Soil layer identification method and system based on static sounding data

Abstract

The invention relates to the technical field of geotechnical engineering investigation, and provides a soil layer identification method and system based on static sounding data, and the method comprises the steps: carrying out the framing of static sounding sequence data through a sliding window, and carrying out the layering according to the formation process of a soil layer, the method comprises the following steps: inputting each soil layer in an uplink soil layer into an uplink head for soil layer evolution information extraction, inputting each soil layer in a downlink soil layer into a downlink head for soil layer evolution information extraction, carrying out vector splicing on features extracted by the uplink head and the downlink head, and then carrying out classification, so that a soil development and evolution rule can be comprehensively captured; redundant information is abandoned, and the purpose of capturing stratum depth sequence correlation to assist in soil texture type prediction is achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of geotechnical engineering investigation, and in particular relates to a soil layer identification method and system based on static penetration test data. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In nature, soil is usually distributed in layers, with diverse and complex types and significant temporal and spatial variability. Exploring geological structural characteristics (especially the accurate identification and classification of soil types at different depths) is a necessary prerequisite for human activities such as natural disaster prevention and control, engineering safety construction, and efficient exploitation of underground resources, and is of great significance.

[0004] At present, the identification of stratigraphic types is mainly achieved through methods such as engineering analogy, borehole sampling and in-situ testing. Due to the significant temporal and spatial variability of geological conditions, the engineering analogy method has great limitations, especially the reliability of different sites is difficult to guarantee. The borehole sampling method is usually costly, time-consuming, and has large sample disturbances. In addition, the manual analysis and judgment process of soil sampling data is inefficient and highly subjective. In-situ testing methods such as the Cone Penetration Test (CPT) have become the main methods for in-situ detection and identification of soil types. Scholars have proposed a series of charts based on CPT data to determine soil types. However, this type of method often faces the problem of large regional differences in applicability, and usually only classifies soil based on single-point test data, ignoring the continuity of soil layer sequence information, and has a low accuracy rate.

[0005] With the rapid development of artificial intelligence, intelligent identification of soil types has become possible and has also developed rapidly. However, the current relevant algorithms are still driven by single-point test data to identify and classify soil layers, ignoring the continuity of soil layer sequence information and failing to capture the laws of soil development and evolution, resulting in problems such as poor interpretability and generalization of the model. Summary of the invention

[0006] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a soil layer identification method and system based on static penetration data, which uses a sliding window to frame the static penetration sequence data and stratify them according to the formation process of the soil layer, and inputs each soil layer data in the upper soil layer into the upper head in reverse order to extract the soil layer evolution information, and inputs each soil layer data in the lower soil layer into the lower head to extract the soil layer evolution information, and classifies the features extracted by the upper head and the lower head after vector splicing. This can capture the soil development and evolution laws more comprehensively, discard redundant information, and achieve the purpose of capturing the correlation of the stratum depth sequence to assist in soil type prediction.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] A first aspect of the present invention provides a soil layer identification method based on static penetration test data, comprising:

[0009] Obtaining CPT sequence data arranged by depth;

[0010] The static penetration test sequence data is framed using a sliding window along the depth direction, each frame is centered on the soil layer to be classified, the soil layer before the formation of the soil layer to be classified is the upper soil layer, and the soil layer after the formation of the soil layer to be classified is the lower soil layer;

[0011] For each frame, a soil layer classification model is used based on the upper soil layer and the lower soil layer to obtain the soil layer type of the soil layer to be classified; the soil layer classification model includes a vector splicing operation, a soil quality classification layer, an upper header for extracting upper information, and a lower header for extracting lower information; the upper soil layer data is input into the upper header in reverse order, and the lower soil layer data is input into the lower header for information extraction; after performing a vector splicing operation on the features extracted from the upper header and the lower header, the soil quality classification layer is used to obtain the soil layer type of the soil layer to be classified.

[0012] Further, the static penetration sequence data includes the cone tip resistance and side friction resistance at each depth;

[0013] The cone tip resistance is the force acting on the cone tip divided by the cross-sectional area of ​​the cone base of the probe;

[0014] The side friction force is the force acting on the friction sleeve divided by the surface area of ​​the sleeve.

[0015] Furthermore, the soil layer type of the soil layer classification model during the training process is obtained by drilling and taking soil samples or by prior geological data.

