundant feature pairs (e.g., correlation coefficients with an absolute value > 0.9). By removing redundant terms from strongly correlated features and retaining features that have weak correlations with other features but show significant differences in distribution under different geological conditions (e.g., "moisture content" shows significant differences between clay and sand), a subset of candidate features with high information density and low redundancy is formed.
[0081] Through the above mechanism, this disclosure can select a subset of features with significant distribution differences, low information redundancy, and large structural contribution from the initial multidimensional feature set, which can serve as the basis for subsequent supervised screening, effectively reducing computational complexity and improving feature screening efficiency.
[0082] In step S302, based on the candidate feature set and the pre-labeled marine soil category labels, a supervised feature selection mechanism is used to perform a second feature screening to determine the key soil features.
[0083] In this step, after obtaining the candidate feature set, class labels can be introduced as a supervision signal to quantitatively evaluate the classification and discrimination capabilities of each feature in the candidate feature set, thereby identifying and retaining key features that significantly contribute to the soil identification model construction task. Specifically, the above-mentioned supervised feature selection mechanism analyzes the effectiveness of each feature in distinguishing different marine soil types (such as sandy soil, silty soil, and clay soil) by constructing a first-level classification model or calculating the statistical correlation between features and categories. This process not only considers the distribution differences of feature values but also focuses on their actual role in the formation of decision boundaries, ensuring that the selected features have strong generalization ability and physical interpretability. Preferably, a cross-validation strategy can be used to verify the stability of the feature selection results, avoiding overfitting problems caused by sample bias or noisy data.
[0084] The supervised feature selection mechanism mentioned above may include, but is not limited to, the following methods: feature importance assessment based on tree models, recursive feature elimination, mutual information analysis, L1 regularization, or analysis of variance. For example, random forests or XGBoost (eXtreme Gradient Boosting Model) can be used to output the split gain or permutation importance score of each feature, and a preset importance threshold can be set to retain features with scores higher than the threshold; or a recursive feature elimination method can be used to remove the feature with the smallest contribution after each round of model training until the classification performance tends to stabilize or the target number of features is reached. In addition, features with high information gain and strong class correlation can be screened by calculating the mutual information value or chi-square statistic between features and class labels. This mechanism can effectively identify feature combinations with significant discriminative power in multiple soil types, such as the synergistic discriminative effect of "friction ratio" and "plasticity index" in distinguishing between sandy soil and clayey soil, thereby improving the accuracy of subsequent classification models.
[0085] Next, refer to Figure 2 In step S202, soil survey measured data that are compatible with key soil characteristics are acquired as the first training data, and soil surface laser point cloud data that are spatiotemporally aligned with the survey points are collected as the second training data.
[0086] In this step, marine geotechnical survey data containing physical and mechanical parameters such as particle size distribution, water content, liquid limit, plastic limit, and cone tip resistance can be obtained through geological database retrieval or field survey. Samples that are consistent with the key soil characteristics in terms of spatial location and stratigraphic depth are selected as the first training data.
[0087] Using a high-resolution laser scanning system mounted on an ROV (Remotely Operated Vehicle) or AUV (Autonomous Underwater Vehicle), three-dimensional laser point cloud data of the seabed sediment surface were collected at the same exploration point. This ensured that the data was spatiotemporally aligned with the exploration data in terms of spatial coordinates (positioning error ≤ 1m) and survey time sequence (same voyage or sediment stabilization period), serving as the second training data.
[0088] In step S203, a target machine learning model for marine soil category identification is trained and generated based on the first training data and the second training data.
[0089] In this step, a target machine learning model for marine soil identification can be trained based on the first and second training data mentioned above. For example, the structured soil features from the first training data can be first input into a fully connected neural network branch to extract high-level semantic features. Then, the laser point cloud data of the soil surface from the second training data can be preprocessed and converted into a depth map or directly input as a point cloud sequence into a convolutional neural network (CNN) or PointNet-like network branch to extract spatial geometric features. Afterward, the features extracted from the two branches are fused, using feature concatenation or attention fusion to achieve intermodal information interaction. The soil category probability distribution is then output through a classification head, and end-to-end training is performed using a cross-entropy loss function. During training, regularization strategies (such as Dropout and L2 regularization) and learning rate scheduling mechanisms are introduced to prevent overfitting and improve the model's generalization ability. Finally, the optimal model parameters are saved as the target machine learning model for marine soil category identification.
