Geological survey report creation support server, geological survey report creation support system, geological survey report creation support program, and geological survey report creation support method

Abstract

This invention provides a geological survey report creation support server, a geological survey report creation support system, a geological survey report creation support program, and a geological survey report creation support method that enable the automatic creation of highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers, even without prior experience or expertise in geological surveys. [Solution] The system includes a geological survey data acquisition unit 41, an N-value determination unit 42 that determines the validity of the N-value using an N-value determination model M1, a corrected N-value prediction unit 43 that predicts the corrected N-value using a corrected N-value prediction model M2, a ground constant calculation unit 45 that calculates ground constants, and a geological survey report creation unit 48 that performs consistency evaluation of ground constants, fine-tuning, and creation of a geological survey report using an integrated inference model M6.

Description

Technical Field

[0001] The present invention relates to a geological survey report creation support server, a geological survey report creation support system, a geological survey report creation support program, and a geological survey report creation support method suitable for assisting in the creation of geological survey reports.

Background Art

[0002] Conventionally, a geological survey report is created through a process of creating a columnar diagram using dedicated software that only creates a columnar diagram, creating the text using document creation software, calculating various numerical values such as ground constants using spreadsheet software and empirical rules, pasting on-site photos using photo processing software, and finally binding. Note that Japanese Patent Application Laid-Open No. 2002-227176 discloses a list of ground geological survey results (Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, the conventional work of creating a geological survey report has a problem of taking a lot of time and effort because it requires many steps as described above. Also, when a skilled technician finally determines numerical data such as ground constants, they judge the validity in consideration of the geological tendency peculiar to the region, the feel during excavation, and the consistency with other numerical data, and make fine adjustments as necessary. For this reason, it is a highly personal work that depends on the empirical rules and tacit knowledge of skilled technicians, and there is also a problem that geological survey reports created by technicians lacking experience and skill are inferior in terms of accuracy and reliability.

[0005] The present invention was made to solve these problems and aims to provide a geological survey report creation support server, a geological survey report creation support system, a geological survey report creation support program, and a geological survey report creation support method that can automatically create highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers, even without experience or expertise in geological surveys. [Means for solving the problem]

[0006] The geological survey report creation support server according to the present invention solves the problem of automatically creating highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers in an easy and rapid manner, even without experience or expertise in geological surveys. The server is a geological survey report creation support server for supporting the creation of geological survey reports and includes: a geological survey data acquisition unit that acquires geological survey data from a user terminal; an N-value determination unit that determines whether the N-values ​​of each depth included in the geological survey data are valid using an N-value determination model; a corrected N-value prediction unit that predicts corrected N-values ​​using a corrected N-value prediction model for N-values ​​that have been determined to be invalid by the N-value determination unit; a ground constant calculation unit that calculates ground constants using the N-values ​​that have been determined to be valid by the N-value determination unit or the N-values ​​predicted by the corrected N-value prediction unit; and a geological survey report creation unit that determines the consistency of the ground constants calculated by the ground constant calculation unit and performs a process of creating a geological survey report after making fine adjustments to ground constants that are inconsistent using an integrated inference model.

[0007] Furthermore, in one aspect of the present invention, in order to solve the problem of predicting coefficients suitable for the survey area and soil layer classification and calculating more accurate ground constants, the present invention may have a coefficient prediction unit that predicts coefficients to be used in calculating ground constants based on the survey area and soil layer classification included in the geological survey data using a coefficient prediction model, and the ground constant calculation unit may calculate ground constants using the N value determined to be appropriate by the N value determination unit or the N value predicted by the corrected N value prediction unit and the coefficient predicted by the coefficient prediction unit.

[0008] Furthermore, in one aspect of the present invention, in order to solve the problem of determining the validity of ground constants, predicting appropriate ground constants according to the survey area and soil layer classification, and determining the consistency of ground constants with higher accuracy, the present invention includes a ground constant determination unit that determines whether the ground constants calculated by the ground constant calculation unit are valid or not using a ground constant determination model, and a corrected ground constant prediction unit that predicts corrected ground constants using a corrected ground constant prediction model for ground constants that the ground constant determination unit has determined to be invalid, and the geological survey report creation unit may determine the consistency of the ground constants determined to be valid by the ground constant determination unit or the ground constants predicted by the corrected ground constant prediction unit using the integrated inference model.

[0009] Furthermore, in one aspect of the present invention, in order to solve the problem of accurately determining whether or not the N-value of geological survey data is noise such as boulders, the N-value determination model is machine-trained to output a determination result of whether or not each N-value is noise when an N-value at each depth is input, and the N-value determination unit may determine the validity of the N-value based on the determination result.

[0010] Furthermore, in one aspect of the present invention, in order to solve the problem of predicting a reasonable N-value according to depth and soil layer classification, the corrected N-value prediction model is machine-trained to output a corrected N-value when it receives the depth and soil layer classification of an N-value that has been determined to be inappropriate by the N-value determination unit as input, and the corrected N-value prediction unit may predict the corrected N-value as the corrected N-value.

