A method for testing formation rock and soil parameters based on a controlled neutron source, a ground acquisition computer, a system and a computer-readable storage medium

Abstract

The present invention discloses a formation rock and soil parameter testing method based on a controllable neutron source, a ground acquisition computer, a system and a computer-readable storage medium, the method comprising: obtaining geotechnical mechanics and physical property synthetic data and geotechnical mechanics and physical property in-situ measured data, and performing preprocessing to obtain target rock and soil synthetic data and target rock and soil in-situ measured data; determining a preset dual neural network model, and training the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model; obtaining current geotechnical mechanics and physical property in-situ measured data, and inputting them into the formation rock and soil parameter prediction model, and outputting the formation rock and soil parameter prediction results. The present invention combines a controllable neutron source, static penetration test and data processing method, and introduces a neural network model, so as to achieve effective measurement and accurate analysis of key formation parameters and obtain more accurate formation parameter data.

Description

Technical Field

[0001] The present invention relates to the technical field of marine geological survey, and in particular to a method for testing formation rock and soil parameters based on a controllable neutron source, a ground acquisition computer, a system and a computer-readable storage medium. Background Art

[0002] In offshore wind power, natural gas hydrate development, offshore carbon storage and other new energy application scenarios, accurate assessment of formation characteristics is crucial to the successful implementation of the project. As a core means of evaluating formation engineering characteristics, the cone penetration test (CPT) technology has been widely used in geotechnical engineering surveys. The current CPT technology is mainly divided into two common forms: seabed cone penetration test (Seabed CPT) and downhole cone penetration test (Downhole CPT).

[0003] Seabed static penetration testing is usually used in marine environments. By penetrating the probe into the stratum from the seabed, mechanical parameters such as cone tip resistance, sidewall friction resistance and pore water pressure are tested. However, due to the limitation of test depth, seabed static penetration testing can usually only be carried out within a range of tens of meters to tens of meters below the seabed, which is difficult to meet the needs of deep geological exploration.

[0004] In contrast, downhole static penetration testing uses drilling technology to test at the bottom of the borehole, significantly increasing the penetration depth, breaking through the depth limit of seabed static penetration testing, and becoming a key tool in deep geotechnical exploration. However, in actual applications, existing downhole static penetration testing equipment can only measure in-situ mechanical parameters, but not physical parameters, which is difficult to meet the needs of new energy exploration for in-situ mechanical and physical parameters of the formation. Further technical improvements are urgently needed to improve its comprehensive analysis capabilities of formation characteristics.

[0005] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0006] The main purpose of the present invention is to provide a formation rock and soil parameter testing method based on a controllable neutron source, a ground acquisition computer, a system and a computer-readable storage medium, aiming to solve the problem in the prior art that the downhole static penetration measurement capability is limited, resulting in inaccurate measured rock and soil parameter data.

[0007] To achieve the above object, the present invention provides a method for testing formation rock and soil parameters based on a controlled neutron source, and the method for testing formation rock and soil parameters based on a controlled neutron source comprises the following steps:

[0008] Acquire geotechnical mechanics and physical property synthetic data and geotechnical mechanics and physical property in-situ measured data, and preprocess the geotechnical mechanics and physical property synthetic data and the geotechnical mechanics and physical property in-situ measured data to obtain target geotechnical synthetic data and target geotechnical in-situ measured data;

[0009] Determine a preset dual neural network model, and perform model training on the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model;

[0010] The current in-situ measured data of rock and soil mechanics and physical properties are obtained, and the current in-situ measured data of rock and soil mechanics and physical properties are input into the formation rock and soil parameter prediction model, and the formation rock and soil parameter prediction results are output.

[0011] Optionally, the method for testing formation rock and soil parameters based on a controlled neutron source, wherein:

[0012] The obtaining of geotechnical mechanics and physical property synthesis data specifically includes:

[0013] Acquire existing static penetration test mechanical parameter data, and use a preset density formula to calculate density data corresponding to the existing static penetration test mechanical parameter data;

[0014] Acquire indoor experimental test data, and perform interpolation processing on the indoor experimental test data to obtain indoor experimental test physical property parameter data;

[0015] The geotechnical mechanics and physical property composite data are obtained based on the existing static penetration mechanical parameter data, the density data and the indoor experimental test physical property parameter data.

[0016] Optionally, the method for testing formation rock and soil parameters based on a controlled neutron source, wherein obtaining the in-situ measured data of rock and soil mechanics and physical properties, specifically includes:

[0017] Acquiring downhole static penetration mechanical parameter data through a cone-tip detection probe in a downhole static penetration probe, and acquiring controllable neutron source physical property parameter data through a controllable neutron source measuring device in the downhole static penetration probe;

[0018] The in-situ measured data of rock and soil mechanics and physical properties are obtained based on the downhole static penetration mechanical parameter data and the controllable neutron source physical property parameter data.

[0019] Optionally, the method for testing formation rock and soil parameters based on a controlled neutron source, wherein the preprocessing of the rock and soil mechanics and physical property synthetic data and the rock and soil mechanics and physical property in-situ measured data to obtain target rock and soil synthetic data and target rock and soil in-situ measured data, specifically includes:

[0020] Performing data enhancement processing and normalization processing on the geotechnical mechanics and physical property synthetic data to obtain target geotechnical synthetic data, wherein the data enhancement processing includes time slicing processing, jitter processing and scaling processing;

[0021] The in-situ measured data of rock and soil mechanics and physical properties are subjected to outlier removal processing, data enhancement processing and normalization processing to obtain the in-situ measured data of target rock and soil.

[0022] Optionally, the method for testing formation rock and soil parameters based on a controlled neutron source, wherein the step of determining a preset dual neural network model and training the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model, specifically includes:

[0023] Determining a preset dual neural network model, wherein the preset dual neural network model includes a convolutional neural network and a recursive neural network;

[0024] The preset dual neural network model is subjected to model training processing, model fine-tuning processing and incremental training processing according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model.

[0025] Optionally, the method for testing formation rock and soil parameters based on a controlled neutron source, wherein the preset dual neural network model is subjected to model training processing, model fine-tuning processing and incremental training processing according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model, specifically includes:

[0026] Inputting the target rock and soil synthetic data into the convolutional neural network in the preset dual neural network model, extracting geological features from the target rock and soil synthetic data through the convolutional neural network, and obtaining a geological type classification result;

[0027] The geological type classification result is input into the recursive neural network in the preset dual neural network model, and the geological parameter prediction processing is performed on the geological type classification result by the recursive neural network, so that the model training processing is completed and the initial formation rock and soil parameter prediction model is obtained;

[0028] Inputting the target rock and soil in-situ measured data into the initial formation rock and soil parameter prediction model, and performing model fine-tuning processing on the initial formation rock and soil parameter prediction model according to the target rock and soil in-situ measured data to obtain an improved formation rock and soil parameter prediction model;

[0029] The improved stratum rock and soil parameter prediction model is incrementally trained to obtain a stratum rock and soil parameter prediction model.