[0016] Furthermore, both the upstream head and the downstream head are embedded with a long short-term memory network.

[0017] Furthermore, there is overlap in the soil layers between two adjacent frames.

[0018] A second aspect of the present invention provides a soil layer identification system based on static penetration test data, comprising:

[0019] A data acquisition module is configured to: acquire static penetration sequence data arranged by depth;

[0020] A framing module is configured to: frame the static penetration sequence data using a sliding window along the depth direction, with each frame centered on the soil layer to be classified, the soil layer before the formation of the soil layer to be classified is the upper soil layer, and the soil layer after the formation of the soil layer to be classified is the lower soil layer;

[0021] The soil layer classification model module is configured as follows: for each frame, based on the upper soil layer and the lower soil layer, a soil layer classification model is used to obtain the soil layer type of the soil layer to be classified; the soil layer classification model includes a vector splicing operation, a soil quality classification layer, an upper header for extracting upper information, and a lower header for extracting lower information; the upper soil layer data is input into the upper header in reverse order, and the lower soil layer data is input into the lower header for information extraction; after the features extracted from the upper header and the lower header are subjected to a vector splicing operation, the soil quality classification layer is used to obtain the soil layer type of the soil layer to be classified.

[0022] Further, the static penetration sequence data includes the cone tip resistance and side friction resistance at each depth;

[0023] The cone tip resistance is the force acting on the cone tip divided by the cross-sectional area of ​​the cone base of the probe;

[0024] The side friction force is the force acting on the friction sleeve divided by the surface area of ​​the sleeve.

[0025] Furthermore, the soil layer type of the soil layer classification model during the training process is obtained by drilling and taking soil samples or by prior geological data.

[0026] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the soil layer identification method based on static penetration data as described above.

[0027] The fourth aspect of the present invention provides a computer device, including a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, wherein when the processor executes the program, the steps in the soil layer identification method based on static penetration data as described above are implemented.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] The present invention uses a sliding window to frame the static penetration sequence data, and stratifies them according to the formation process of the soil layer. The data of each soil layer in the upper soil layer is input into the upper head in reverse order to extract the soil layer evolution information, and the data of each soil layer in the lower soil layer is input into the lower head to extract the soil layer evolution information. The features extracted by the upper head and the lower head are vector-joined and then classified. This can capture the soil development and evolution laws more comprehensively, discard redundant information, and achieve the purpose of capturing the correlation of the stratum depth sequence to assist in soil type prediction.

[0030] The present invention uses static penetration test sequence data as the basis for soil layer classification, which is efficient and easy to obtain and contains rich soil development and evolution information.

[0031] The present invention utilizes prior geological information as model training data to solve the limitations of traditional soil classification models, such as poor universality and low accuracy, which are caused by their inability to capture the evolution relationship of strata.

[0032] The overlapping frame data reading method proposed in the present invention can better integrate information of different depths and improve the continuity and effectiveness of data; in addition, under the overlapping reading method, a group of data will be repeatedly used many times, greatly improving the utilization rate of data information. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0034] Figure 1 is a flow chart of model training of embodiment 1 of the present invention;

[0035] Figure 2 Schematic diagram of overlapping frame data reading according to the first embodiment of the present invention;

[0036] Figure 3 is a schematic diagram of an underground soil layer classification model according to the first embodiment of the present invention;

[0037] Figure 4 Schematic diagram of LSTM network of embodiment 1 of the present invention;

[0038] Figure 5 is a schematic diagram of classification results of the second embodiment of the present invention;

[0039] Figure 6 This is a schematic diagram of a file selection page according to a second embodiment of the present invention;

[0040] Figure 7 This is a schematic diagram of the first jump pop-up window in the second embodiment of the present invention;

[0041] Figure 8 This is a schematic diagram of a second jump pop-up window in the second embodiment of the present invention;

[0042] Fig. 9 This is a structural diagram of a soil layer identification system based on static penetration test data according to a second embodiment of the present invention;

[0043] Fig.10 It is a structural diagram of a computer device according to a fourth embodiment of the present invention. DETAILED DESCRIPTION

[0044] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0045] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0046] Embodiment 1

[0047] This embodiment provides a soil layer identification method based on static penetration test data.