[0011] Furthermore, in one aspect of the present invention, in order to solve the problem of integrating the output results of each trained model and making a final decision that reflects the empirical rules and tacit knowledge of skilled engineers, the integrated inference model may determine the consistency of each output result by integrally processing the output results of the N-value determination model, the corrected N-value prediction model, the coefficient prediction model, the ground constant determination model, and the corrected ground constant prediction model, along with the geological survey data.

[0012] Furthermore, the geological survey report creation support system according to the present invention comprises a geological survey report creation support server in any of the above-described embodiments and the user terminal, in order to solve the problem of automatically creating highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers in a simple and rapid manner, even without experience or expertise in geological surveys.

[0013] Furthermore, the geological survey report creation support program according to the present invention solves the problem of automatically creating highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers in an easy and rapid manner, even without experience or expertise in geological surveys. The program is a geological survey report creation support program for assisting in the creation of geological survey reports, and it functions as a computer comprising: a geological survey data acquisition unit that acquires geological survey data from a user terminal; an N-value determination unit that determines whether the N-values ​​of each depth included in the geological survey data are valid using an N-value determination model; a corrected N-value prediction unit that predicts corrected N-values ​​for N-values ​​that have been determined to be invalid by the N-value determination unit using a corrected N-value prediction model; a ground constant calculation unit that calculates ground constants using the N-values ​​determined to be valid by the N-value determination unit or the N-values ​​predicted by the corrected N-value prediction unit; and a geological survey report creation unit that uses an integrated inference model to execute a process of determining the consistency of the ground constants calculated by the ground constant calculation unit, and making fine adjustments to ground constants that are inconsistent before creating a geological survey report.

[0014] Furthermore, the geological survey report creation support method according to the present invention is a geological survey report creation support method for supporting the creation of geological survey reports in order to solve the problem of automatically creating highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers in an easy and quick manner, even without experience or expertise in geological surveys, and comprises: a geological survey data acquisition step of acquiring geological survey data from a user terminal; an N-value determination step of determining whether the N-values ​​of each depth included in the geological survey data are valid using an N-value determination model; a corrected N-value prediction step of predicting corrected N-values ​​using a corrected N-value prediction model for N-values ​​that were determined to be invalid in the N-value determination step; a ground constant calculation step of calculating ground constants using the N-values ​​determined to be valid in the N-value determination step or the N-values ​​predicted in the corrected N-value prediction step; and a geological survey report creation step of executing a process by an integrated inference model to determine the consistency of the ground constants calculated in the ground constant calculation step, and to create a geological survey report after making fine adjustments to ground constants that are inconsistent. [Effects of the Invention]

[0015] According to the present invention, even without experience or expertise in geological surveys, it is possible to automatically create highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers in a simple and rapid manner. [Brief explanation of the drawing]

[0016] [Figure 1] This is a block diagram showing one embodiment of the geological survey report creation support server and geological survey report creation support system according to the present invention. [Figure 2] This figure shows an example of the display screen of the geological survey report creation support site in this embodiment. [Figure 3] This diagram shows the various pre-trained models that the geological survey report creation support server possesses in this embodiment. [Figure 4]It is a flowchart showing the processing steps of a geological survey report creation support method executed by a geological survey report creation support server and a geological survey report creation support program according to the present invention.

Embodiment for Carrying Out the Invention

[0017] Hereinafter, embodiments of a geological survey report creation support server, a geological survey report creation support program, a geological survey report creation support method, and a geological survey report creation support system according to the present invention will be described with reference to the drawings.

[0018] [1] System Configuration As shown in FIG. 1, the geological survey report creation support system 100 of the present embodiment includes a user terminal 10 for inputting geological survey data and obtaining a geological survey report, and a geological survey report creation support server 1 for supporting the creation of a geological survey report. The user terminal 10 and the geological survey report creation support server 1 are communicably connected to each other via a communication network such as the Internet. Hereinafter, each configuration will be described.

[0019] [2] User Terminal 10 The user terminal 10 is composed of a computer such as a smartphone, a tablet, or a personal computer, and is used by a user engaged in geological survey work or the like. In the present embodiment, a browser capable of browsing a geological survey report creation support site or the like as shown in FIG. 2 is installed in the user terminal 10. Then, the geological survey report creation support program 1a provided by the geological survey report creation support server 1 as a web application is used via the browser.

[0020] Specifically, the user terminal 10 transmits (uploads) various geological survey data input by the user on the data input screen of the geological survey report creation support site to the geological survey report creation support server 1, or obtains (downloads) the geological survey report created by the geological survey report creation support server 1.

[0021] The geological survey data includes all the data necessary for the creation of a geological survey report, such as columnar diagram information obtained through boring surveys and image data of on-site photos. Further, the columnar diagram information includes the survey area, soil layer classification at each depth (e.g., sandy soil, clayey soil, gravelly soil, etc.), boundary depth of the soil layer, layer thickness of each soil layer, continuity relationship of the soil layer within the boring hole, and N-values obtained through standard penetration tests, etc.

[0022] [3] Geological Survey Report Creation Support Server 1 The geological survey report creation support server 1 is composed of a computer such as a cloud server or a physical server, and supports the creation of a geological survey report based on the geological survey data received from the user terminal 10. In this embodiment, as shown in FIG. 1, the geological survey report creation support server 1 mainly has a communication means 2, a storage means 3, and an arithmetic processing means 4. Hereinafter, each constituent means will be described.