[0030] Optionally, the method for testing formation geotechnical parameters based on a controlled neutron source, wherein the obtaining of current in-situ measured data of geotechnical mechanics and physical properties, and inputting the current in-situ measured data of geotechnical mechanics and physical properties into the formation geotechnical parameter prediction model, and outputting the formation geotechnical parameter prediction results, specifically includes:

[0031] Acquire current geotechnical mechanics and physical property in-situ measured data, and input the current geotechnical mechanics and physical property in-situ measured data into the formation geotechnical parameter prediction model;

[0032] The geological parameter prediction processing is performed on the current in-situ measured data of rock and soil mechanics and physical properties through the formation rock and soil parameter prediction model to obtain formation rock and soil parameter prediction results, wherein the formation rock and soil parameter prediction results include predicted density data and predicted porosity data.

[0033] In addition, to achieve the above-mentioned purpose, the present invention also provides a ground acquisition computer, wherein the ground acquisition computer includes: a memory, a processor, and a formation rock and soil parameter testing program based on a controlled neutron source stored in the memory and executable on the processor, wherein the formation rock and soil parameter testing program based on a controlled neutron source, when executed by the processor, implements the steps of the formation rock and soil parameter testing method based on a controlled neutron source as described above.

[0034] In addition, to achieve the above-mentioned purpose, the present invention also provides a formation rock and soil parameter testing system based on a controllable neutron source, wherein the formation rock and soil parameter testing system based on a controllable neutron source comprises: a downhole static penetration probe, a penetration and control system, and a data and transmission system;

[0035] The downhole static penetration probe comprises a cone-tip detection probe and a controllable neutron source measurement device;

[0036] The cone tip detection probe includes a cone tip resistance sensor, a side wall friction sensor and a pore water pressure sensor;

[0037] The controllable neutron source measuring device comprises a neutron receiver, a gamma receiver and a conversion joint;

[0038] The penetration and control system includes a propulsion device and a control system;

[0039] The data and transmission system includes an underground central control module, a data transmission line and the surface acquisition computer as described above.

[0040] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a formation rock and soil parameter testing program based on a controlled neutron source, and when the formation rock and soil parameter testing program based on a controlled neutron source is executed by a processor, the steps of the formation rock and soil parameter testing method based on a controlled neutron source as described above are implemented.

[0041] In the present invention, synthetic data of geotechnical mechanics and physical properties and in-situ measured data of geotechnical mechanics and physical properties are obtained, and the synthetic data of geotechnical mechanics and physical properties and the in-situ measured data of geotechnical mechanics and physical properties are preprocessed to obtain target geotechnical synthetic data and target geotechnical in-situ measured data; a preset dual neural network model is determined, and the preset dual neural network model is trained according to the target geotechnical synthetic data and the target geotechnical in-situ measured data to obtain a formation geotechnical parameter prediction model; current in-situ measured data of geotechnical mechanics and physical properties are obtained, and the current in-situ measured data of geotechnical mechanics and physical properties are input into the formation geotechnical parameter prediction model, and the formation geotechnical parameter prediction result is output. The present invention obtains synthetic data of geotechnical mechanics and physical properties and in-situ measured data of geotechnical mechanics and physical properties, and performs model training and model fine-tuning on the preset dual neural network model according to the synthetic data of geotechnical mechanics and physical properties and the in-situ measured data of geotechnical mechanics and physical properties, so as to construct a formation geotechnical parameter prediction model. Furthermore, the formation geotechnical parameter prediction model can realize the effective measurement and accurate analysis of the key physical parameters of the underground formation, and obtain more accurate and comprehensive formation physical parameter data. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is the first flow chart of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0043] Figure 2 It is a data processing flow diagram of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0044] Figure 3 This is a second flow chart of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0045] Figure 4 This is a third flow chart of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0046] Figure 5 It is a schematic diagram of a controlled neutron source CPT probe of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0047] Figure 6It is a schematic diagram of a penetration and control system and an underground static penetration system of a preferred embodiment of a formation rock and soil parameter testing method based on a controllable neutron source of the present invention;

[0048] Figure 7 It is a schematic diagram of a data acquisition and transmission system and an underground static penetration system of a preferred embodiment of a formation rock and soil parameter testing method based on a controllable neutron source of the present invention;

[0049] Figure 8 This is a fourth flow chart of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0050] Fig. 9 This is a fifth flow chart of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0051] Fig.10 It is a schematic diagram of a dual neural network architecture of a preferred embodiment of a formation rock and soil parameter testing method based on a controllable neutron source of the present invention;

[0052] Fig.11 This is a sixth flow chart of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0053] Fig.12 It is a schematic diagram of a model training process of a preferred embodiment of a formation rock and soil parameter testing method based on a controlled neutron source of the present invention;

[0054] Fig.13 This is the seventh flow chart of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0055] Fig.14 It is a schematic diagram of on-site drilling operation of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0056] Fig.15 It is a schematic diagram of on-site CPT operation of a preferred embodiment of the formation rock and soil parameter testing method based on a controlled neutron source of the present invention;

[0057] Fig.16 It is a schematic diagram of synthetic data of a preferred embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source of the present invention;

[0058] Fig.17 It is a structural diagram of a preferred embodiment of the ground acquisition computer of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0060] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 and Figure 2 The first embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source in the embodiment of the present invention includes:

[0061] The basic scheme of the present invention:

[0062] Step S10, obtaining geotechnical mechanics and physical property synthetic data and geotechnical mechanics and physical property in-situ measured data, and preprocessing the geotechnical mechanics and physical property synthetic data and the geotechnical mechanics and physical property in-situ measured data to obtain target geotechnical synthetic data and target geotechnical in-situ measured data.

[0063] The present invention generates synthetic geotechnical mechanics and physical property data based on existing geotechnical mechanics and physical property data, obtains in-situ measured data of geotechnical mechanics and physical properties through a controlled neutron source CPT, and performs data enhancement and normalization processing on the synthetic geotechnical mechanics and physical property data and the in-situ measured data of geotechnical mechanics and physical properties to generate training data for the model.

[0064] Step S20, determining a preset dual neural network model, and training the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model.

[0065] The present invention constructs a preset dual neural network model through a combination of dual neural networks, such as a convolutional neural network and a recursive neural network, and uses pre-processed target rock and soil synthetic data to train the preset dual neural network model. Afterwards, the preset dual neural network model is fine-tuned using the pre-processed target rock and soil in-situ measured data, which can optimize the weight parameters of the preset dual neural network model, so that the formation rock and soil parameter prediction model obtained after subsequent training can output more accurate geological prediction parameters.

[0066] Step S30, obtaining current in-situ measured data of geotechnical mechanics and physical properties, and inputting the current in-situ measured data of geotechnical mechanics and physical properties into the formation geotechnical parameter prediction model, and outputting the formation geotechnical parameter prediction result.

[0067] After the formation geotechnical parameter prediction model is trained, the formation geotechnical parameter prediction model can be used to perform geotechnical parameter prediction processing on the input current geotechnical mechanics and physical property in-situ measured data, and the required formation density data and formation porosity data (i.e., the formation geotechnical parameter prediction results in the present invention, which can also be referred to as formation physical property parameter data) can be generated quickly and accurately.

