JP7347887B1 - Learning device, estimation device, estimation system, learning method, estimation method and program - Google Patents

Abstract

[Problem] To shorten the period required for soil cement mixing tests. [Solution] A learning control unit calculates the amount of solidified material and water content for each of multiple types of soil cement mixing cases, and the first period strength which is the strength after the first period of soil cement for each of the multiple types of mixing cases. The intensity estimation model is learned by machine learning using teacher data indicating the intensity of the second period and the second period intensity that is the intensity after the elapse of the second period, which is longer than the first period. The estimating unit inputs, into the strength estimation model, the amount of solidified material and water content of each of the plurality of combination cases of soil cement to be estimated for the second period strength, and the first period strength of one or more combination cases. The output of the second period strength of the soil cement is obtained for each of a plurality of combination cases of the solidified material amount and water content. [Selection diagram] Figure 12

Description

The present invention relates to a learning device, an estimation device, an estimation system, a learning method, an estimation method, and a program.

Landslides are occurring frequently due to abnormal weather conditions around the world. In recent years, in place of the conventionally constructed concrete sabo dams (sabo dams), sabo dams made of sabo soil cement, which is made from local earth and sand and is strongly consolidated, have been increasingly used at these disaster recovery sites. .

This construction method differs from the soil-cement construction method, which is conventionally used for ground improvement work, and is a construction method that uses earth and sand washed away by debris flows, etc., and soil leftover from construction work as the main materials. In the sabo soil-cement method, it is extremely important to carry out an appropriate mix design, such as the type and amount of solidifying agent (cement, etc.) and moisture content settings, in accordance with the material properties, in order to meet the specified strength. In order to perform this formulation design, a formulation test is conducted.

The outline of a typical sabo soil cement mixing test is as follows.
(1) After collecting 500kg to 600kg of earth and sand on site and conducting a soil test according to JIS standards, basically nine types of combination cases will be considered.
(2) In the testing room, create 54 specimens in total, 6 specimens for each of the 9 combination cases.
(3) For each combination case, measure the 7-day strength of the three specimens and the 28-day strength of the three specimens, analyze the results, and determine the appropriate formulation.
(4) If the results obtained in (3) do not satisfy the required quality, consider the combination case again, perform additional tests from (2), analyze the results, and determine an appropriate formulation.

On the other hand, there is a technique for early determining that the strength of sabo soil cement is equal to or higher than a target strength (for example, see Patent Document 1).

JP 2021-9062 Publication

As mentioned above, in the blending test, additional tests are repeated until test results that satisfy the required quality are obtained. However, one mixing test requires at least 28 days to obtain strength, and if there is insufficient soil for the test, soil must be collected from the site again. Further, in the technique of Patent Document 1, the strength is measured after steam curing the specimen at a predetermined curing temperature for a predetermined time. Therefore, it was necessary to carry out the test by transporting the soil to a facility where the specimen could be steam-cured.

The present invention has been made in view of the above-mentioned circumstances, and provides a technique that makes it possible to shorten the period required for soil cement mixing tests.

One aspect of the present invention is to determine the amount of solidified material and water content of each of the plurality of types of mixing cases of soil cement, and the first period strength which is the strength after the first period of the soil cement of each of the plurality of types of mixing cases. Using teacher data indicating the second period strength which is the strength after a second period which is longer than the first period, the solidified material amount and water content of each of the plurality of combination cases and one or more of the combination cases are determined. Using machine learning, a strength estimation model is created that takes as input the first period strength of soil cement and outputs the second period strength of soil cement for each of the plurality of combination cases of the input solidification material amount and water content. This is a learning device including a learning control section that performs learning.

One aspect of the present invention is the learning device described above, wherein the teacher data further includes information on the type of solidifying material, and the learning control unit is configured to learn the type of solidifying material and the solidifying material for each of the plurality of types of combination cases. amount, water content, and the first period strength of the soil cement of one or more of the mixing cases, and the second period strength of the soil cement of each of the plurality of mixing cases of the input solidified material amount and the water content. An intensity estimation model that outputs period intensity is learned by machine learning.

One aspect of the present invention is to determine the amount of solidified material and water content of each of the plurality of types of mixing cases of soil cement, and the first period strength which is the strength after the first period of the soil cement of each of the plurality of types of mixing cases. The intensity estimation model trained using training data indicating the second period intensity, which is the intensity after the elapse of the second period which is longer than the first period, is applied to each of the plurality of combination cases for which the second period intensity is to be estimated. By inputting the amount of solidified material and water content, and the first period strength of soil cement of one or more combination cases among the plurality of combination cases for which the second period strength is estimated, the amount of solidified material input is determined. and an estimation unit that obtains an output of second period strength of soil cement for each of the plurality of combination cases of the water content.