[0023] The communication means 2 implements a communication function in the geological survey report creation support server 1. In this embodiment, the communication means 2 is composed of a communication module compatible with carrier network standards such as the 5th generation mobile communication network (5G) and wireless LAN standards such as Wi-Fi (registered trademark). And the communication means ② is adapted to transmit and receive various data to and from the user terminal 10 via a communication network such as the Internet.

[0024] The storage means 3 stores various data and functions as a working area when the arithmetic processing means 4 performs arithmetic processing. In this embodiment, the storage means 3 is composed of a solid state drive (SSD), a hard disk drive (HDD), a ROM (Read Only Memory), a RAM (Random Access Memory), a flash memory, etc. As shown in FIG. 1, it has a program storage unit 31, a learned model storage unit 32, and a learning data storage unit 33.

[0025] The program storage unit 31 has the geological survey report creation support program 1a of this embodiment installed. The calculation processing means 4 then executes the geological survey report creation support program 1a, causing the computer acting as the geological survey report creation support server 1 to function as one of the components described later. In this embodiment, the geological survey report creation support program 1a is provided as a web application from the geological survey report creation support server 1.

[0026] The trained model storage unit 32 stores multiple trained models that form the core of artificial intelligence (AI). In this embodiment, as shown in Figure 1, the trained model storage unit 32 stores an N-value determination model M1, a corrected N-value prediction model M2, a coefficient prediction model M3, a ground constant determination model M4, a corrected ground constant prediction model M5, and an integrated inference model M6.

[0027] These trained models are uniquely designed and built specifically for geological survey work. They are trained using machine learning based on numerous geological survey reports accumulated in the past, as well as ground parameters and bearing layer determination results adopted by skilled geotechnical engineers as final judgments. As a result, the internal parameters of each trained model reflect the empirical rules and tacit knowledge of skilled engineers, such as subtle ground tendencies that cannot be calculated using general formulas and are difficult to express numerically.

[0028] The N-value judgment model M1 is a pre-trained model for determining the validity of the N-value, an index indicating the hardness and compaction of the ground. Geological survey data includes N-values ​​measured at predetermined depth intervals. These N-values ​​often contain noise, such as a sharp increase when encountering boulders (larger stones) within the ground. However, attempting to remove this noise using a uniform threshold carries the risk of mistakenly excluding truly hard ground (bearing layers).

[0029] Therefore, in this embodiment, the N-value judgment model M1 is trained to learn the waveform pattern formed by the data sequence of N-values ​​along the depth direction from a large number of training data points. Based on the continuity with the N-values ​​at the upper and lower depths, it is made possible to distinguish between "noise-specific waveform patterns" and "true support layer waveform patterns." As a result, as shown in Figure 3, when the N-value judgment model M1 is input at each depth, it outputs a judgment result indicating whether or not each N-value is noise.

[0030] The corrected N-value prediction model M2 is a trained model that predicts a corrected N-value (a reasonable N-value that should have been observed) by correcting for an inappropriate N-value. In this embodiment, the corrected N-value prediction model M2 has learned the relationship between depth / soil layer classification and N-value from a large number of training data, which has been judged as reasonable by skilled engineers. Therefore, as shown in Figure 3, the corrected N-value prediction model M2 outputs a corrected N-value when it receives the depth and soil layer classification of an N-value that has been judged as inappropriate by the N-value determination unit 42, which will be described later.

[0031] The coefficient prediction model M3 is a pre-trained model that predicts the coefficients used to calculate ground parameters. Ground parameters are numerical data used in design analysis and mainly include design N-value, unit weight (γ), cohesion (C), internal friction angle (φ), deformation modulus (E), etc. In geological survey reports, each ground parameter is calculated for each soil layer classification using general calculation formulas that utilize coefficients.

[0032] For example, the cohesion (C) = 6.25·(N value), the internal friction angle (φ) = √(20·(N value)) + 15, and the deformation modulus (E) = 700·(N value), where the numerical part multiplied by the N value is the coefficient. However, the properties of the ground vary greatly depending on the survey area and soil layer classification. Therefore, in order to calculate more realistic ground constants, it is necessary for skilled engineers to set coefficients according to the survey area and soil layer classification.

[0033] Therefore, in this embodiment, the coefficient prediction model M3 has learned the relationship between region / soil layer classification and coefficients from a large number of training data, which has been judged as valid by skilled engineers. As a result, the coefficient prediction model M3 has the knowledge that "this coefficient is optimal for this soil layer in this area," and as shown in Figure 3, when the survey area and soil layer classification included in the geological survey data are input, it outputs a coefficient suitable for that region and soil layer.

[0034] The M4 soil parameter determination model is used to determine the validity of soil parameters (theoretical values) calculated using general formulas. As mentioned above, various soil parameters are calculated for each soil layer, but the final decision on whether to adopt them as is is made by skilled engineers, taking into account whether they are consistent with the properties of the soil layer in question and whether they are natural in relation to the upper and lower soil layers.