[0090] After obtaining the trained target machine learning model, the soil characteristics of the marine soil to be identified are first extracted. Specifically, key soil characteristics selected through feature importance analysis during the model training phase, such as particle size distribution, water content, liquid limit and plastic limit, and cone tip resistance, are core parameters that effectively reflect its physical and mechanical properties. Simultaneously, three-dimensional laser point cloud data of the marine soil surface to be identified is acquired. Through preprocessing operations such as denoising, registration, normal estimation, and geometric feature extraction, point cloud features characterizing micromorphology, such as surface roughness, curvature, normal vector distribution, fractal dimension, or depth map, are extracted. Subsequently, the soil characteristics and point cloud features are modally aligned and format normalized to form a structured multimodal feature vector, which is then fed into the trained target machine learning model. Based on the nonlinear mapping relationship between multi-source features and soil categories learned during the training phase, the model performs fusion modeling and classification inference on the input features, and finally outputs the soil category determination result of the marine soil to be identified, i.e., the target soil category, thereby achieving intelligent and high-precision identification of seabed sediment types.
[0091] Next, refer to Figure 1 In step S120, a candidate stratification method corresponding to the target soil type is determined based on the preset mapping relationship between soil type and stratification method.
[0092] In this step, the candidate stratification method corresponding to the target soil type can be determined based on the preset mapping relationship between soil type and stratification method.
[0093] In one exemplary implementation, reference Figure 4 , Figure 4 This embodiment of the present disclosure illustrates a flowchart of how to determine a candidate stratification method corresponding to a target soil type based on a preset mapping relationship between soil types and stratification methods, including steps S401-S402:
[0094] In step S401, a pre-constructed mapping table is obtained, which records each marine soil type and at least one corresponding adaptive stratification method.
[0095] In this step, the mapping table mentioned above can be retrieved from the local storage module or the cloud knowledge base. The mapping table is a structured rule base built based on marine geological sedimentary patterns, engineering survey specifications, and historical modeling experience. Each record item in the mapping table can include: marine soil type (such as "silty clay", "fine sand", "silty soil", "gravel layer", etc.); corresponding adaptive stratification method, including stratigraphic division grain size, modeling strategy, and constraints. The stratification method is configured according to the physical properties of the soil, sedimentary structure characteristics, and engineering application requirements to ensure that the selected method can accurately reflect the spatial distribution and mechanical behavior of this type of soil.
[0096] It should be noted that the above mapping table can be dynamically updated through expert experience input, historical project data mining, or machine learning cluster analysis to improve its adaptability and accuracy.
[0097] In step S402, the entry matching the target soil type is searched in the mapping table, and the set of candidate stratification methods associated with it is extracted.
[0098] In this step, the target soil category can be used as the search key to perform exact matching or fuzzy matching queries in the mapping table. If a completely matching entry exists, one or more associated adaptive stratification methods are extracted to form a candidate stratification method set. If no completely matching item is found, but there are semantically similar or physically similar soil categories (e.g., similarity calculated through ontology knowledge graphs or embedding vectors), their corresponding stratification methods are included as candidates in the set. The candidate stratification method set includes at least one partitioning strategy that can be used for subsequent stratigraphic modeling. Each strategy includes stratification granularity, modeling algorithm suggestions, and geological constraints. Furthermore, the stratification methods in the candidate set can be prioritized based on factors such as matching confidence, historical usage frequency, and engineering applicability score.
[0099] Next, refer to Figure 1In step S130, the marine soil to be identified is processed by layering based on the candidate layering method, the layering depth of each soil layer interface is determined, and a marine soil layering result containing soil layer sequence and corresponding depth information is generated.