[0068] The present invention provides a controlled neutron source in-situ CPT testing method and system for marine engineering development scenarios such as offshore wind power, natural gas hydrate development, and carbon sequestration. The system includes testing equipment hardware and software and analysis and calculation software. The testing equipment is used to obtain the original data of physical parameters of porosity and density, as well as mechanical parameters such as cone tip resistance, side wall resistance and pore water pressure. The analysis and calculation software processes the synthetic data and the original data through artificial intelligence consistency analysis technology, establishes a machine learning model, and establishes the relationship between physical parameters and mechanical parameters to obtain more accurate and comprehensive formation physical parameter data.

[0069] Further technical solutions of the present invention:

[0070] See also Figure 3 The second embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source in the embodiment of the present invention includes:

[0071] S101, obtaining existing static penetration test mechanical parameter data, and using a preset density formula to calculate density data corresponding to the existing static penetration test mechanical parameter data;

[0072] It is understandable that the present invention obtains the physical property parameters of a certain position by obtaining the existing mechanical parameter data of the position using an empirical formula. For example, the following preset density formula can be used to calculate the density data corresponding to the existing mechanical parameter data:

[0073] The expression of the preset density formula is:

[0074] ;in, is the density, , , , are constants, For depth, is the cone tip resistance, is the side wall friction, It is a logarithmic operation.

[0075] S102, acquiring indoor experimental test data, and performing interpolation processing on the indoor experimental test data to obtain indoor experimental test physical property parameter data;

[0076] S103, obtaining the geotechnical mechanics and physical property synthesis data according to the existing static penetration test mechanical parameter data, the density data and the indoor experimental test physical property parameter data;

[0077] Furthermore, the constants in the empirical formula are corrected by acquiring the physical property parameter data of the indoor experimental test at the location, so as to obtain the optimal synthetic data of the rock and soil mechanics and physical properties.

[0078] See also Figure 4 The third embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source in the embodiment of the present invention includes:

[0079] S104, obtaining downhole static penetration mechanical parameter data through a cone-tip detection probe in the downhole static penetration probe, and obtaining controllable neutron source physical property parameter data through a controllable neutron source measurement device in the downhole static penetration probe;

[0080] S105. Obtain the in-situ measured data of rock and soil mechanics and physical properties according to the downhole static penetration mechanical parameter data and the controllable neutron source physical property parameter data.

[0081] In order to obtain more comprehensive information on formation physical properties, the nuclear density cone probe (ND-CP) technology was introduced. The nuclear density cone probe can supplement the mechanical parameters of the cone tip resistance, sidewall friction and pore water pressure of the traditional CPT by measuring the soil density in a small range around the cone head. However, the nuclear density cone probe technology uses radioactive cesium isotopes as gamma ray sources, which has potential radioactive risks. Once the radioactive source is lost or leaked, it may cause serious environmental pollution problems.

[0082] In order to solve this problem, the present invention proposes a CPT test method combined with a controlled neutron source. The controlled neutron source is not radioactive when not activated and will not pollute the environment, greatly improving the safety of operation and the environmental friendliness of the equipment. In addition, the controlled neutron source can measure not only density data, but also porosity data, which is an important physical property parameter in geotechnical engineering and can provide more key data for the analysis of underground formation characteristics.

[0083] The data acquisition device in the present invention (the present invention is provided with an underground CPT system, i.e., the formation rock and soil parameter testing system based on a controllable neutron source in the present invention) is constructed as follows:

[0084] The main components of the downhole CPT (Downhole Cone Penetration Test) system include downhole static penetration probe, penetration and control system, and data acquisition and transmission system.

[0085] like Figure 5As shown, for the downhole static penetration probe: the downhole static penetration probe is the core component of the downhole CPT system, including two important parts: the cone tip detection probe and the controlled neutron source measurement device.

[0086] Among them, the cone tip detection probe: three sensors are integrated in the probe to obtain the mechanical properties of the formation. The three sensors are: 1. Cone tip resistance sensor: measures the resistance encountered by the cone tip during advancement, reflecting the shear strength of the soil; 2. Side wall friction sensor: measures the friction between the probe surface and the soil, provides information on the friction characteristics of the soil, and helps to understand the uniformity and density of the formation structure; 3. Pore water pressure sensor: monitors the changes in pore water pressure in the formation, and evaluates the permeability and pore characteristics of the soil (a key parameter for determining saturation and formation stability).

[0087] Among them, the controllable neutron source measurement device: a small controllable neutron source system is integrated in the probe to measure the density and porosity of the formation. The device includes: 1. Neutron emitter: used to generate neutrons and emit them into the surrounding formations; 2. Neutron receiver and gamma receiver: detect the signals after neutron scattering and absorption, analyze the density and porosity of the formation. The controllable neutron source system is non-radioactive in the unactivated state, has high safety, and is suitable for long-term underground operations; 3. Conversion joint: used to connect the controllable neutron source device with the hydraulic cylinder to ensure the effective transmission of signals and power.

[0088] like Figure 6 As shown, for the penetration and control system: the penetration and control system is responsible for pushing the CPT probe accurately and stably into the formation at the bottom of the borehole, and usually includes a control system, a propulsion device, hydraulic pipelines, drilling equipment and a downhole static penetration system.

[0089] Among them, the propulsion device: generally composed of a winch and a hydraulic or electric propulsion system, is used to ensure that the probe maintains a constant speed and force during the penetration process to reduce formation disturbance. Taking hydraulics as an example, the system includes: 1. Hydraulic propulsion cylinder: The propulsion cylinder on the deck is connected to the hydraulic cylinder of the downhole static penetration system through a hydraulic pipeline. The hydraulic cylinder provides a constant propulsion force to ensure that the speed of the probe is controllable when penetrating the formation and adapt to the changes in the density of different formations. As an independent system, the hydraulic pipeline is not connected to the casing, drill bit or downhole central control system underground. 2. Winch: used to control the lifting position of the probe, so as to facilitate the adjustment of the probe position in boreholes of different depths.

[0090] Among them, the control system: including the control computer, propulsion controller, etc., is used to manage the propulsion process. The operator can monitor and adjust the propulsion parameters in real time through the operation interface of the control computer to ensure the stability of the test process and the accuracy of the measurement.

[0091] like Figure 7As shown, for the data acquisition and transmission system: the data acquisition and transmission system is used to monitor and record the formation parameters obtained by the CPT probe in real time, and transmit the data to the ground processing equipment, wherein the data acquisition and transmission system generally includes a ground acquisition computer, a propulsion device, a data transmission line, a drilling equipment and an underground static penetration system.

[0092] The underground static penetration system includes an underground central control module: used to control the measurement process of each sensor and transmit the collected data to the surface system. The central control module is connected to each sensor to achieve efficient data collection and transmission. The underground central control passes through the hydraulic cylinder, but the data cable is not connected to the hydraulic cylinder, but directly connected to the static penetration probe at the bottom.

[0093] Data transmission line: used to transmit measurement data to ground data acquisition equipment in real time, suitable for exploration scenarios with greater depths underground, to ensure data stability and real-time performance. The upper part of the data transmission line is connected to the deck winch and the ground acquisition computer, while the lower part is connected to the underground central control module and probe to form a complete transmission path, supporting continuous transmission of data from the probe to the ground.