[0035] Therefore, in this embodiment, the ground constant determination model M4 has learned the relationship between columnar section information and each ground constant from a large number of training data, which has been judged as valid by skilled engineers. As a result, as shown in Figure 3, when the calculated ground constant and columnar section information are input to the ground constant determination model M4, it refers to them as contextual information. Then, in that context, it determines whether the input ground constant is valid or not (whether it is consistent with the properties of the soil layer, whether it is unnatural in relation to the upper and lower soil layers, etc.) based on criteria learned from the final judgment tendencies of skilled engineers, and outputs the determination result.

[0036] The corrected ground parameter prediction model M5 is a trained model that predicts corrected ground parameters after correcting for inappropriate ground parameters. In this embodiment, the corrected ground parameter prediction model M5 has been trained using a large number of training data to learn the relationship between columnar section information and each ground parameter, which has been judged as appropriate by skilled engineers. Therefore, as shown in Figure 3, the corrected ground parameter prediction model M5 outputs corrected ground parameters when it receives the N value and columnar section information used to calculate ground parameters that have been judged as inappropriate by the ground parameter determination unit 46 (described later).

[0037] In this embodiment, the following information can be used as the columnar section information to be input into the corrected ground parameter prediction model M5. • Local information • Soil layer classification at each depth (e.g., sandy soil, cohesive soil, gravelly soil, etc.) • Depth of soil layer boundary • Thickness of each soil layer • Continuity of soil layers within a borehole

[0038] Furthermore, in this embodiment, the corrected ground constant prediction model M5 is configured to output the correction direction (increase or decrease) and correction amount for the ground constant (theoretical value) and then calculate the corrected ground constant. However, the configuration is not limited to this, and the corrected ground constant may be output directly.

[0039] The integrated inference model M6 is an inference engine that comprehensively processes the output results of each trained model M1 to M5 to make a final decision. In this embodiment, the integrated inference model M6 is composed of an inference engine that has been fine-tuned based on a Large Language Model (LLM). The integrated inference model M6 then executes processes to evaluate the consistency of the output results of each trained model M1 to M5, to make a final determination (fine-tune) of the ground constants, and to create a geological survey report, according to prompts output from the survey report creation unit 48, which will be described later.

[0040] Specifically, the integrated inference model M6 evaluates the consistency of each output result by integrally processing the output results of the N-value determination model M1, the corrected N-value prediction model M2, the coefficient prediction model M3, the ground constant determination model M4, and the corrected ground constant prediction model M5 described above, along with the geological survey data. In this embodiment, as described later, the integrated inference model M6 evaluates the consistency of the ground constants determined to be valid by the ground constant determination unit 46 or the ground constants predicted by the corrected ground constant prediction unit 47.

[0041] In this evaluation, the integrated inference model M6 considers the consistency with other ground parameters, the determination result of the soil layer ultimately adopted as the bearing layer by skilled engineers, and the context of the entire columnar section information. As will be described later, if the ground parameter determination unit 46 and the corrected ground parameter prediction unit 47 are not provided, the integrated inference model M6 evaluates the consistency of the ground parameters calculated by the ground parameter calculation unit 45.

[0042] Based on the evaluation results, the integrated inference model M6 adopts the ground parameters that are consistent and uses them to create the geological survey report. On the other hand, it makes minor adjustments to the ground parameters that are inconsistent before creating the geological survey report. When making these adjustments, the integrated inference model M6 considers the consistency of the output results of each trained model M1 to M5, as well as past final judgment cases by skilled engineers, and inferencely generates correction proposals that improve consistency.

[0043] Furthermore, when creating a geological survey report, the integrated inference model M6 automatically generates a predetermined form by importing geological survey data, ground parameters, and on-site photographs, and also clearly indicates the soil layer that should be adopted as the supporting layer. In this embodiment, the integrated inference model M6 overlays a blackboard image containing the project name, hole number, construction details, etc., onto the on-site photographs included in the geological survey report.

[0044] The usage of the geological survey report creation support program 1a and each of the pre-trained models M1 to M6 is not limited to the above configuration. For example, the geological survey report creation support program 1a and each of the pre-trained models M1 to M6 may be stored on a computer-readable non-temporary storage medium such as a USB memory stick or CD-ROM, and then read and executed directly from that storage medium. Alternatively, the geological survey report creation support program 1a and each of the pre-trained models M1 to M6 may be packaged and made available as a downloadable application.

[0045] The learning data storage unit 33 stores learning data for fine-tuning (retraining) each trained model M1 to M6. In this embodiment, the learning data storage unit 33 stores input data, output results of each trained model M1 to M6, and final judgments by skilled engineers (ground constants, bearing layer, etc.) as learning data for each survey project. After a predetermined verification process, this data is used in batches to fine-tune each trained model M1 to M6. As a result, the internal parameters are fine-tuned in response to changes in regional characteristics and ground conditions, thus maintaining a continuously adaptable trained model.

[0046] The calculation processing unit 4 consists of a CPU (Central Processing Unit) and the like, and by executing the geological survey report creation support program 1a installed in the program storage unit 31, it functions as a geological survey data acquisition unit 41, an N-value determination unit 42, a corrected N-value prediction unit 43, a coefficient prediction unit 44, a ground constant calculation unit 45, a ground constant determination unit 46, a corrected ground constant prediction unit 47, and a geological survey report creation unit 48, as shown in Figure 1. Each component will be described in more detail below.