One aspect of the present invention is the above-mentioned estimating device, wherein the teacher data further includes information on the type of solidifying material, and the estimating section adds the type of solidifying material and the second period strength to the strength estimation model. Input the amount of solidified material and water content of each of the plurality of combination cases to be estimated, and the first period strength of soil cement for one or more of the plurality of combination cases to be estimated for the second period strength. As a result, the output of the second period strength of the soil cement for each of the plurality of combination cases of the input solidified material amount and the water content is obtained.

One aspect of the present invention is to determine the amount of solidified material and water content of each of the plurality of types of mixing cases of soil cement, and the first period strength which is the strength after the first period of the soil cement of each of the plurality of types of mixing cases. Using teacher data indicating the second period strength which is the strength after a second period which is longer than the first period, the solidified material amount and water content of each of the plurality of combination cases and one or more of the combination cases are determined. Using machine learning, a strength estimation model is created that takes as input the first period strength of soil cement and outputs the second period strength of soil cement for each of the plurality of combination cases of the input solidification material amount and water content. A learning control unit for learning, the strength estimation model, the amount of solidified material and water content of each of the plurality of combination cases targeted for second period strength estimation, and the amount of solidified material and water content of each of the plurality of combination cases targeted for second period strength estimation. By inputting the first period strength of soil cement for one or more combination cases, the output of the second period strength of soil cement for each of the plurality of combination cases of the input solidified material amount and the water content is output. An estimating system includes an estimating unit that obtains the information.

One aspect of the present invention is to determine the amount of solidified material and water content of each of the plurality of types of mixing cases of soil cement, and the first period strength which is the strength after the first period of the soil cement of each of the plurality of types of mixing cases. Using teacher data indicating the second period strength which is the strength after a second period which is longer than the first period, the solidified material amount and water content of each of the plurality of combination cases and one or more of the combination cases are determined. Using machine learning, a strength estimation model is created that takes as input the first period strength of soil cement and outputs the second period strength of soil cement for each of the plurality of combination cases of the input solidification material amount and water content. This is a learning method having a learning step for learning.

One aspect of the present invention is to determine the amount of solidified material and water content of each of the plurality of types of mixing cases of soil cement, and the first period strength which is the strength after the first period of the soil cement of each of the plurality of types of mixing cases. The intensity estimation model trained using training data indicating the second period intensity, which is the intensity after the elapse of the second period which is longer than the first period, is applied to each of the plurality of combination cases for which the second period intensity is to be estimated. By inputting the amount of solidified material and water content, and the first period strength of soil cement of one or more combination cases among the plurality of combination cases for which the second period strength is estimated, the amount of solidified material input is determined. and an estimating step of obtaining an output of second period strength of soil cement for each of the plurality of combination cases of the water content.

One aspect of the present invention is a program for causing a computer to function as the above-mentioned learning device.

One aspect of the present invention is a program for causing a computer to function as the above estimation device.

According to the present invention, it is possible to shorten the period required for soil cement mixing tests.

It is a figure which shows the example of the combination case of the object of a combination test. FIG. 2 is a flow diagram showing the procedure for conducting a blending test. FIG. 3 is a diagram showing how a blending test is carried out. FIG. 3 is a diagram showing how a blending test is carried out. FIG. 3 is a diagram showing how a blending test is carried out. FIG. 3 is a diagram showing how a blending test is carried out. FIG. 3 is a diagram showing how a blending test is carried out. FIG. 3 is a diagram showing how a blending test is carried out. FIG. 3 is a diagram showing how a blending test is carried out. It is a figure which shows the example of strength distribution by the difference in the water content ratio of earth and sand and sabo soil cement. It is a figure which shows the example of the relationship between the water content ratio and 28-day strength of sabo soil cement. FIG. 1 is a schematic block diagram showing the system configuration of an estimation system according to an embodiment. FIG. 2 is a schematic block diagram showing a specific example of the functional configuration of a terminal device according to an embodiment. It is a figure showing an example of teacher data of an embodiment. FIG. 2 is a schematic block diagram showing a specific example of the functional configuration of the learning device according to the embodiment. It is a flowchart which shows a specific example of processing of the learning device of an embodiment. FIG. 2 is a schematic block diagram showing a specific example of the functional configuration of the estimation device according to the embodiment. It is a flow chart which shows a specific example of processing of an estimation device of an embodiment. It is a figure showing an example of a screen displayed on a terminal device of an embodiment. 1 is a diagram schematically showing an example of a hardware configuration of an information processing apparatus applied to this embodiment. It is a schematic block diagram showing a modification of the functional configuration of the estimation device of the embodiment. It is a schematic block diagram showing a modification of the functional configuration of the estimation device of the embodiment. It is a figure showing the estimation result using the estimation system of an embodiment.