[0048] As the background technology introduces, the existing soil classification methods have the limitations of high manpower and financial costs and strong subjectivity; in addition, due to the lack of effective algorithm tools, the valuable historical prior geological information of the site cannot be fully utilized, which is undoubtedly a waste of resources. In recent years, supported by powerful computing power, deep learning technology has developed rapidly, bringing new ideas for the classification of underground soil layers based on prior knowledge. The present invention uses deep learning algorithms to capture the internal connections of stratum evolution and development, complete the rapid classification of underground soil layers, and construct a visualization map of soil types in each exploration hole with depth, so as to achieve the purpose of accurately and efficiently obtaining the spatial distribution of complex underground soil layers.

[0049] The present embodiment provides a soil layer identification method based on static penetration test data, including a CPT data reading method named "overlapping frame" and an underground soil layer classification model, using CPT to obtain the cone tip resistance (q c ), lateral friction (f s ) values ​​are used as feature data, and the soil layer types at different depths are identified by drilling holes with a drilling rig, or prior geological survey data are obtained; the proposed "overlapping frame" data reading method is used for data processing, and the data is input into the underground soil layer classification model to complete the intelligent identification of the soil layer.

[0050] The present embodiment provides a soil layer identification method based on static penetration test data, comprising the following steps:

[0051] Step 1: Model training. Collect the CPT sequence data and borehole soil labels of the site to be tested, and use the underground soil layer classification algorithm based on CPT data to obtain indicators such as classification accuracy, precision, recall rate and F1 score. Figure 1 As shown, the following steps are included:

[0052] Step 101: Collect CPT data of the site to be tested, obtain a CPT data sequence arranged by depth, and record features including cone tip resistance (q c ), lateral friction (f s) and depth (h). At the same time, a drilling rig is used to drill soil holes near the CPT exploration holes or to obtain the stratum distribution sequence from reliable prior geological survey data to obtain the soil category labels corresponding to the CPT data (i.e., soil layer types at different depths). Based on the depth, cone tip resistance and side friction resistance of each CPT exploration hole in the geological survey data and the corresponding soil type information, a dataset is constructed.

[0053] The CPT sequence data is collected by moving the static penetration test equipment to the designated location of the site to be tested, drilling a CPT test hole, and using the double-bridge static penetration test technology. The equipment automatically collects and records the cone tip resistance (q c ) and lateral friction (f s The cone tip resistance is calculated by dividing the measured force on the cone tip by the area A. c , and the sleeve friction (i.e., side friction) is calculated by dividing the force acting on the friction sleeve by the surface area of ​​the sleeve. c ) ranges from 10 to 15 cm 2 , the cone tip angle (ɑ) is 60°.

[0054] As an implementation method, for the double bridge static penetration test, the cross-sectional area of ​​the cone bottom of the probe is 10cm 2 , side wall area is 150cm 2 , cone tip angle is 60°, probe penetration rate is 2cm / s, and sampling interval is 0.1m.

[0055] To obtain soil type labels, reliable prior geological data are used, or soil samples are collected by drilling holes. A hole is drilled to a preset depth near the CPT exploration hole, and the soil layer types at different depths of the drilling location are identified. The drilling location, drilling depth and corresponding soil layer type information are obtained. In particular, due to the slow lateral geological changes, it can be considered that the soil sequence of the CPT exploration hole is consistent with the nearby boreholes.

[0056] If the drilling method is adopted, the specific steps are as follows: move the drilling rig to the vicinity of each CPT exploration hole, and level the drilling rig to place it stably and horizontally. Drill a hole at the specified location to the same depth as the CPT exploration hole, and identify the soil layer types at different depths of the drilling location. Obtain geological survey data consisting of the drilling depth and the corresponding soil layer type information, and obtain the soil type label corresponding to the CPT feature data.

[0057] Step 102: Use the "overlapping frame" data reading method to perform a "frame splitting" operation on the above data to construct a feature sequence with the target soil layer as the center point to provide sequence information containing the potential evolution relationship between the upper soil layer and the lower soil layer of the soil layer to be classified.

[0058] The feature sequence is constructed by the "overlapping frame" method, specifically: a data sequence with a frame length of L and an overlap of n is constructed with the soil layer x to be classified as the center point, where L = 2x-1. In this embodiment, L is set to 31, and the sliding step n is set to 1, that is, the length of each set of data is 31 lines, and one line of data is slid down each time to read the next set of data.