[0047] The geological survey data acquisition unit 41 is a functional unit that acquires geological survey data from the user terminal 10. In this embodiment, the geological survey data acquisition unit 41 acquires geological survey data transmitted from the user terminal 10 via the communication means 2 and provides it to other functional units. In this embodiment, the geological survey data acquisition unit 41 acquires geological survey data (raw data) directly entered on the geological survey report creation support site, but the system is not limited to this configuration, and columnar section data (in XML file format, etc.) created by existing columnar section creation software, etc., may also be acquired as geological survey data.

[0048] The N-value determination unit 42 is a functional unit that determines the validity of the N-value using the N-value determination model M1. In this embodiment, the N-value determination unit 42 determines whether the N-values ​​for each depth included in the geological survey data acquired by the geological survey data acquisition unit 41 are valid, using the N-value determination model M1 stored in the trained model storage unit 32. Specifically, the N-value determination unit 42 inputs the N-value at each depth to the N-value determination model M1 and determines the validity of the N-value based on the determination result output from the N-value determination model M1, which determines whether or not the N-value is noise.

[0049] The corrected N-value prediction unit 43 is a functional unit that predicts the corrected N-value using the corrected N-value prediction model M2. In this embodiment, the corrected N-value prediction unit 43 predicts the corrected N-value for N-values ​​that have been determined to be invalid by the N-value determination unit 42, using the corrected N-value prediction model M2 stored in the trained model storage unit 32. Specifically, the corrected N-value prediction unit 43 inputs the depth and soil layer classification of the N-value that has been determined to be invalid into the corrected N-value prediction model M2, and predicts the corrected N-value as the valid N-value (corrected N-value) that should have been observed at that depth and soil layer, which is output from the corrected N-value prediction model M2.

[0050] The coefficient prediction unit 44 predicts the coefficients to be used in calculating the ground constants using the coefficient prediction model M3. In this embodiment, the coefficient prediction unit 44 inputs the survey area and soil layer classification included in the geological survey data into the coefficient prediction model M3, and predicts the coefficients output from the coefficient prediction model M3 as coefficients suitable for the survey area and soil layer classification. Note that the coefficient prediction unit 44 is not an essential component; if reasonable ground constants can be obtained using the coefficients defined by the general formula, it is not necessary to provide the coefficient prediction unit 44 and the coefficient prediction model M3.

[0051] The ground constant calculation unit 45 is a functional unit that calculates ground constants using the N value. In this embodiment, the ground constant calculation unit 45 calculates various ground constants using the N value determined to be valid by the N value determination unit 42 or the corrected N value predicted by the corrected N value prediction unit 43 and the coefficient predicted by the coefficient prediction unit 44. That is, the ground constant calculation unit 45 modifies the calculation formula with the coefficient predicted by the coefficient prediction unit 44 and calculates the ground constants by substituting the N value into the calculation formula.

[0052] If the coefficient prediction unit 44 is not provided, the ground constant calculation unit 45 calculates various ground constants by substituting the N value determined to be appropriate by the N value determination unit 42 or the corrected N value predicted by the corrected N value prediction unit 43 into a general calculation formula or the like.

[0053] The ground constant determination unit 46 is a functional unit that determines the validity of ground constants using a ground constant determination model M4. In this embodiment, the ground constant determination unit 46 determines whether the ground constants calculated by the ground constant calculation unit 45 are valid using the ground constant determination model M4 stored in the learned model storage unit 32. Specifically, the ground constant determination unit 46 inputs the ground constants calculated by the ground constant calculation unit 45 and the columnar section information into the ground constant determination model M4, and determines the validity of the ground constants based on the determination result output from the ground constant determination model M4.

[0054] In this embodiment, the ground constant determination unit 46 performs preprocessing as necessary when determining the validity of the ground constants. For example, based on columnar section information, N-value distribution, soil layer classification, etc., it dynamically extracts soil layers that satisfy the conditions for generally being a supporting layer (for example, soil layers that exist below a certain depth and continuously show high N-values) as candidates. Then, the ground constants of the candidate soil layers are determined from the perspective of "whether the ground constants are reasonable values ​​if they were to be adopted as a supporting layer."

[0055] Furthermore, in this embodiment, the ground constant determination unit 46 is configured to individually input various ground constants into the ground constant determination model M4 and execute a determination process for each ground constant. Specifically, the ground constant determination unit 46 individually executes the design N value determination process, the unit volume weight determination process, the cohesion determination process, the internal friction angle determination process, and the deformation coefficient determination process, among others.

[0056] The corrected ground constant prediction unit 47 is a functional unit that predicts corrected ground constants using the corrected ground constant prediction model M5. In this embodiment, the corrected ground constant prediction unit 47 predicts corrected ground constants for ground constants that the ground constant determination unit 46 has determined to be invalid, using the corrected ground constant prediction model M5 stored in the learned model storage unit 32.

[0057] Specifically, the corrected ground parameter prediction unit 47 inputs the N-value and columnar section information used to calculate the ground parameters that the ground parameter determination unit 46 determined to be inappropriate into the corrected ground parameter prediction model M5, and predicts the corrected ground parameters output from the corrected ground parameter prediction model M5 as the corrected ground parameters.