[0094] Surface acquisition computer: used to receive and display the measured data of the downhole CPT sensor in real time, such as cone tip resistance, sidewall friction, pore water pressure, density and porosity, etc. The computer is also responsible for running the data acquisition and analysis software to achieve data storage, real-time monitoring and subsequent analysis and processing.

[0095] The various components of the underground CPT system work together to achieve high-precision exploration and multi-parameter measurement tasks in deep soil. Through the organic combination of sensors, control systems and data transmission, the underground CPT system can efficiently and accurately record the mechanical and physical properties of the formation. Furthermore, through data processing and modeling, continuous stratigraphic profile data is generated to provide scientific data support for marine geological surveys. This system is particularly suitable for new energy projects in complex geological environments, and provides key geological data for the implementation of offshore wind power, natural gas hydrate development and other projects.

[0096] See also Figure 8 The fourth embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source in the embodiment of the present invention includes:

[0097] S106, performing data enhancement processing and normalization processing on the geotechnical mechanics and physical property synthetic data to obtain target geotechnical synthetic data, wherein the data enhancement processing includes time slicing processing, dithering processing and scaling processing;

[0098] S107, performing outlier removal processing, data enhancement processing and normalization processing on the in-situ measured data of rock and soil mechanics and physical properties to obtain in-situ measured data of target rock and soil.

[0099] The data processing process is as follows:

[0100] like Figure 2 As shown, the first step of data processing is data preprocessing. Before there is measured data, the preprocessing stage generates synthetic data through the in-situ CPT mechanical parameter data (i.e., the existing static penetration mechanical parameter data in the present invention) and the sample indoor test test physical parameter data (i.e., the indoor experimental test physical parameter data in the present invention) collected at adjacent positions. At the same time, data enhancement techniques (such as time slicing, jittering, scaling, etc.) are used to diversify the data to increase the amount and diversity of data, thereby improving the training effect and generalization ability of the model. In addition, for the measured data collected in real time (i.e., the in-situ measured data of rock and soil mechanics and physical properties in the present invention), it is also necessary to perform similar preprocessing steps on the CPT mechanical parameters (i.e., the downhole static penetration mechanical parameter data in the present invention) and the controllable neutron source physical parameter data (i.e., the controllable neutron source physical parameter data in the present invention) to ensure that they match the format and range of the synthetic data. Afterwards, the synthetic data and the measured data collected by the device of the present invention need to be normalized.

[0101] See also Fig. 9 The fifth embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source in the embodiment of the present invention includes:

[0102] S201, determining a preset dual neural network model, wherein the preset dual neural network model includes a convolutional neural network and a recursive neural network;

[0103] Due to the limited downhole space and advancement length, the existing nuclear density cone probe can only obtain local limited density data within the penetration depth, which has great data limitations. Based on this, the present invention proposes a CPT test method using a machine learning algorithm by combining software and hardware. This method uses the measured parameters such as cone tip force, sidewall friction resistance and pore water pressure, and uses a machine learning model to predict density and porosity, thereby expanding the application scope and applicable scenarios of the equipment and improving the application effect of CPT equipment in deep geological exploration.

[0104] S202, performing model training processing, model fine-tuning processing and incremental training processing on the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data, to obtain a formation rock and soil parameter prediction model.

[0105] Among them, for the dual neural network structure used in the present invention: the machine learning model used in the present invention is a dual neural network structure, in which the first layer of the neural network is used for the classification of the formation type (such as Fig.10 As shown). The first layer inputs synthetic and measured data including key formation parameters such as burial depth, cone tip resistance, sidewall friction resistance and pore water pressure. This layer divides the formation into different types such as sand and clay by learning these input features, laying the foundation for the prediction of subsequent physical property parameters. When the classification is completed, the second layer of neural network predicts key physical property parameters such as density and porosity of sand and clay respectively according to the classification results (that is, the first layer of neural network outputs the formation type, and the second layer of neural network outputs density and porosity).

[0106] See also Fig.11 The sixth embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source in the embodiment of the present invention includes:

[0107] S2021, inputting the target rock and soil synthetic data into the convolutional neural network in the preset dual neural network model, extracting geological features of the target rock and soil synthetic data through the convolutional neural network, and obtaining a geological type classification result;

[0108] S2022, inputting the geological type classification result into the recursive neural network in the preset dual neural network model, performing geological parameter prediction processing on the geological type classification result through the recursive neural network, then the model training processing is completed, and an initial stratum rock and soil parameter prediction model is obtained;

[0109] S2023, inputting the target rock and soil in-situ measured data into the initial formation rock and soil parameter prediction model, and performing model fine-tuning processing on the initial formation rock and soil parameter prediction model according to the target rock and soil in-situ measured data to obtain an improved formation rock and soil parameter prediction model;

[0110] S2024. Perform incremental training processing on the improved formation rock and soil parameter prediction model to obtain a formation rock and soil parameter prediction model.

[0111] For the training phase of the two-layer neural network model used in the present invention: the training process of the model is divided into three phases: synthetic data training, measured data fine-tuning, and incremental training (such as Fig.12 as shown).

[0112] 1. Synthetic data training phase: In the initial training phase, the model uses synthetic data for basic training of general learning. Through the law of marine sedimentation and stratigraphic distribution, the data enhancement technology in the preprocessing phase is used to simulate and generate a large amount of synthetic data with different stratigraphic characteristics, so that the model has the ability to learn on a wide range of input-output relationships. This phase focuses on the basic classification of stratigraphic types and the universal prediction of physical parameters (such as density and porosity). The first output head is used to generate basic prediction results, obtain the approximate weights of the neural network, capture the overall trend of stratigraphic parameters, and provide a solid initial model for subsequent specialized learning.

[0113] 2. Measured data fine-tuning stage: After initial training, the model enters the measured data fine-tuning stage, in which the model is specifically learned in combination with real data collected from the field, and the weights of the neural network model are fine-tuned to further optimize the model performance. Unlike the basic prediction of the first output head, the second output head focuses on using measured data to make local adjustments to the model weights to adapt to the complexity and diversity of specific geological conditions.

[0114] Among them, the process of weight fine-tuning is mainly to refine the model's understanding of the actual geological environment based on the general model constructed in the initial training stage. Specifically, by introducing measured data, the model is able to correct the deviations in the synthetic data. This fine-tuning not only targets the local optimization of the deep network weights, but also affects the high-level feature extraction, allowing the model to more accurately predict actual formation parameters such as density and porosity. In addition, a small step learning rate is used in the fine-tuning process to avoid damage to the original capabilities of the model, while ensuring that the impact of measured data on weights can fully reflect the actual geological characteristics. In this way, while retaining the general prediction ability, the model significantly improves its adaptability and prediction accuracy in specific geological scenarios, thereby providing more reliable data support for stratigraphic analysis under complex geological conditions.

[0115] 3. Incremental training phase: In order to keep the model highly accurate in the ever-changing geological conditions, incremental training optimizes the model's weights and parameters by continuously introducing new field data. At regular intervals, the system introduces new collected data into the model, updates the general learning output, and adjusts the specialized learning output to ensure that the model still has high predictive power and adaptability when facing new data. This closed-loop iterative structure not only helps the model adapt quickly when responding to new geological information, but also ensures the long-term stability and accuracy of the model. Through this continuous optimization and adaptive adjustment, the model can continuously improve the reliability and accuracy of the prediction. The model can maintain high efficiency and superior adaptability when facing new geological exploration scenarios. The incremental learning structure allows the model to update itself over time, allowing it to gradually adapt to the changing needs of different geological environments and exploration depths.