Embodiments of the present invention will be described in detail below with reference to the drawings. First, we will explain the procedure for carrying out a blending test for sabo soil cement. Sabo soil cement is an example of soil cement. Here, a case will be described using as an example a case where nine types of combination cases are considered after conducting a soil test according to the JIS standard for soil and sand to be tested.

FIG. 1 shows an example of a combination case to be subjected to a combination test. The same type of solidifying agent is used in the nine combination cases shown in FIG. In the following, a case where the solidifying material is cement will be explained as an example. First, three cement cases, CASE1, CASE2, and CASE3, are set. The unit amounts of cement in CASE1, CASE2, and CASE3 are C1, C2, and C3, respectively. Then, three types of water content ratios Wn1, Wn2, and Wn3 are set for each cement case CASEn (n=1, 2, and 3), and are respectively set as CASEn-1, CASEn-2, and CASEn-3. As a result, a total of nine combination cases CASE1-1 to CASE3-3 are set. The unit amount of cement is C1<C2<C3, and the water content ratio is W11<W12<W13, W21<W22<W23, and W31<W32<W33. Although C2-C1=C3-C2 is often used, it is not limited to this. In addition, although W12-W11=W13-W12=W22-W21=W23-W22=W32-W31=W33-W32 is often set, it is not limited to this. In the blending test, the 7-day strength and 28-day strength of sabo soil cement for each of these nine blending cases are generally measured.

FIG. 2 is a flow diagram showing the procedure of the blending test. Conventionally, steps S1 to S9 are performed for each of the nine combination cases. Further, FIGS. 3 to 9 are diagrams showing how each procedure shown in FIG. 2 is implemented.

First, the test person confirms the mixing amount of the sabo soil cement materials to be tested (step S1, FIG. 3). The materials are the soil sample to be tested, cement, and added water. The mixing amount of these materials is determined based on the type of cement, unit cement amount, and water content ratio of each mixing case shown in FIG.

Next, the test person puts the sample and cement into a mixer for each combination case and performs dry kneading for a specified time (Step S2, FIG. 4). Subsequently, the test person puts added water into the mixer and performs main kneading for a specified time (step S3, FIG. 5). After mixing the main kneading, the test person confirms the mixing state visually and with a feeler (step S4, FIG. 6). The test person measures the water content ratio of the sample after mixing in order to confirm the consistency between the test water content ratio and the post-mixing water content ratio (step S5, FIGS. 7(a) and 7(b)). Next, the tester measures the mixed sample using a slump cone (step S6, FIG. 8).

Thereafter, the test person prepares three specimens for the 7-day compressive strength test and three specimens for the 28-day compressive strength test for each of the nine combination cases (step S7 ). In the production of these specimens, the mixed sample is put into the mold, and then the sample in the mold is compacted by compacting with a push rod, compacting with a vibrator, etc. (FIG. 9(a)). In addition, sample input and compaction are performed in several steps. After the specimen is prepared (FIG. 9(b)), the top end is capped with mortar or the like, if necessary. The strength of sabo soil cement is affected by unit cement amount, water content ratio, curing temperature, humidity control, etc., so these factors are kept constant in each mixing test.

On the 7th day after the preparation of the specimens, the test person measures the compressive strength of the three specimens for the 7-day compressive strength test of each combination case to obtain the 7-day strength (step S8). Furthermore, the tester measures the compressive strength of the three 28-day age compressive strength test specimens of each combination case on the 28th day after creating the specimen to obtain the 28-day strength (step S9).

The purpose of the blending test was to confirm the 28-day strength in each blending case. However, sabo soil cement is intended to harden through cement hydration reaction (chemical reaction), and the increase from 7-day strength to 28-day strength is also an important factor when determining the specific formulation (mixture that satisfies required quality). Since this is data, the 7-day strength of the specimen was also measured. In addition, for each combination case, three specimens were prepared for the 7-day age compressive strength test and three specimens for the 28-day age compressive strength test, but this did not result in abnormal values being obtained. This is to make it possible to detect this in case of a problem, and the number is not limited to three.