[0058] For example, if the design N-value that was determined to be inappropriate is 30, and the result of that determination is "overestimated for the region and soil layer in question," the corrected ground parameter prediction model M5 will output "decrease (-)" as the correction direction and "5" as the correction amount, resulting in a corrected design N-value of 25. Also, the cohesion that was determined to be inappropriate is 20 (kN / m 2 If the result of that assessment is "underestimated as a supporting layer", the corrected ground parameter prediction model M5 will set the correction direction to "increase (+)" and the correction amount to "3 (kN / m 2 The output is ""), and the corrected adhesive strength is 23 (kN / m 2 )

[0059] While the ground parameter determination unit 46 and the corrected ground parameter prediction unit 47 improve the accuracy and automation level of the ground parameters, they are not essential components. If the ground parameter determination unit 46 and the corrected ground parameter prediction unit 47 are not provided, the ground parameter determination model M4 and the corrected ground parameter prediction model M5 also become unnecessary.

[0060] The geological survey report creation unit 48 is a functional unit that creates a geological survey report using an integrated inference model M6. In this embodiment, the geological survey report creation unit 48 inputs the output results of the N-value determination model M1, the corrected N-value prediction model M2, the coefficient prediction model M3, the ground constant determination model M4, and the corrected ground constant prediction model M5, along with the geological survey data, into the integrated inference model M6, and determines the consistency of the ground constants based on the consistency evaluation results output from the integrated inference model M6.

[0061] The geological survey report creation unit 48 adopts the ground parameters that are consistent based on the above evaluation results and uses the integrated inference model M6 to create the geological survey report. On the other hand, for ground parameters that are inconsistent, it makes fine adjustments considering the consistency of the output results of each trained model M1 to M5 and past final judgment cases by skilled engineers. Then, the integrated inference model M6 executes the process of creating the geological survey report using these finely adjusted ground parameters.

[0062] In this embodiment, if the geological survey report creation unit 48 determines that there are ground parameters that are valid by the ground parameter determination unit 46, it has the integrated inference model M6 evaluate the consistency of those ground parameters. On the other hand, if there are ground parameters predicted by the corrected ground parameter prediction unit 47, it has the integrated inference model M6 evaluate the consistency of those ground parameters. Furthermore, if the ground parameter determination unit 46 and the corrected ground parameter prediction unit 47 are not provided, the geological survey report creation unit 48 has the integrated inference model M6 determine the consistency of the ground parameters calculated by the ground parameter calculation unit 45.

[0063] Furthermore, in this embodiment, if the geological survey report creation unit 48 includes a field map in the geological survey report, it may obtain a URL (Uniform Resource Locator) of the Geospatial Information Authority of Japan map or similar from the user terminal 10. This allows the position of the field map to be adjusted as needed, and the adjusted field map to be easily obtained.

[0064] Next, the operation of the geological survey report creation support server 1, geological survey report creation support system 100, geological survey report creation support program 1a, and geological survey report creation support method of this embodiment will be described.

[0065] When creating a geological survey report using the geological survey report creation support server 1, geological survey report creation support program 1a, geological survey report creation support method, and geological survey report creation support system 100 of this embodiment, first, the user terminal 10 accesses the geological survey report creation support site provided by the geological survey report creation support server 1. Then, various geological survey data is input from the geological survey report creation support site and transmitted to the geological survey report creation support server 1.

[0066] When geological survey data is transmitted from the user terminal 10, the geological survey data acquisition unit 41 acquires the geological survey data from the user terminal 10, as shown in Figure 4 (Step S1: Geological survey data acquisition step). Next, the N-value determination unit 42 inputs the N-values ​​for each depth included in the geological survey data acquired in step S1 into the N-value determination model M1 and determines the validity of each N-value (Step S2: N-value determination step). This allows for highly accurate determination of whether the N-value is noise such as boulders, based on the waveform pattern formed by the data sequence of N-values ​​along the depth direction.

[0067] If the result of the judgment in step S2 is determined to be invalid (step S2: NO), the corrected N-value prediction unit 43 inputs the depth and soil layer classification of the N-value determined to be invalid into the corrected N-value prediction model M2, and predicts the corrected N-value output from this corrected N-value prediction model M2 as the corrected N-value (step S3: corrected N-value prediction step). This predicts the valid N-value that should have been observed according to the depth and soil layer classification. On the other hand, if all N-values ​​are determined to be valid in step S2 (step S2: YES), the process in step S3 is skipped.

[0068] Next, the coefficient prediction unit 44 inputs the survey area and soil layer classification included in the geological survey data into the coefficient prediction model M3 and predicts the coefficients to be used in calculating the ground constants (step S4: coefficient prediction step). As a result, coefficients suitable for the survey area and soil layer classification are predicted, and the resulting ground constants are closer to reflecting actual ground characteristics compared to the ground constants (theoretical values) calculated by general calculation formulas.

[0069] Next, the ground constant calculation unit 45 calculates various ground constants using the N value determined to be valid in step S2 or the corrected N value predicted in step S3, and the coefficient predicted in step S4 (step S5: ground constant calculation step). This allows for the calculation of more accurate ground constants by comparing the N value included in the geological survey data with the ground constants (theoretical values) calculated using general calculation formulas.