[0100] In this step, the marine soil to be identified can be stratified based on candidate stratification methods to determine the stratification depth of each soil layer interface and generate marine soil stratification results containing soil layer sequences and corresponding depth information. Specifically, based on the candidate stratification methods determined in the previous steps, combined with field-acquired geotechnical survey data (such as CPTu (Cone Penetration Test with Pore Pressure Measurement) penetration resistance curves and sampling results) and auxiliary sensing data (such as seabed surface laser point clouds), the sedimentary structure of the target area can be modeled and analyzed; for the typical sedimentary characteristics corresponding to different soil types, appropriate algorithm logic is used to identify the location of soil layer interfaces. For example, in homogeneous clay, the stability of physical properties is used to determine the bottom interface; in sand-mud interlayers, resistance abrupt changes or texture variations are used to detect thin-layer boundaries. Furthermore, using the seabed surface as a reference plane, the vertical depth of each identified interface is calculated downwards to form an ordered sequence of layered depths. Finally, the information such as lithology, sequence number, top and bottom depths, thickness, and layering method used for each soil layer is structured and integrated to generate a complete marine soil stratification result. The output is in a standardized data format (such as GeoJSON or database table entries) for subsequent 3D geological modeling, engineering stability assessment, or unmanned platform decision support, achieving a closed-loop output from soil identification to stratigraphic structure analysis.
[0101] In one exemplary implementation, reference Figure 5 , Figure 5 This illustration shows a flowchart of how to perform stratification processing on marine soil to be identified based on a candidate stratification method, and determine the stratification depth of each soil layer interface, according to an embodiment of the present disclosure, including steps S501-S502:
[0102] In step S501, the target stratification method is determined from the candidate stratification method set according to a preset selection strategy.
[0103] In this step, the target stratification method can be determined from the set of candidate stratification methods according to a preset selection strategy.
[0104] The aforementioned preset selection strategy may include at least one of cost priority, accuracy priority, and efficiency priority, as detailed below:
[0105] Cost priority: Prioritize the stratification method that requires the fewest types of survey equipment, the shortest on-site operation time, and the lowest input of manpower and resources. This is suitable for budget-constrained or large-scale survey scenarios.
[0106] Precision Priority: Prioritize the layering method that can most accurately reflect the soil deposition structure, interface changes and physical and mechanical properties, which is suitable for fine modeling of key engineering areas (such as offshore wind power foundations and submarine pipeline routes).
[0107] Efficiency First: Prioritize hierarchical methods with low computational complexity, fast modeling speed, and support for real-time or near-real-time processing. This approach is suitable for online decision-making or emergency reconnaissance tasks on unmanned platforms (AUV / ROV).
[0108] The system can load corresponding strategies based on user configuration or project requirements, score and sort the items in the candidate stratification method set, and finally select the stratification method with the highest score as the target stratification method for subsequent stratigraphic modeling.
[0109] In step S502, the marine soil to be identified is layered using a target stratification method to determine the stratification depth of each soil layer interface.
[0110] In this step, the soil to be identified can be layered based on the aforementioned target stratification method to determine the stratification depth of each soil layer interface. For example, if the target stratification method is a single-layer model, the area to be identified can be regarded as a homogeneous soil body, and a uniform bottom interface depth can be output (such as based on the minimum bearing capacity requirement); if it is a multi-layer model, the abrupt change points of the cone tip resistance curve or shear wave can be analyzed, combined with the changes in point cloud surface morphology (such as scour surface, depositional discontinuity), to identify multiple soil layer interfaces and determine their stratification depth; if it is a variable thickness layer or lenticular body model, spatial interpolation algorithms (such as Kriging interpolation, inverse distance weighting) or machine learning regression models can be used to predict the interface depth distribution at each location.