[0116] See also Fig.13 The seventh embodiment of the method for testing formation rock and soil parameters based on a controlled neutron source in the embodiment of the present invention includes:

[0117] S301, obtaining current in-situ measured data of geotechnical mechanics and physical properties, and inputting the current in-situ measured data of geotechnical mechanics and physical properties into the formation geotechnical parameter prediction model;

[0118] S302. Perform geological parameter prediction processing on the current in-situ measured data of rock and soil mechanics and physical properties through the formation rock and soil parameter prediction model to obtain formation rock and soil parameter prediction results, wherein the formation rock and soil parameter prediction results include predicted density data and predicted porosity data.

[0119] Using CPT equipment based on a controlled neutron source, the current in-situ measured data of geotechnical mechanics and physical properties are collected on site and input into the geotechnical parameter prediction model. For the part that contains both mechanical parameters and physical parameters, the model further optimizes the prediction ability through the trained weights; for data that only contains mechanical parameters, the model directly outputs the prediction results of physical parameters.

[0120] During the data processing, model building and optimization process, the model combines synthetic data with measured data, and applies machine learning algorithms to establish the relationship between the mechanical parameters of the formation (such as cone tip resistance, sidewall friction, pore water pressure) and physical parameters (such as density and porosity), thereby generating continuous density and porosity profiles. The model not only makes predictions based on current data, but can also continuously optimize and continuously improve the accuracy and reliability of the model through adaptive adjustment and combined with measured data collected in the future. Ultimately, the system can achieve long-term and stable formation predictions, provide effective data support and scientific decision-making basis for complex geological exploration, thereby greatly improving the implementation efficiency and safety of projects such as offshore wind power and natural gas hydrate mining.

[0121] Example application:

[0122] The field testing steps include drilling operations, downhole CPT testing, data acquisition and measurement, and the integration and continuous operation of drilling and testing.

[0123] 1. Drilling operation: First, drilling operation is carried out on the geological exploration ship (such as Fig.14). A borehole that meets the CPT test requirements is drilled in the target area using dedicated drilling equipment to ensure that the depth and diameter of the borehole meet the test requirements, such as a borehole diameter of 100 mm. Before drilling, a preliminary assessment of the formation conditions is required to select the appropriate drill bit type, drilling parameters and mud mix. These preparations help avoid potential risks such as wellbore collapse and drill cuttings blockage. During the drilling process, a top drive or power head is used to rotate the drill pipe and the bottom hole assembly (BHA), and mud is pumped through the inside of the drill pipe to remove the cuttings generated during the drilling process. The flow of mud not only keeps the borehole unobstructed, but also prevents the wellbore from collapsing. When the BHA reaches the preset position (such as 10 meters below the seabed), drilling stops, but the mud continues to flow to stabilize the wellbore. During the entire drilling operation, the seabed template has been lowered to the seabed, and the CPT probe remains parked on the rig in standby mode so that subsequent static penetration tests can be carried out at any time.

[0124] 2. Downhole CPT test: After drilling is completed, the seabed base plate fixes the drill pipe and disconnects the top drive or power head from the top of the drill pipe. The umbilical cable head is deployed into the drill pipe through the active lifting compensation umbilical winch and connected to the top of the CPT probe through the umbilical cable head connector (such as Fig.15 ). The operator uses the propulsion controller to lower the probe to the bottom of the borehole and locks it with the BHA to prepare for the static penetration test. The operator sends a start command to the CPT probe to start the probe's propulsion system. The CPT probe slowly extends from the BHA at a constant speed (usually 1-2 cm / s) and gradually advances into the formation. During the advancement process, the probe collects key data such as the mechanical and physical properties of the formation in real time, including cone tip resistance, sidewall friction, pore water pressure, density and porosity. A constant speed is maintained during the advancement process to ensure the accuracy of the data until the preset maximum penetration depth (for example, 3 meters) is reached. At this point, the probe is 13 meters below the seabed.

[0125] 3. Data acquisition and measurement: During the test, the controlled neutron source system of the CPT probe is started synchronously, and the density and porosity of the formation are measured using neutron scattering and absorption technology. Compared with traditional radioactive measurement methods, the controlled neutron source is non-radioactive when not activated, which greatly improves the safety of operation and avoids environmental pollution and radioactive risks. This non-radioactive technology is more environmentally friendly and safe, and is particularly suitable for long-term operation in marine environments. During the penetration process, the probe maintains real-time data transmission to ensure continuous monitoring and recording of key parameters (such as cone tip resistance, sidewall friction, pore water pressure, density and porosity). It is worth noting that the controlled neutron source detector is located 1-2 meters above the CPT probe. For example, if the maximum penetration depth of the CPT probe is 3 meters, 3 meters of mechanical parameter data (such as cone tip resistance, sidewall friction, pore water pressure) and 1-2 meters of real-time density and porosity data can be collected. These data need to be further processed (i.e., preprocessing of measured data) to generate a complete and continuous formation mechanical and physical property characteristic profile. After the test, the CPT probe will be retracted into the BHA and recovered to the ship through the umbilical cable system. At this time, the operator can download data, detect faults and perform equipment maintenance with the probe through wired communication. This method not only ensures the progress of continuous drilling, but also completes the drill pipe connection and unloading cycle before the next penetration test, preparing for the next drilling operation.

[0126] 4. Integration and continuous operation of drilling and testing: This testing method allows drilling and CPT testing to be carried out alternately, providing efficient continuous operation support. At each stage of drilling, equipment recovery and redeployment do not require waiting, allowing drilling and testing to be carried out alternately in deep exploration.

[0127] For example, after the drilling equipment drills the BHA to 10 meters below the seabed, the CPT probe is lowered and penetrated from 3 meters to 13 meters below the seabed. Subsequently, the CPT probe is recovered and the drilling equipment drills to a depth of 13 meters. Then, the CPT probe is lowered again and penetrated to a depth of 16 meters. Through this alternating drilling and testing method, the project can achieve accurate and continuous measurement of deep formations, ensuring the integrity and consistency of formation exploration data.

[0128] After data collection is completed, data analysis and model building are carried out:

[0129] Specifically, the data processing and model optimization of the present invention includes three parts: 1. data preprocessing; 2. construction of a dual neural network model; 3. training of a dual neural network model.

[0130] The data preprocessing process is as follows:

[0131] 1. Preprocessing the synthetic data: Data preprocessing includes processing the synthetic data (i.e., the synthetic data of geotechnical mechanics and physical properties in the present invention) and the actual field data (i.e., the in-situ measured data of geotechnical mechanics and physical properties in the present invention). For the composition of the synthetic data, taking density as an example, firstly, the data of depth, cone tip resistance and sidewall friction (i.e., the existing static penetration mechanics parameter data in the present invention) are read from a measured CPT borehole, and the density is calculated by a preset formula. The expression of the preset density formula is:

[0132] ;in, is the density, , , , are constants, For depth, is the cone tip resistance, is the side wall friction, Next, the unit weight measured in the laboratory (also known as experimental density data) is obtained by combining the indoor experimental test data corresponding to the borehole, and these experimental data are interpolated so that they can match the calculated data in the depth range. This step ensures the consistency and accuracy between the calculated data and the actual measured data.