[0070] Next, the ground constant determination unit 46 determines the validity of each ground constant calculated in step S5 using the ground constant determination model M4 (step S6: ground constant determination step). This allows for a highly accurate determination of the validity of the ground constants, such as whether they are consistent with the properties of the soil layer and whether they are unnatural in relation to the upper and lower soil layers, based on the final judgment tendencies of skilled engineers.

[0071] If the determination in step S6 is found to be inappropriate (step S6: NO), the corrected ground constant prediction unit 47 predicts the corrected ground constant using the corrected ground constant prediction model M5 (step S7: corrected ground constant prediction step). This ensures that inappropriate ground constants are not adopted as is, and predicts ground constants that are consistent with the properties of the soil layer and natural in relation to the upper and lower soil layers. On the other hand, if all ground constants are found to be appropriate in step S6 (step S6: YES), the process in step S7 is skipped.

[0072] Next, the geological survey report creation unit 48 inputs the output results of the aforementioned trained models M1 to M5 and the geological survey data into the integrated inference model M6, and determines the consistency of the ground parameters based on the consistency evaluation results output from this integrated inference model M6 (Step S8). This determines the consistency considering the consistency with other ground parameters, the determination results of the soil layer ultimately adopted as the bearing layer by the skilled engineer, and the context of the entire columnar section information.

[0073] If the judgment in step S8 determines that there is no consistency (step S8: NO), the integrated inference model M6 fine-tunes the ground parameters (step S9). As a result, the final adopted ground parameters take into account the consistency of the output results of each trained model M1 to M5, as well as past final judgment cases by skilled engineers, improving accuracy and reliability. On the other hand, if the ground parameters are determined to be consistent (step S8: YES), the process in step S9 is skipped.

[0074] Next, the geological survey report creation unit 48 creates a geological survey report using the ground constants determined to be consistent in step S8, or the ground constants fine-tuned in step S9, along with other geological survey data (step S10). This allows for the automatic creation of highly accurate and reliable geological survey reports that reflect the empirical rules and tacit knowledge of skilled engineers, easily and quickly, even without experience or expertise in geological surveys. Steps S8 to S10 described above correspond to the geological survey report creation steps.

[0075] The created geological survey report is transmitted to the user terminal 10 upon request from the user terminal 10 (step S11: geological survey report transmission step) and used as appropriate. In this embodiment, when a geological survey report is created, the input data, the output results of each trained model M1 to M6, and the final judgment by the skilled engineer (ground constants, bearing layer, etc.) are stored as training data in the training data storage unit 33 for each survey case. Then, using this stored training data, each trained model M1 to M6 is retrained (fine-tuned) periodically or at predetermined intervals. As a result, each trained model M1 to M6 can continuously learn about changes in regional characteristics and ground tendencies to maintain adaptability and autonomously improve its accuracy and reliability.

[0076] Finally, if a request to create a new geological survey report is received from user terminal 10 (step S12: YES), the process returns to step S1 and the above-described process is repeated. On the other hand, if there is no request to create a new geological survey report from user terminal 10 (step S12: NO), this flowchart is terminated.

[0077] The geological survey report creation support server 1, geological survey report creation support system 100, geological survey report creation support program 1a, and geological survey report creation support method according to the present invention, as described above, provide the following effects. 1. Even without experience or expertise in geological surveys, it is possible to easily and quickly automatically generate highly accurate and reliable geological survey reports that reflect the experience and tacit knowledge of skilled engineers. 2. It is possible to predict coefficients suitable for the survey area and soil layer classification, and calculate more accurate ground parameters. 3. It is possible to determine the validity of the ground parameters, predict appropriate ground parameters according to the survey area and soil layer classification, and determine the consistency of the ground parameters with higher accuracy. 4. It is possible to determine with high accuracy whether the N-value of geological survey data is noise such as boulders. 5. It is possible to predict a reasonable N-value according to the depth and soil layer classification. 6. The output results of each trained model can be integrated to make a final decision that reflects the experience and tacit knowledge of skilled engineers. 7. By retraining (fine-tuning) each trained model, it is possible to continuously adapt to changes in regional characteristics and ground conditions, thereby autonomously improving accuracy and reliability.

[0078] Furthermore, the geological survey report creation support server 1, geological survey report creation support system 100, geological survey report creation support program 1a, and geological survey report creation support method according to the present invention are not limited to the embodiments described above and can be modified as appropriate.

[0079] For example, in the embodiment described above, the geological survey report creation support program 1a is provided as a web application from the geological survey report creation support server 1, but the configuration is not limited to this, and it may also be installed on the user terminal 10. In this case, all of the above-described functional units are implemented on the user terminal 10, so it can be used in a standalone state.

[0080] Furthermore, in the embodiment described above, the coefficient prediction model M3, the ground parameter determination model M4, and the corrected ground parameter prediction model M5 are each configured as a single trained model and are processed individually by inputting the type of ground parameter as identification information. However, the configuration is not limited to this. In other words, a completely separate and independent trained model may be provided for each type of ground parameter.