FIG. 10 is a diagram showing an example of strength distribution due to the difference in water content ratio between earth and sand and sabo soil cement. Curve A0 shows the relationship between soil moisture content and compaction strength, and curves A1, A2, and A3 are quadratic curves calculated from the 28-day strength against the moisture content ratio of sabo soil cement with the same cement type and unit cement amount. FIG. Curve A0 is obtained, for example, by material testing. w1, w2, w3 are the water content ratios of sabo soil cement for CASEn-1, CASEn-2, and CASEn-3, respectively, and C1, C2, and C3 are the 28-day strength for CASEn-1, CASEn-2, and CASEn-3, respectively. It is. Further, P is the peak intensity and B is the minimum control intensity. The minimum controlled strength is the strength that can ensure that the material is cured by a hydration reaction.

FIG. 10(a) shows the measurement results of the 28-day intensity when the intensity peak P could be captured in CASEn-1 to CASEn-n-3. In this case, there is a peak P on the curve A1. FIG. 10(b) shows the measurement results when the amount of water added exceeds the intensity peak P in CASEn-1 to n-3. Curve A2 decreases as the water content increases. FIG. 10(c) shows the measurement results when the amount of water added was below the intensity peak in CASEn-1 to n-3. Curve A3 decreases as the water content increases. As shown in Figures 10(b) and 10(c), if the peak P cannot be captured from the 28-day strength results, the formulation may be reconsidered and a blending test is conducted to capture the peak P. .

FIG. 11 is a diagram showing an example of the relationship between the water content ratio and the 28-day strength of sabo soil cement obtained from each cement case of CASE 1, 2, and 3. Curve A0 shows the relationship between the water content of soil and compaction strength, and curves D1, D2, and D3 show the relationship between the water content ratio and 28-day strength of sabo soil cement obtained from CASE 1, 2, and 3, respectively. α1, α2, and α3 indicate the peak strength water content ratio of sabo soil cement of CASE 1, 2, and 3, respectively. Figure 11(a) shows the case where the peak strength of all three cement cases exceeds the minimum control strength, and Figure 11(b) shows the case where the peak strength of the two cement cases exceeds the minimum control strength. be. In these cases, the indicated formulation can be determined based on two or more peak intensities (eg, peak intensities of 2 N/mm ² or more). On the other hand, Fig. 11(c) shows the case where the peak strength of one cement case exceeds the minimum control strength, and Fig. 11(d) shows the case where the peak strength of all three cement cases does not exceed the minimum control strength. This is the case. In these cases, the combination case may be reconsidered and a combination test may be performed to obtain two or more peak intensities. Conventionally, repeated blending tests increased the testing period and even increased costs.

Next, the estimation system of this embodiment will be explained. FIG. 12 is a schematic block diagram showing the system configuration of the estimation system 100 according to an embodiment of the present invention. Estimation system 100 includes a terminal device 10, a learning device 20, and an estimation device 30. The terminal device 10 and the estimation device 30 are communicably connected via a network 70. The learning device 20 and the estimation device 30 may be communicably connected via the network 70. The network 70 may be a network using wireless communication or a network using wired communication. The network 70 may be configured using, for example, the Internet or a local area network (LAN). The network 70 may be configured by combining a plurality of networks.

The learning device 20 performs machine learning on the strength estimation model using blending test data indicating the results of blending tests conducted in the past as teacher data according to the procedure shown in FIG. The strength estimation model takes as input the type of cement, the unit cement amount (amount of solidified material) and water content for each of the nine mixing cases, and the 7-day strength of sabo soil cement for one or more mixing cases. The 28-day strength of sabo soil cement is output for each of nine combination cases of unit cement and water content.

When conducting a mixture test on new soil, the tester examines 9 types of mixture cases for the soil and sand, and tests the 7-day strength of the sabo soil cement of at least one combination case according to the procedure shown in Figure 2. Measure. The estimation device 30 adds the type of cement, the unit cement amount, and water content of each of the nine types of blending cases studied to the strength estimation model after learning by the learning device 20 (hereinafter also referred to as the “trained model”). By inputting the ratio and the 7-day strength information (hereinafter referred to as "input information") of one to nine types of mixture cases measured among the mixture cases considered, the input unit cement and its content can be calculated. Estimated values of the 28-day strength of sabo soil cement for each of nine combination cases of water amount are obtained as output.