[0133] After that, a nonlinear curve fitting method (such as the curve_fit function, i.e., the fitting function) is used to adjust the parameters in the formula (such as Fig.16 The calculation results are closer to the laboratory data. This process ensures that the calculated density value can more accurately reflect the physical properties of the actual formation by optimizing the parameters of the formula. Through these steps, a set of data including depth, cone tip resistance, sidewall friction and density is generated.

[0134] In order to further improve the generalization ability of the model, data enhancement techniques (such as adding random perturbations, scaling, translation, etc.) are used to expand the diversity of the data set to enhance the model's adaptability under different geological conditions, such as clay and sandy soil conditions. Multiple sets of processed data ultimately provide high-quality input for the training of the machine learning model, improving the accuracy and stability of the model's prediction results.

[0135] After the preprocessing of the synthetic data is completed, the measured data needs to be preprocessed. The process is as follows: For the measured data of the equipment, it is first necessary to clean up the outliers in the first and last data of the measured data to ensure the accuracy of the data. In addition, the controllable neutron source used in the present invention does not directly measure density and porosity, but is derived through indirect physical principles. Taking density measurement as an example, based on the Compton scattering principle of gamma rays, when gamma rays pass through matter, they collide and scatter with atomic electrons, part of the gamma rays are absorbed by the matter, and the other part reaches the detector. The number of gamma rays reaching the detector is proportional to the density of the matter, so the density of the matter can be derived by measuring the number of rays.

[0136] The derivation process of density depends on the ratio of gamma-ray counts detected by near and far detectors, and usually involves multiple energy windows and specific empirical coefficients. These empirical coefficients are determined during the calibration process by comparing the ratio of gamma-ray counts in different energy windows, combined with the mineral composition, porosity and fluid properties of the formation, so that the density derivation can more accurately reflect the actual situation of the formation (these calibrations are usually completed before offshore construction).

[0137] In order to ensure the accuracy of the measurement, the commonly used fitting formula or rock physics model needs to be adjusted according to the specific conditions of the formation to ensure that the model is consistent with the actual geological conditions. In some complex formation conditions, field sampling and experimental verification are necessary, which can help verify and correct the density derivation model, thereby further improving the prediction accuracy of formation physical property parameters.

[0138] Furthermore, after the data preprocessing is completed, the machine learning model needs to be trained using the preprocessed data. The process is as follows:

[0139] 1. Determine the dual neural network model to be used: In the process of geological parameter prediction, the present invention uses a dual neural network architecture to improve the prediction accuracy of geotechnical parameters. For example, a combination of convolutional neural network (CNN) and recurrent neural network (RNN) is used to construct a neural network (each neural network has its own unique role, and the combined use can better extract features from multi-parameter data and make accurate predictions).

[0140] 2. Strata classification: In the stratigraphic classification stage, machine learning algorithms (such as convolutional neural networks, CNNs) are called to extract and classify multi-parameter data (such as cone tip resistance, sidewall friction, pore water pressure, etc.). The core advantage of CNN lies in its convolution operation, which can effectively capture the local patterns and spatial relationships of data, and is particularly suitable for processing multi-dimensional inputs such as stratigraphic data. In exploration, the physical properties of strata vary significantly in different types of strata. For example, the cone tip resistance of sand is often much greater than that of clay, and the pore water pressure of sand is smaller. This local feature is crucial for stratigraphic classification. The convolution layer of CNN can extract these local features layer by layer by applying convolution kernels (filters) to the input data, such as detecting changes in cone tip resistance and sidewall friction at different depths, and identifying the relationship between these changes and stratigraphic types. Compared with traditional machine learning methods, CNN is particularly good at processing spatially structured data and can more effectively learn the differences between different strata.

[0141] In this system, the input layer of CNN receives raw data from CPT (static penetration test), including burial depth, cone tip resistance, sidewall friction, pore water pressure, etc. Through the convolution layer, CNN can extract high-dimensional features from these raw data layer by layer and generate feature maps. These feature maps can reflect local patterns in the strata, such as the significant differences in physical parameters between shallow sand and deep clay.

[0142] The convolution operation of CNN can not only identify these local features, but also further extract key information through pooling layers (such as maximum pooling or average pooling), reduce the dimension of the data, and retain important spatial structure information. After the superposition of multiple convolutional layers and pooling layers, the model can capture the significant feature differences between different stratum types and generate feature maps with strong discriminative power.

[0143] In the final classification layer of the model, the extracted features will be used to distinguish different formation types such as sand and clay. The classification layer outputs the probability of belonging to a certain formation type (such as sand or clay). These probability results provide a basis for subsequent physical property parameter predictions, allowing the model to make more accurate density and porosity predictions for different formation types (such as sand and clay).

[0144] Furthermore, for density and porosity output: Based on the results of the stratum classification, the model will call different machine learning models (such as recurrent neural networks (RNN) or other regression models) to predict the density and porosity of each stratum. After the classification is completed, the model will make specific parameter predictions for different types of strata (such as sand and clay). In particular, when using RNN, the model can process data with sequence structures or implicit time series characteristics.

[0145] In theory, the strata can be regarded as the manifestation of soil in time series. During the CPT test, as the probe gradually advances, the strata data at different depths show a certain order. The characteristics of each stratum are often related to the state of the previous layer, especially when predicting physical parameters such as density and porosity, the change in depth will affect the physical properties of the stratum. This dependency is similar to the "long dependency" in time series data, and RNN has significant advantages in dealing with such continuously changing trends and characteristics.

[0146] RNN can effectively capture historical information in the layered structure through its internal memory units (such as LSTM, Long Short-Term Memory or GRU, Gated Recurrent Unit, recurrent neural network), so as to better predict the parameter changes of the subsequent layers using the data of the previous layer. In the density and porosity prediction, the data input to the RNN network comes from the high-dimensional feature mapping results previously extracted by the convolutional neural network (CNN), such as sand and clay. These features include important parameters such as cone tip resistance, sidewall friction resistance and pore water pressure.

[0147] By integrating sequential features at different depths, RNN further improves the overall prediction accuracy of the model for parameters such as density and porosity, making it particularly suitable for geological exploration tasks with complex stratigraphic structures. By capturing and utilizing the order and dependencies in the strata, RNN can provide more accurate and predictive analysis of geological physical parameters.

[0148] 3. Model training: Model training is divided into three stages: initial training stage, data fine-tuning stage and incremental training stage.

[0149] Initial training phase: In the initial training phase, the model is mainly trained using synthetic data. By combining the existing geological sedimentary pattern library and CPT data, a large amount of synthetic data is generated, covering a variety of stratigraphic types and complex geological features. The synthetic data is preprocessed to generate a broadly representative simulated data set, mainly clay and sand data sets. These data provide sufficient samples for convolutional neural networks (CNNs) and recurrent neural networks (RNNs), and the model can identify the association between various geological parameters by learning these data. The goal of the initial training is to enable the model to have basic predictive capabilities and be able to predict stratigraphic conditions, namely physical parameters density and porosity, without measured data.