[0081] Furthermore, while the above-described embodiment exemplified a finely tuned large-scale language model as the integrated inference model M6, it is not dependent on any particular model or vendor. The integrated inference model M6 may be a large-scale language model with a different architecture, or another type of inference engine capable of deriving logical conclusions from multiple sources. [Explanation of symbols]

[0082] 1. Geological survey report creation support server 1a Geological Survey Report Preparation Support Program 2. Means of communication 3 Memory means 31 Program Storage Unit 32. Pre-trained model memory 33. Learning data storage unit 4. Calculation processing means 41 Geological Survey Data Acquisition Department 42 N-value determination unit 43 Corrected N-value prediction section 44 Coefficient prediction unit 45. Ground Constant Calculation Section 46 Ground constant determination section 47 Corrected Ground Constant Prediction Section 48. Department for Preparing Investigation Reports 10 User terminals 100 Geological Survey Report Preparation Support System M1 N-value judgment model M2-corrected N-value prediction model M3 coefficient prediction model M4 Ground Constant Determination Model M5 Corrected Ground Temperature Prediction Model M6 Integrated Inference Model

Claims

1. A geological survey report creation support server for assisting in the creation of geological survey reports, A geological survey data acquisition unit that acquires geological survey data from the user terminal, An N-value determination unit that determines whether the N-values ​​for each depth included in the geological survey data are valid using an N-value determination model, For N values ​​that the N value determination unit has determined to be invalid, the corrected N value prediction unit predicts the corrected N value using a corrected N value prediction model, A ground constant calculation unit calculates a ground constant using the N value determined to be valid by the N value determination unit or the N value predicted by the corrected N value prediction unit, A geological survey report creation unit, using an integrated inference model, evaluates the consistency of the ground parameters calculated by the ground parameter calculation unit, and performs fine-tuning on any inconsistent ground parameters before creating a geological survey report. A server that supports the creation of geological survey reports.

2. The system includes a coefficient prediction unit that predicts coefficients used to calculate ground parameters based on the survey area and soil layer classification included in the aforementioned geological survey data, using a coefficient prediction model. The ground constant calculation unit calculates the ground constant using the N value determined to be valid by the N value determination unit or the N value predicted by the corrected N value prediction unit and the coefficient predicted by the coefficient prediction unit. A server for supporting the creation of geological survey reports as described in claim 1.

3. A ground constant determination unit determines whether the ground constant calculated by the ground constant calculation unit is appropriate using a ground constant determination model, For ground parameters that the ground parameter determination unit has determined to be inappropriate, the corrected ground parameter prediction unit predicts the corrected ground parameters using a corrected ground parameter prediction model, It has, The geological survey report preparation unit determines, using the integrated inference model, the consistency between the ground parameters determined to be valid by the ground parameter determination unit or the ground parameters predicted by the corrected ground parameter prediction unit. A geological survey report preparation support server according to claim 1 or claim 2.

4. The aforementioned N-value determination model is machine-trained to take an N-value at each depth as input and output a determination result indicating whether or not each N-value is noise. The N-value determination unit determines the validity of the N-value based on the determination result. A server for supporting the creation of geological survey reports as described in claim 1.

5. The corrected N-value prediction model is machine-trained to output a corrected N-value when it receives the depth and soil layer classification of an N-value that has been determined to be invalid by the N-value determination unit. The corrected N-value prediction unit predicts the corrected N-value as the corrected N-value. A server for supporting the creation of geological survey reports as described in claim 1.

6. The geological survey report creation support server according to claim 3, which is dependent on claim 2, wherein the integrated inference model determines the consistency of each output result by integrally processing the output results of the N-value determination model, the corrected N-value prediction model, the coefficient prediction model, the ground constant determination model, and the corrected ground constant prediction model with the geological survey data.

7. The geological survey report creation support server described in claim 1, The user terminal and, A geological survey report creation support system.

8. A geological survey report preparation support program to assist in the preparation of geological survey reports, A geological survey data acquisition unit that acquires geological survey data from the user terminal, An N-value determination unit that determines whether the N-values ​​for each depth included in the geological survey data are valid using an N-value determination model, For N values ​​that the N value determination unit has determined to be invalid, the corrected N value prediction unit predicts the corrected N value using a corrected N value prediction model, A ground constant calculation unit calculates a ground constant using the N value determined to be valid by the N value determination unit or the N value predicted by the corrected N value prediction unit, The geological survey report creation unit uses an integrated inference model to evaluate the consistency of the ground parameters calculated by the ground parameter calculation unit, and to fine-tune any inconsistent ground parameters before creating a geological survey report. A program that assists in creating geological survey reports by enabling computers to function as such.

9. A method for supporting the preparation of geological survey reports, A geological survey data acquisition step involves acquiring geological survey data from the user terminal, An N-value determination step in which an N-value determination model is used to determine whether the N-values ​​for each depth included in the geological survey data are valid, For N values ​​that were determined to be invalid in the N value determination step, a corrected N value prediction step is performed to predict the corrected N value using a corrected N value prediction model. A ground constant calculation step which calculates a ground constant using the N value determined to be valid in the N value determination step or the N value predicted in the corrected N value prediction step, A geological survey report creation step is performed by an integrated inference model to evaluate the consistency of the ground parameters calculated in the ground parameter calculation step, and to fine-tune any inconsistent ground parameters before creating a geological survey report. A method for supporting the creation of geological survey reports, which includes the following features.

Images