[0059] like Figure 2 As shown in FIG. 1 , the "overlap frame" data reading method is specifically implemented as follows: the CPT data sequence 1 along the depth direction is divided into continuous small segments along the depth direction, each segment is recorded as a frame, the depth 2 of the CPT data sequence 1 along the depth direction is h, the length 3 of each frame 5 is recorded as the frame length L, and the soil layer to be classified is recorded as x, that is, the x-th soil layer in each frame is used as the soil layer to be classified, L=2x-1. It is worth noting that the "soil layer" proposed in the present invention is defined as a certain thickness of underground soil to be classified, and the specific thickness value corresponds to the CPT data recording interval. In order to ensure the continuity and effective use of data, "overlap" is set between adjacent frames, and a certain sliding step size 4 (that is, sliding down n rows of data each time to read the next set of data) is used to achieve it, and the sliding step size 4 of the sliding window is n. Slide from top to bottom according to the depth sequence to obtain the sequence representation of all samples in the complete data set, and obtain the CPT sequence after the frame operation.

[0060] Step 103: adopt the proposed underground soil layer classification model, use the data obtained after the above-mentioned "framing" operation to train, and then complete the soil classification task, store the model parameters, and output the soil classification accuracy, precision, recall rate and F1 score (F1 score) indicators.

[0061] like Figure 3 As shown in the figure, the underground soil layer classification model includes two parts: information extraction layer 6 and soil classification layer 7. The soil type obtained by drilling cores near the exploration hole is used as the label of the corresponding exploration hole depth to train the underground soil layer classification model to classify soil. The information extraction layer of the model is used to extract the sequence information of the upper and lower soil layers of the target soil layer, and then the information features are processed by the soil classification layer after splicing, and the soil type prediction value of the target soil layer is output and the classification accuracy is given.

[0062] The information extraction layer 6 is mainly composed of an upstream header 10, a downstream header 9 and a vector concatenation operation 12. The upstream header 10 and the downstream header 9 are used to extract the upstream CPT sequence information 11 and the downstream soil layer CPT sequence information 8 respectively.

[0063] The upstream and downstream heads are each embedded with a 1-layer LSTM network. The input data are the framed CPT upstream and downstream sequences (batch=64), and the output is a 64×128 feature vector.

[0064] like Figure 4 As shown, the LSTM network converts the input x of the current time step into t and the external state h of the previous time step t-1 After the vector concatenation operation, the Sigmoid function 16 is used to obtain the forget gate f t , input gate i t and output gate o t , using the tanh function to get the storage unit See formulas (1) to (4); set the forget gate f t The memory cell c of the previous time step t-1 Perform vector element multiplication operation 14 to obtain the first result; input gate i t With storage unit Perform a vector element multiplication operation to obtain a second result; as shown in formula (5), perform a vector sum calculation 15 on the first result and the second result to obtain a memory unit c t ; Memory unit c t After activation by the tanh function and the output gate o t Perform vector element multiplication to obtain the external state h t , see formula (6); external state h t After passing the appropriate activation function, the final probability y of each category is obtained n .

[0065] f t =σ(W f x t +U f h t-1 +b f ) (1)

[0066] i t =σ(W i x t +U i h t-1 +b i ) (2)

[0067] o t =σ(W o x t +U o h t-1 +b o ) (3)

[0068]

[0069] h t =o t⊙tanh(c t ) (6)

[0070] Among them, σ(·) is the Logistic function, and its output interval is (0,1); W * , U * , b * is the learnable network parameter, *∈{i,f,o,c}; ⊙ is the product of vector elements.

[0071] Among them, the soil classification layer 7 uses a fully connected network (FC) to summarize and extract the upstream and downstream features, and output the probability of each type of the soil layer to be classified 13.

[0072] The upper soil layer is defined as the soil layer before the formation of any soil layer x, and the lower soil layer is the soil layer formed after any soil layer x.

[0073] Among them, the training of the underground soil layer classification model uses the soil type obtained by drilling near the CPT exploration hole, which is converted into a one-hot encoding and used as a label for supervised learning.