[0111] For example, the above-mentioned stratification depth can be based on the seabed surface, with downward as the positive direction, and the unit is meters (m); the final structured stratigraphic division result can include: the number of each soil layer, lithological description, top and bottom depth, thickness and spatial distribution range, which can be used for subsequent engineering design or three-dimensional geological modeling.
[0112] In one exemplary implementation, reference Figure 6 , Figure 6 This illustration shows another embodiment of the present disclosure of how to perform stratification processing on the marine soil to be identified based on a candidate stratification method, and determine the stratification depth of each soil layer interface, including steps S601-S602:
[0113] In step S601, multiple stratification methods from the candidate stratification method set are used to independently stratify the marine soil to be identified, resulting in multiple preliminary stratification results.
[0114] In this step, each stratification method in the candidate stratification method set can be traversed and treated as an independent modeling path, and parallel stratigraphic division can be performed based on the same input data (including geotechnical investigation measured data, laser point cloud features, etc.).
[0115] Each stratification method generates corresponding preliminary stratification results based on its modeling logic. These preliminary stratification results may include: the predicted depth of each soil layer interface (e.g., first interface: 0.3m, second interface: 0.7m); the inference of the number and lithology of soil layers; and the stratification confidence score (derived from the model's internal uncertainty assessment). The various stratification methods mentioned above may include, but are not limited to: single-layer models, multi-layer interpolation models, variable-thickness layer models, and machine learning-based sequence segmentation models. All preliminary stratification results are stored in a structured format to form a preliminary stratification result set for subsequent fusion analysis.
[0116] In step S602, consistency analysis or weighted fusion is performed on multiple preliminary stratification results to determine the final stratification depth of each soil layer interface.
[0117] In this step, at least one of the following fusion strategies can be used to integrate the preliminary hierarchical result set:
[0118] Consistency analysis: By spatial alignment and depth matching, soil interfaces in multiple preliminary stratification results are compared to determine whether they point to the same stratigraphic location. A reasonable depth tolerance range (e.g., ±0.2m) can be set. When two or more stratification methods identify soil interfaces with the same sequence number (such as the first or second interface) within this range, the interface is considered to have high confidence, is retained as a consistent result, and marked as a high-confidence output. For interfaces identified by only a single method and lacking support from other methods, they are marked as "interfaces to be verified," and their source method and the strength of support from the corresponding data are recorded to form auxiliary decision-making information for subsequent manual review or supplementary exploration.
[0119] Weighted Fusion: A dynamic weighting mechanism is introduced, assigning corresponding weights based on the credibility of each stratification method to more rationally fuse different results. Credibility considers multiple factors, including the method's accuracy in historical projects, the quality of the current input data (e.g., CPTu signal integrity, laser point cloud density), and the confidence level of the model's output. For each soil interface, the depth values predicted by each method are weighted and averaged to calculate the final stratification depth. This approach effectively reduces the impact of low-quality or unstable methods on the results, improving the accuracy and stability of the overall stratification results.
[0120] Based on the above technical solutions, this disclosure has at least the following technical effects:
[0121] First, it improves the accuracy of marine soil classification. By integrating historical geotechnical survey data (structured engineering parameters) with three-dimensional laser point cloud data of soil surface (microscopic morphological features), a multimodal input machine learning model is constructed. This fully utilizes the complementarity of the two types of data, effectively overcoming the shortcomings of single data sources being susceptible to noise interference and incomplete information, and significantly improving the accuracy of soil classification in complex marine environments.
[0122] Secondly, it achieves an intelligent closed loop from soil identification to stratigraphic modeling. This solution not only completes soil type identification, but also automatically recommends suitable stratigraphic division strategies through a preset "soil type - stratification method" mapping relationship, realizing full-process automation from "perception-identification-decision-modeling" and promoting the development of marine geotechnical exploration towards intelligence and unmanned operation.
[0123] Third, it enhances the adaptability and engineering practicality of stratigraphic division. A target stratification method determination mechanism based on multi-dimensional selection strategies such as cost priority, accuracy priority, and efficiency priority is introduced. This allows for flexible configuration of the optimal modeling path according to different engineering scenarios (such as general surveys, detailed explorations, and emergency responses), balancing economy, reliability, and timeliness to meet diverse engineering needs.