[0150] Data fine-tuning phase: After the initial training is completed, the model enters the data fine-tuning phase. At this point, the measured data collected from the field is introduced into the model to further train and optimize the model. Compared with synthetic data, measured data can better reflect the actual geological conditions. Therefore, at this stage, the model can correct the synthetic data deviation that may exist in the initial training by learning from the measured data. The key to the fine-tuning phase is to improve the model's adaptability to specific geological environments, especially different types of clay and sand. By using measured data, the model can better capture subtle geological features, thereby improving its prediction accuracy in practical applications. At the same time, the model's weights and parameters will be adjusted based on the measured data to ensure that it can more accurately predict the physical parameters density and porosity.

[0151] Incremental training phase: As field exploration continues, new measured mechanical parameter data and physical property parameter data will continue to be generated. In the incremental training phase, the model is incrementally updated and trained by continuously introducing new field data. In this way, the model can gradually update its weights and parameters to adapt to changing geological conditions and maintain efficient prediction performance. Incremental training enables the model to self-adjust when responding to new data. It can not only absorb the latest exploration results, but also avoid the overfitting problem of the model through continuous optimization. As the amount of mechanical parameter and physical property parameter data increases and time passes, the prediction accuracy and generalization ability of the model will gradually improve, ensuring that it can maintain high-precision prediction effects in different geological environments.

[0152] The resulting model combines mechanical parameters (cone tip resistance, sidewall friction, pore water pressure) and physical parameters (density and porosity) data to form a continuous geotechnical mechanical and physical parameter profile (i.e., cone tip resistance, sidewall friction, pore water pressure, density and porosity profile) to establish the complex nonlinear relationship between the physical and mechanical properties of the formation. In addition, the static calibration data will be further used to fine-tune the geotechnical physical model to improve the response accuracy to changes in formation properties and ensure the reliability and accuracy of the model's prediction of density and porosity.

[0153] The downhole controlled neutron source static penetration test (CPT) system of the present invention and its application have the following effects and advantages:

[0154] 1. Integration of controlled neutron source technology: Traditional static penetration technology relies on radioactive sources for measurement, which poses safety and environmental risks. The present invention uses controlled neutron source technology to measure the density and porosity of soil through neutron scattering and absorption. Since the controlled neutron source is non-radioactive when not in operation, it greatly improves the safety and environmental protection of the operation, avoids the potential hazards of radioactive sources, and facilitates transportation and storage.

[0155] 2. Expansion of underground static penetration testing technology: Compared with the traditional seabed CPT, the present invention breaks through the depth limit of seabed CPT by performing penetration testing at the bottom of the borehole, and realizes effective exploration of deeper soil. In addition, the present invention extends the traditional CPT, which is limited to the measurement of geotechnical parameters, to the simultaneous acquisition of geotechnical and formation physical properties, including key physical parameters such as density and porosity, making the geological analysis more comprehensive.

[0156] 3. Intelligent modeling improves formation assessment accuracy: This invention introduces machine learning technology to construct multi-parameter models such as cone tip resistance, sidewall friction resistance, pore water pressure, density and porosity. These data are used to generate continuous geotechnical profiles through machine learning models, which significantly improves the analysis accuracy of formation characteristics and can more accurately predict and evaluate the physical and mechanical characteristics of formations.

[0157] 4. Combination of hardware and software and self-iterative optimization of the model: This invention combines the advantages of hardware and software, collects data in real time through hardware devices, and provides training data for machine learning models. Through a closed-loop iterative structure, the model continuously introduces new field data for updating, which improves the reliability and adaptability of the prediction. With the continuous introduction of new data, the model is gradually optimized to achieve more accurate geological surveys and assessments.

[0158] In summary, by introducing machine learning algorithms, CPT equipment can not only generate more comprehensive density and porosity data in deep geological surveys, but also improve its ability to adapt to different formation conditions. Based on a large amount of historical data, the machine learning model realizes dynamic adjustment of new data through adaptive optimization, further enhancing the prediction accuracy and reliability of the model. This innovative CPT testing method effectively broadens the application effect of the equipment in deep geological surveys and provides a more accurate geological analysis method for the development of new marine energy.

[0159] Furthermore, if Fig.17 As shown, based on the above-mentioned formation rock and soil parameter testing method based on a controllable neutron source, the present invention also provides a ground acquisition computer accordingly, and the ground acquisition computer includes a processor 10, a memory 20 and a display 30. Fig.17 Only some of the components of the surface acquisition computer are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0160] In some embodiments, the memory 20 may be an internal storage unit of the ground acquisition computer, such as a hard disk or memory of the ground acquisition computer. In other embodiments, the memory 20 may also be an external storage device of the ground acquisition computer, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the ground acquisition computer. Further, the memory 20 may also include both an internal storage unit of the ground acquisition computer and an external storage device. The memory 20 is used to store application software and various types of data installed on the ground acquisition computer, such as the program code of the ground acquisition computer installed. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a formation geotechnical parameter test program 40 based on a controlled neutron source is stored on the memory 20, and the formation geotechnical parameter test program 40 based on a controlled neutron source can be executed by the processor 10, thereby realizing the formation geotechnical parameter test method based on a controlled neutron source in the present application.

[0161] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run the program code or process data stored in the memory 20, such as executing the formation geotechnical parameter testing method based on a controlled neutron source.

[0162] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information collected on the ground computer and to display a visual user interface.

[0163] In one embodiment, when the processor 10 executes the formation rock and soil parameter testing program 40 based on a controlled neutron source in the memory 20, the steps of the formation rock and soil parameter testing method based on a controlled neutron source as described above are implemented.

[0164] Furthermore, if Figure 5 , Figure 6 as well as Figure 7 As shown, based on the above-mentioned formation rock and soil parameter testing method based on a controllable neutron source, the present invention also provides a formation rock and soil parameter testing system based on a controllable neutron source, wherein the formation rock and soil parameter testing system based on a controllable neutron source includes: a downhole static penetration probe, a penetration and control system, and a data and transmission system;

[0165] The downhole static penetration probe comprises a cone-tip detection probe and a controllable neutron source measurement device;

[0166] The cone tip detection probe includes a cone tip resistance sensor, a side wall friction sensor and a pore water pressure sensor;

[0167] The controllable neutron source measuring device comprises a neutron receiver, a gamma receiver and a conversion joint;

[0168] The penetration and control system includes a propulsion device and a control system;

[0169] The data and transmission system includes an underground central control module, a data transmission line and the surface acquisition computer as described above.

[0170] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a formation rock and soil parameter testing program based on a controlled neutron source, and when the formation rock and soil parameter testing program based on a controlled neutron source is executed by a processor, the steps of the formation rock and soil parameter testing method based on a controlled neutron source as described above are implemented.