[0074] Specifically, the specific process of training is:

[0075] (1) Obtain CPT borehole data of the area to be tested, arrange them by depth, and drill cores near each borehole to obtain the soil type at the corresponding depth, perform one-hot encoding operations, and use them as labels for supervised learning. Use the "overlapping frame" data reading method to perform a "frame splitting" operation with a frame length of L on the CPT data and corresponding labels collected on site, construct a feature sequence with the target soil layer as the center point, and divide the training set and test set into 3:1 to train the underground soil layer classification model to classify soil.

[0076] (2) Based on the training data set, the model is trained with the features of each frame as input and the soil type of the soil layer x to be classified as the label. Specifically:

[0077] For the soil layer x to be classified, select the downstream soil layer of x in each frame, that is, the 1st to xth groups of data in the frame where x is the middle soil layer, and input them into the model's downstream header. At the same time, select the upstream soil layer, that is, the xth to 2x-1th groups of data in the frame where x is the middle soil layer, and input them into the model's upstream header in reverse order to extract sequence information. Set the input dimension to 3 and the hidden state to 64 to improve the model's expression and feature extraction capabilities. Then, concatenate the output features to obtain a feature vector with a dimension of 128. Finally, pass the concatenated features into the soil classification layer to process the information features, output the soil type prediction value of the target soil layer, and give the test set classification accuracy, precision, recall rate, and F1 score indicators.

[0078] Based on the training data set, the model is trained with the features of each frame as input and the soil type of the soil layer x to be classified as the label.

[0079] Specifically, the data after "framing" is input into the information extraction layer of the proposed underground soil layer classification model based on CPT data: the data sequence with a frame length of 31 is divided into two groups, each containing the soil layer to be classified, and the 1st to 16th rows of data of the descending soil layer are input into the descending head of the model, and the 16th to 31st rows of data of the upper soil layer are input into the upper head of the model in reverse order to extract the soil layer evolution information, where the input dimension is set to 3 and the hidden state is set to 64. The feature vector with a dimension of 128 obtained by splicing the features output by the information extraction layer is passed into the soil classification layer to obtain the final soil classification result. Repeat the above steps until the category information of all soil layers to be classified is obtained, and the overall accuracy, precision, recall rate and F1 score are calculated at the same time.

[0080] Step 2, soil layer identification. Obtain CPT sequence data arranged by depth; Use a sliding window to frame the CPT sequence data along the depth direction, with each frame centered on the soil layer to be classified, the soil layer before the formation of the soil layer to be classified is the upper soil layer, and the soil layer after the formation of the soil layer to be classified is the lower soil layer; For each frame, based on the upper soil layer and the lower soil layer, a soil layer classification model is used to obtain the soil layer type of the soil layer to be classified; The soil layer classification model includes a vector splicing operation, a soil quality classification layer, an upper header for extracting upper information, and a lower header for extracting lower information; Input the upper soil layer data into the upper header in reverse order, and input the lower soil layer data into the lower header for information extraction; After performing a vector splicing operation on the features extracted from the upper header and the lower header, the soil quality classification layer is used to obtain the soil layer type of the soil layer to be classified.

[0081] This embodiment provides a soil layer identification method based on static penetration test data, which uses static penetration test to obtain the cone tip resistance (q c ), lateral friction (f s ) values ​​are used as feature data, and the soil types at different depths are identified by drilling holes using a drilling rig to obtain training labels. The proposed “overlapping frame” data reading method is used for data processing, and the data is input into the proposed underground soil classification model based on CPT data to complete soil classification.

[0082] CPT can quickly obtain continuous soil test records (cone tip resistance q c and lateral friction f s ), with high resolution, fast and efficient, and wide application range. Using CPT data as model training and test data is efficient and easy to obtain, and contains rich soil development and evolution information.

[0083] The "overlapping frame" data reading method proposed in this embodiment can better integrate information at different depths and improve the continuity and effectiveness of data; in addition, under the overlapping reading method, a set of data will be reused repeatedly. Second, it greatly improves the utilization rate of data information.

[0084] This embodiment uses an intelligent soil layer identification algorithm to more comprehensively capture the laws of soil development and evolution, discard redundant information, and achieve the purpose of capturing the correlation of stratum depth sequences to assist in soil type prediction.

[0085] This embodiment achieves accurate prediction of the spatial distribution of complex rock and soil strata, promotes the informatization of underground engineering, speeds up the construction progress and improves the construction quality and safety.