[0124] Fourth, it improves the reliability and anti-interference ability of stratigraphic interface identification. By adopting a multi-layered parallel computing and fusion analysis mechanism, and by performing consistency checks or weighted fusion on multiple preliminary stratification results, it effectively suppresses the bias and abnormal output of a single model, and improves the stability and reliability of the final stratification depth. It is especially suitable for marine environments with complex sedimentary structures and fluctuating data quality.
[0125] Fifth, it supports collaborative modeling driven by knowledge and data. By constructing a mapping table of "soil type - stratification method", expert experience and industry standards are transformed into a computable knowledge base, realizing the digital expression of geological cognition; at the same time, combined with the data-driven capabilities of machine learning models, a dual-driven intelligent modeling system of "knowledge + data" is formed, which has both interpretability and adaptability.
[0126] Sixth, it provides a deployable technical path for unmanned exploration platforms. The method can be integrated into unmanned systems such as AUVs and ROVs, supporting the completion of seabed soil identification and stratigraphic division without real-time human intervention. It is suitable for autonomous exploration tasks in high-risk areas such as deep water and open sea, and has good engineering application prospects and industrialization value.
[0127] This disclosure also provides a marine soil stratification device. Figure 7This diagram illustrates the structure of a marine soil stratification device in an exemplary embodiment of this disclosure; as shown... Figure 7 As shown, the marine soil stratification device 700 may include a category identification module 710, a stratification method determination module 720, and a stratification result generation module 730. Wherein:
[0128] The category identification module 710 is used to identify the soil type of the marine soil to be identified and obtain the target soil type of the marine soil to be identified.
[0129] The stratification method determination module 720 is used to determine the candidate stratification method corresponding to the target soil type based on the preset mapping relationship between soil type and stratification method;
[0130] The stratification result generation module 730 is used to perform stratification processing on the marine soil to be identified based on the candidate stratification method, determine the stratification depth of each soil layer interface, and generate marine soil stratification results containing soil layer sequence and corresponding depth information.
[0131] In an exemplary embodiment of this disclosure, the category identification module 710 performs soil category identification on the marine soil to be identified to obtain the target soil category of the marine soil to be identified, including:
[0132] Extract the soil characteristics of the marine soil to be identified, and extract the point cloud features corresponding to the surface laser point cloud data of the marine soil to be identified;
[0133] The soil features and the point cloud features are input into the trained target machine learning model;
[0134] The target machine learning model is used to identify the soil type of the marine soil to be identified, thereby obtaining the target soil type of the marine soil to be identified.
[0135] In an exemplary embodiment of this disclosure, the target machine learning model is obtained in the following manner:
[0136] Key soil features are selected from a variety of marine soil characteristics; the feature importance score of the key soil features is higher than a preset score threshold;
[0137] Soil survey and measurement data that are compatible with the key soil characteristics are obtained as the first training data, and soil surface laser point cloud data that are spatiotemporally aligned with the survey points are collected as the second training data.
[0138] Based on the first and second training data, a target machine learning model for marine soil category identification is trained and generated.
[0139] In an exemplary embodiment of this disclosure, the marine soil characteristics include: physical properties, mechanical properties, and hydrological characteristics;
[0140] The physical characteristics include at least one of particle size distribution, density, and moisture content;
[0141] The mechanical properties include at least one of the following: tip resistance, sleeve resistance, and friction ratio in a static cone penetration test;
[0142] The hydrological characteristics include at least one of pore water pressure and pore water capacity.
[0143] In an exemplary embodiment of this disclosure, the stratification method determination module 720 determines a candidate stratification method corresponding to the target soil type based on a preset mapping relationship between soil types and stratification methods, including:
[0144] Obtain a pre-constructed mapping table, which records each marine soil type and at least one corresponding adaptive stratification method;
[0145] The entry matching the target soil type is found in the mapping table, and the associated set of candidate stratification methods is extracted.