[0171] In summary, the present invention provides a formation rock and soil parameter testing method, system, ground acquisition computer and computer-readable storage medium based on a controllable neutron source, the method comprising: obtaining geotechnical mechanics and physical property synthetic data and geotechnical mechanics and physical property in-situ measured data, and preprocessing the geotechnical mechanics and physical property synthetic data and the geotechnical mechanics and physical property in-situ measured data to obtain target rock and soil synthetic data and target rock and soil in-situ measured data; determining a preset dual neural network model, and training the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model; obtaining current rock and soil mechanics and physical property in-situ measured data, and inputting the current rock and soil mechanics and physical property in-situ measured data into the formation rock and soil parameter prediction model, and outputting formation rock and soil parameter prediction results. The present invention obtains geotechnical mechanics and physical property synthetic data and geotechnical mechanics and physical property in-situ measured data, and performs model training and model fine-tuning on a preset dual neural network model according to the geotechnical mechanics and physical property synthetic data and geotechnical mechanics and physical property in-situ measured data, thereby constructing a formation geotechnical parameter prediction model. Furthermore, the formation geotechnical parameter prediction model can realize effective measurement and accurate analysis of key physical property parameters of underground formations, and obtain more accurate and comprehensive formation physical property parameter data.

[0172] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or ground acquisition computer including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or ground acquisition computer. In the absence of more restrictions, an element defined by the sentence "comprises one..." does not exclude the existence of other identical elements in the process, method, article or ground acquisition computer including the element.

[0173] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the program can be stored in a computer-readable storage medium that can be read by a computer, and the program can include the processes of the above-mentioned method embodiments when executed. The computer-readable storage medium can be a memory, a disk, an optical disk, etc.

[0174] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for testing formation rock and soil parameters based on a controlled neutron source, characterized in that: The method for testing formation rock and soil parameters based on a controlled neutron source includes: Acquire geotechnical mechanics and physical property synthetic data and geotechnical mechanics and physical property in-situ measured data, and preprocess the geotechnical mechanics and physical property synthetic data and the geotechnical mechanics and physical property in-situ measured data to obtain target geotechnical synthetic data and target geotechnical in-situ measured data; The obtaining of geotechnical mechanics and physical property synthesis data specifically includes: Acquire existing static penetration test mechanical parameter data, and use a preset density formula to calculate density data corresponding to the existing static penetration test mechanical parameter data; Acquire indoor experimental test data, and perform interpolation processing on the indoor experimental test data to obtain indoor experimental test physical property parameter data; Obtaining the geotechnical mechanics and physical property synthesis data according to the existing static penetration test mechanical parameter data, the density data and the indoor experimental test physical property parameter data; Obtaining the in-situ measured data of geotechnical mechanics and physical properties, specifically including: Acquiring downhole static penetration mechanical parameter data through a cone-tip detection probe in a downhole static penetration probe, and acquiring controllable neutron source physical property parameter data through a controllable neutron source measuring device in the downhole static penetration probe; Obtaining the rock and soil mechanics and physical property in-situ measured data according to the downhole static penetration mechanical parameter data and the controllable neutron source physical property parameter data; The preprocessing of the geotechnical mechanics and physical property synthetic data and the geotechnical mechanics and physical property in-situ measured data to obtain target geotechnical synthetic data and target geotechnical in-situ measured data specifically includes: Performing data enhancement processing and normalization processing on the geotechnical mechanics and physical property synthetic data to obtain target geotechnical synthetic data, wherein the data enhancement processing includes time slicing processing, jitter processing and scaling processing; Performing outlier removal processing, data enhancement processing, and normalization processing on the in-situ measured data of rock and soil mechanics and physical properties to obtain in-situ measured data of target rock and soil; Determine a preset dual neural network model, and perform model training on the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model; The step of determining a preset dual neural network model and training the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model specifically includes: Determining a preset dual neural network model, wherein the preset dual neural network model includes a convolutional neural network and a recursive neural network; According to the target rock and soil synthetic data and the target rock and soil in-situ measured data, the preset dual neural network model is subjected to model training processing, model fine-tuning processing and incremental training processing to obtain a formation rock and soil parameter prediction model; The method of performing model training processing, model fine-tuning processing and incremental training processing on the preset dual neural network model according to the target rock and soil synthetic data and the target rock and soil in-situ measured data to obtain a formation rock and soil parameter prediction model specifically includes: Inputting the target rock and soil synthetic data into the convolutional neural network in the preset dual neural network model, extracting geological features from the target rock and soil synthetic data through the convolutional neural network, and obtaining a geological type classification result; The geological type classification result is input into the recursive neural network in the preset dual neural network model, and the geological parameter prediction processing is performed on the geological type classification result by the recursive neural network, so that the model training processing is completed and the initial formation rock and soil parameter prediction model is obtained; Inputting the target rock and soil in-situ measured data into the initial formation rock and soil parameter prediction model, and performing model fine-tuning processing on the initial formation rock and soil parameter prediction model according to the target rock and soil in-situ measured data to obtain an improved formation rock and soil parameter prediction model; Performing incremental training processing on the improved stratum rock and soil parameter prediction model to obtain a stratum rock and soil parameter prediction model; The current in-situ measured data of rock and soil mechanics and physical properties are obtained, and the current in-situ measured data of rock and soil mechanics and physical properties are input into the formation rock and soil parameter prediction model, and the formation rock and soil parameter prediction results are output.

2. The method for testing formation rock and soil parameters based on a controlled neutron source according to claim 1, characterized in that: The obtaining of current in-situ measured data of geotechnical mechanics and physical properties, and inputting the current in-situ measured data of geotechnical mechanics and physical properties into the formation geotechnical parameter prediction model, and outputting the formation geotechnical parameter prediction results specifically include: Acquire current geotechnical mechanics and physical property in-situ measured data, and input the current geotechnical mechanics and physical property in-situ measured data into the formation geotechnical parameter prediction model; The geological parameter prediction processing is performed on the current in-situ measured data of rock and soil mechanics and physical properties through the formation rock and soil parameter prediction model to obtain formation rock and soil parameter prediction results, wherein the formation rock and soil parameter prediction results include predicted density data and predicted porosity data.

3. A ground collection computer, characterized in that: The ground acquisition computer includes: a memory, a processor, and a formation rock and soil parameter testing program based on a controlled neutron source stored in the memory and executable on the processor. When the formation rock and soil parameter testing program based on a controlled neutron source is executed by the processor, the steps of the formation rock and soil parameter testing method based on a controlled neutron source as described in any one of claims 1-2 are implemented.

4. A formation rock and soil parameter testing system based on a controllable neutron source, characterized in that: The controllable neutron source-based formation rock and soil parameter testing system includes: an underground static penetration probe, a penetration and control system, and a data and transmission system; The downhole static penetration probe comprises a cone-tip detection probe and a controllable neutron source measurement device; The cone tip detection probe includes a cone tip resistance sensor, a side wall friction sensor and a pore water pressure sensor; The controllable neutron source measuring device comprises a neutron receiver, a gamma receiver and a conversion joint; The penetration and control system includes a propulsion device and a control system; The data and transmission system includes a downhole central control module, a data transmission line and a surface acquisition computer as described in claim 3.

5. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a formation rock and soil parameter testing program based on a controlled neutron source, and when the formation rock and soil parameter testing program based on a controlled neutron source is executed by a processor, the steps of the formation rock and soil parameter testing method based on a controlled neutron source as described in any one of claims 1-2 are implemented.

Images