[0086] Embodiment 2

[0087] The present embodiment provides a soil layer identification system based on static penetration sounding data, which embeds the underground soil layer classification model in the first embodiment. After using limited field data to quickly generate model parameters for the current site, it can realize on-site soil classification of newly input CPT feature data and provide a visualization of the soil type of each exploration hole, thereby achieving the purpose of quickly and efficiently performing on-site soil classification using prior geological information, and solving the limitations of traditional soil classification models such as poor universality and low accuracy caused by their inability to capture the evolution relationship of the strata.

[0088] This embodiment provides a soil layer identification system based on static penetration test data, such as Fig. 9 As shown, it specifically includes:

[0089] The model training module is configured to collect CPT sequence data and borehole soil type labels of the site to be tested, and use the underground soil layer classification algorithm based on CPT data to quickly train to obtain the pkl file of model parameters. The pkl file records the status of the trained model, including the model parameters, weights, etc.

[0090] The soil layer identification module is configured to: through the underground soil layer classification model, after using limited field data to quickly generate model parameters for the current site, it can quickly and efficiently classify the newly input CPT feature data on site, and give a visualization of the soil type of each exploration hole. Specifically, the above-mentioned pkl file is loaded, the soil type is identified for the newly collected CPT sequence data of the site to be predicted, and the classification accuracy is given, and the visualization of the soil layer classification of each exploration hole is completed.

[0091] The model training module includes:

[0092] A data acquisition module is configured to: acquire static penetration sequence data arranged by depth;

[0093] A framing module is configured to: frame the static penetration sequence data using a sliding window along the depth direction, with each frame centered on the soil layer to be classified, the soil layer before the formation of the soil layer to be classified is the upper soil layer, and the soil layer after the formation of the soil layer to be classified is the lower soil layer;

[0094] The soil layer classification model module is configured as follows: for each frame, based on the upper soil layer and the lower soil layer, a soil layer classification model is used to obtain the soil layer type of the soil layer to be classified; the soil layer classification model includes a vector splicing operation, a soil classification layer, an upper header for extracting upper information, and a lower header for extracting lower information; the upper soil layer data is input into the upper header in reverse order, and the lower soil layer data is input into the lower header for information extraction; after performing a vector splicing operation on the features extracted from the upper header and the lower header, the soil classification layer is used to obtain the soil layer type of the soil layer to be classified, and a pkl file recording the trained model status (including model parameters, weights, etc.) is output.

[0095] The specific implementation of the soil layer identification module is as follows: After loading the model, a pop-up window will be displayed showing the "File" selection page, such as Figure 6 As shown in the figure, the user can select a feature data file (csv format) of a new exploration hole to be classified, click the "Open" button, and jump to the pop-up window "Enter Pile Number". Figure 7 As shown in the figure, the user can enter the exploration hole number; after entering, click the "OK" button to jump to the pop-up window "Do you want to input another file? (yes / no)", as shown in the figure. Figure 8 As shown in the figure, it asks whether to input the next borehole data. The user enters the "yes" command and clicks the "OK" button to jump to the "File" page, and the above file input operation is repeated; if the "no" command is entered, the loop ends, and the model prints out the soil type of each soil layer to be classified in each borehole, and gives the prediction accuracy, and draws a visualization of the soil type along the depth of each borehole. The visualization of the classification result in this embodiment is shown in Figure 5 shown.

[0096] Through the underground soil layer identification system, after using limited field data to quickly generate model parameters for the current site, it is possible to perform on-site soil classification on the newly input CPT feature data, and provide a visualization of the soil type of each exploration hole. Prior geological information is used to quickly and efficiently perform on-site soil classification, solving the limitations of traditional soil classification models such as poor universality and low accuracy due to their inability to capture the evolution relationship of the strata.

[0097] This embodiment utilizes the underground soil layer rapid classification system to realize the rapid visualization of the soil layers of the on-site CPT drilling holes with depth within a certain range, and cleverly utilizes the prior information of the site within the same range to efficiently and conveniently classify the soil.

[0098] It should be noted here that each module in this embodiment corresponds to each step in Example 1 one by one, and the specific implementation process is the same, which will not be repeated here.