[0146] In an exemplary embodiment of this disclosure, the stratification result generation module 730 performs stratification processing on the marine soil to be identified based on the candidate stratification method, and determines the stratification depth of each soil layer interface, including:
[0147] From the set of candidate stratification methods, the target stratification method is determined according to a preset selection strategy;
[0148] The target stratification method is used to stratify the marine soil to be identified, and the stratification depth of each soil layer interface is determined.
[0149] The preset selection strategy includes at least one of cost priority, accuracy priority, and efficiency priority.
[0150] In an exemplary embodiment of this disclosure, the stratification result generation module 730 performs stratification processing on the marine soil to be identified based on the candidate stratification method, and determines the stratification depth of each soil layer interface, including:
[0151] The marine soil to be identified was independently stratified using multiple stratification methods from the candidate stratification method set, resulting in multiple preliminary stratification results;
[0152] Consistency analysis or weighted fusion of the multiple preliminary stratification results are performed to determine the final stratification depth of each soil layer interface.
[0153] The specific details of each module in the above-mentioned marine soil stratification device have been described in detail in the corresponding marine soil stratification method, so they will not be repeated here.
[0154] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.
[0155] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.
[0156] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.
[0157] This disclosure also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device.
[0158] Computer-readable storage media can be, for example—but not limited to—electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.
[0159] A computer-readable storage medium can be sent, propagated, or transmitted for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.
[0160] A computer-readable storage medium carries one or more programs that, when executed by an electronic device, cause the electronic device to perform the methods described in the above embodiments.
[0161] Furthermore, this disclosure also provides an electronic device capable of implementing the above-described method.
[0162] Those skilled in the art will understand that various aspects of this disclosure can be implemented as a system, method, or program product. Therefore, various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."
[0163] The following reference Figure 8 To describe an electronic device 800 according to such an embodiment of the present disclosure. Figure 8 The electronic device 800 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.
[0164] like Figure 8 As shown, the electronic device 800 is presented in the form of a general-purpose computing device. The components of the electronic device 800 may include, but are not limited to: at least one processor 810, at least one memory 820, a bus 830 connecting different system components (including memory 820 and processor 810), and a display 840.
[0165] The memory stores program code that can be executed by the processor 810, causing the processor 810 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure. For example, the processor 810 can perform actions such as... Figure 1 As shown: Step S110, soil type identification is performed on the marine soil to be identified to obtain the target soil type; Step S120, according to the preset mapping relationship between soil type and stratification method, candidate stratification method corresponding to the target soil type is determined; Step S130, the marine soil to be identified is stratified based on the candidate stratification method, the stratification depth of each soil layer interface is determined, and a marine soil stratification result containing soil layer sequence and corresponding depth information is generated.
[0166] The memory 820 may include a readable medium in the form of volatile storage, such as random access memory (RAM) 8201 and / or cache memory 8202, and may further include read-only memory (ROM) 8203.
[0167] The memory 820 may also include a program / utility 8204 having a set (at least one) of program modules 8205, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.
[0168] Bus 830 can represent one or more of several types of bus structures, including a memory bus or memory controller, peripheral bus, graphics acceleration port, processor, or a local bus using any of the various bus structures.
[0169] Electronic device 800 can also communicate with one or more external devices 900 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 800, and / or with any device that enables electronic device 800 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 850. Furthermore, electronic device 800 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 860. As shown, network adapter 860 communicates with other modules of electronic device 800 via bus 830. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 800, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.
[0170] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.
Claims
1. A method for stratifying marine soil, characterized in that, include: Soil type identification is performed on the marine soil to be identified to obtain the target soil type of the marine soil to be identified; The process of identifying the soil type of the marine soil to be identified, and obtaining the target soil type of the marine soil to be identified, includes: Extract the soil characteristics of the marine soil to be identified, and extract the point cloud features corresponding to the surface laser point cloud data of the marine soil to be identified; The soil features and the point cloud features are input into the trained target machine learning model; The target machine learning model is used to identify the soil type of the marine soil to be identified, thereby obtaining the target soil type of the marine soil to be identified. The target machine learning model is obtained through the following methods: Key soil features are selected from a variety of marine soil characteristics; the feature importance score of the key soil features is higher than a preset score threshold; Soil survey and measurement data that are compatible with the key soil characteristics are obtained as the first training data, and soil surface laser point cloud data that are spatiotemporally aligned with the survey points are collected as the second training data. Based on the first and second training data, a target machine learning model for marine soil category identification is trained and generated. Based on the preset mapping relationship between soil type and stratification method, determine the candidate stratification method corresponding to the target soil type; The marine soil to be identified is stratified based on the candidate stratification method, the stratification depth of each soil layer interface is determined, and a marine soil stratification result containing soil layer sequence and corresponding depth information is generated.
2. The method according to claim 1, characterized in that, The marine soil characteristics include: physical properties, mechanical properties, and hydrological characteristics; The physical characteristics include at least one of particle size distribution, density, and moisture content; The mechanical properties include at least one of the following: tip resistance, sleeve resistance, and friction ratio in a static cone penetration test; The hydrological characteristics include at least one of pore water pressure and pore water capacity.
3. The method according to claim 1 or 2, characterized in that, The step of determining the candidate stratification method corresponding to the target soil type based on the preset mapping relationship between soil type and stratification method includes: Obtain a pre-constructed mapping table, which records each marine soil type and at least one corresponding adaptive stratification method; The entry matching the target soil type is found in the mapping table, and the associated set of candidate stratification methods is extracted.
4. The method according to claim 3, characterized in that, The step of performing stratification processing on the marine soil to be identified based on the candidate stratification method, and determining the stratification depth of each soil layer interface, includes: From the set of candidate stratification methods, the target stratification method is determined according to a preset selection strategy; The target stratification method is used to stratify the marine soil to be identified, and the stratification depth of each soil layer interface is determined. The preset selection strategy includes at least one of cost priority, accuracy priority, and efficiency priority.
5. The method according to claim 3, characterized in that, The step of performing stratification processing on the marine soil to be identified based on the candidate stratification method, and determining the stratification depth of each soil layer interface, includes: The marine soil to be identified was independently stratified using multiple stratification methods from the candidate stratification method set, resulting in multiple preliminary stratification results; Consistency analysis or weighted fusion of the multiple preliminary stratification results are performed to determine the final stratification depth of each soil layer interface.
6. A marine soil stratification device, characterized in that, include: The category identification module is used to identify the soil type of the marine soil to be identified, and to obtain the target soil type of the marine soil to be identified. The category recognition module is used to extract the soil characteristics of the marine soil to be identified, and to extract the point cloud features corresponding to the surface laser point cloud data of the marine soil to be identified. The soil features and the point cloud features are input into the trained target machine learning model; The target machine learning model is used to identify the soil type of the marine soil to be identified, thereby obtaining the target soil type of the marine soil to be identified. The target machine learning model is obtained through the following methods: Key soil features are selected from a variety of marine soil characteristics; the feature importance score of the key soil features is higher than a preset score threshold; Soil survey and measurement data that are compatible with the key soil characteristics are obtained as the first training data, and soil surface laser point cloud data that are spatiotemporally aligned with the survey points are collected as the second training data. Based on the first and second training data, a target machine learning model for marine soil category identification is trained and generated. The stratification method determination module is used to determine the candidate stratification method corresponding to the target soil type based on the preset mapping relationship between soil type and stratification method; The stratification result generation module is used to perform stratification processing on the marine soil to be identified based on the candidate stratification method, determine the stratification depth of each soil layer interface, and generate marine soil stratification results containing soil layer sequence and corresponding depth information.
7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the marine soil stratification method according to any one of claims 1 to 5.
8. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the marine soil stratification method according to any one of claims 1 to 5 by executing the executable instructions.