[0099] Embodiment 3

[0100] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the soil layer identification method based on static penetration data as described in the first embodiment above are implemented.

[0101] Embodiment 4

[0102] This embodiment provides a computer device, such as Fig.10 As shown, it includes a computer-readable storage medium 1003, a processor 1001, a communication interface 1002, and a computer program stored on the computer-readable storage medium 1003 and executable on the processor 1001, wherein the processor 1001, the communication interface 1002, and the computer-readable storage medium 1003 can be connected via a bus or other means. The communication interface 1002 is used to receive and send data, and when the processor 1001 executes the program, the steps in the soil layer identification method based on static penetration data as described in the first embodiment are implemented.

[0103] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1A function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A soil layer identification method based on static penetration data, characterized in that: include: Obtaining CPT sequence data arranged by depth; The static penetration test sequence data is framed using a sliding window along the depth direction, each frame is centered on the soil layer to be classified, the soil layer before the formation of the soil layer to be classified is the upper soil layer, and the soil layer after the formation of the soil layer to be classified is the lower soil layer; For each frame, a soil layer classification model is used based on the upper soil layer and the lower soil layer to obtain the soil layer type of the soil layer to be classified; the soil layer classification model includes a vector splicing operation, a soil quality classification layer, an upper header for extracting upper information, and a lower header for extracting lower information; the upper soil layer data is input into the upper header in reverse order, and the lower soil layer data is input into the lower header for information extraction; after performing a vector splicing operation on the features extracted from the upper header and the lower header, the soil quality classification layer is used to obtain the soil layer type of the soil layer to be classified.

2. A soil layer identification method based on static penetration data as claimed in claim 1, characterized in that: The static penetration test sequence data includes the cone tip resistance and the side friction resistance at each depth; The cone tip resistance is the force acting on the cone tip divided by the cross-sectional area of ​​the cone base of the probe; The side friction force is the force acting on the friction sleeve divided by the surface area of ​​the sleeve.

3. A soil layer identification method based on static penetration data as claimed in claim 1, characterized in that: The soil layer type of the soil layer classification model during the training process is obtained by drilling and taking soil samples or by prior geological data.

4. The soil layer identification method based on static penetration data according to claim 1 is characterized in that: The upstream head and the downstream head are both embedded with a long short-term memory network.

5. The soil layer identification method based on static penetration data according to claim 1 is characterized in that: There is overlap in the soil layers between two adjacent frames.

6. A soil layer identification system based on static penetration data, characterized in that: include: A data acquisition module is configured to: acquire static penetration sequence data arranged by depth; A framing module is configured to: frame the static penetration sequence data using a sliding window along the depth direction, with each frame centered on the soil layer to be classified, the soil layer before the formation of the soil layer to be classified is the upper soil layer, and the soil layer after the formation of the soil layer to be classified is the lower soil layer; The soil layer classification model module is configured as follows: for each frame, based on the upper soil layer and the lower soil layer, a soil layer classification model is used to obtain the soil layer type of the soil layer to be classified; the soil layer classification model includes a vector splicing operation, a soil quality classification layer, an upper header for extracting upper information, and a lower header for extracting lower information; the upper soil layer data is input into the upper header in reverse order, and the lower soil layer data is input into the lower header for information extraction; after the features extracted from the upper header and the lower header are subjected to a vector splicing operation, the soil quality classification layer is used to obtain the soil layer type of the soil layer to be classified.

7. A soil layer identification system based on static penetration data as claimed in claim 6, characterized in that: The static penetration test sequence data includes the cone tip resistance and the side friction resistance at each depth; The cone tip resistance is the force acting on the cone tip divided by the cross-sectional area of ​​the cone base of the probe; The side friction force is the force acting on the friction sleeve divided by the surface area of ​​the sleeve.

8. The soil layer identification system based on static penetration data according to claim 6 is characterized in that: The soil layer type of the soil layer classification model during the training process is obtained by drilling and taking soil samples or by prior geological data.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in a soil layer identification method based on static penetration test data as described in any one of claims 1 to 5 are implemented.

10. A computer device comprising a computer-readable storage medium, a processor, and a computer program stored in the computer-readable storage medium and executable on the processor, characterized in that: When the processor executes the program, the steps in the soil layer identification method based on static penetration test data as described in any one of claims 1 to 5 are implemented.

Images