FIG. 13 is a schematic block diagram showing a specific example of the functional configuration of the terminal device 10. As shown in FIG. The terminal device 10 is configured using an information device such as a smartphone, a tablet, a personal computer, or a dedicated device. The terminal device 10 includes a communication section 11, an input section 12, an output section 13, a storage section 14, and a control section 15.

The communication unit 11 is a communication device. The communication unit 11 may be configured as a network interface, for example. The communication unit 11 performs data communication with other devices via the network 70 under the control of the control unit 15 . The communication unit 11 may be a device that performs wireless communication or may be a device that performs wired communication.

The input unit 12 is configured using existing input devices such as a keyboard, pointing device (mouse, tablet, etc.), buttons, touch panel, etc. The input unit 12 is operated by the user when inputting the user's instructions into the terminal device 10 . The input unit 12 may be an interface for connecting an input device to the terminal device 10. In this case, the input unit 12 inputs to the terminal device 10 an input signal generated in response to a user's input on the input device.

The output unit 13 outputs information in a form that can be recognized by the user. The output unit 13 may be an image display device such as a liquid crystal display or an organic EL (Electro Luminescence) display. The output unit 13 may be an interface for connecting the image display device to the terminal device 10. In this case, the output unit 13 generates a video signal for displaying image data, and outputs the video signal to an image display device connected to itself. The output unit 13 may be a device that outputs sound, such as a speaker. The output unit 13 may be an interface for connecting a sound output device such as a speaker or headphones to the terminal device 10. In this case, the output unit 13 generates an audio signal for reproducing the audio data, and outputs the audio signal to the audio output device connected to itself. Note that the output section 13 may be configured as a touch panel integrated with the input section 12.

The storage unit 14 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 14 stores data used by the control unit 15. The storage unit 14 stores data necessary for the control unit 15 to perform processing.

The control unit 15 is configured using a processor such as a CPU (Central Processing Unit) and a memory (main storage device). The control unit 15 functions when a processor executes a program. Note that all or part of each function of the control unit 15 may be realized using hardware such as an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array). The above program may be recorded on a computer-readable recording medium. Computer-readable recording media include portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, semiconductor storage devices (such as SSDs: Solid State Drives), and hard disks and semiconductor storages built into computer systems. It is a storage device such as a device. The above program may be transmitted via a telecommunications line.

The control unit 15 may execute an application installed on its own device (terminal device 10), for example. A specific example of such an application is an application provided to the terminal device 10 as a dedicated application for the estimation system 100. Another specific example of such an application is a WEB (web) browser application. Such an application may be installed in the terminal device 10 in advance, or may be downloaded each time the estimation process is executed. For example, when implemented as a WEB browser application, when the terminal device 10 connects to a specific WEB server, a device specified by the WEB server (for example, the WEB server itself or another server) The terminal device 10 may download and execute the application. The control unit 15 operates according to the program of the application being executed.

The control unit 15 controls the terminal device 10 according to user operations and information received from the estimation device 30. For example, the control unit 15 uses the communication unit 11 to transmit information input by the user by operating the input unit 12 to the estimation device 30. For example, when information transmitted from the estimation device 30 is received by the communication unit 11 via the network 70, the control unit 15 generates screen data based on the received information, and sends the screen data to the output unit 13. Display. Such screen data includes images and characters indicating information transmitted from the estimation device 30. For example, when information transmitted from the estimation device 30 is received by the communication unit 11 via the network 70, the control unit 15 generates audio data based on the received information, and outputs the audio data from the output unit 13. You may also output it.

FIG. 14 is a diagram showing an example of teacher data. The training data includes the type of cement, unit cement amount, water content ratio, 7-day strength of three specimens, and 28-day strength of three specimens for each of mixture cases 1-1 to 3-3 in past mixture tests. including.

FIG. 15 is a schematic block diagram showing a specific example of the functional configuration of the learning device 20. As shown in FIG. The learning device 20 is configured using, for example, an information processing device such as a personal computer or a server device. The learning device 20 includes a communication section 21, a storage section 22, and a control section 23.

The communication unit 21 is a communication device. The communication unit 21 may be configured as a network interface, for example. The communication unit 21 performs data communication with other devices via the network 70 under the control of the control unit 23 . The communication unit 21 may be a device that performs wireless communication or may be a device that performs wired communication.

The storage unit 22 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 22 stores data used by the control unit 23. The storage unit 22 may function as, for example, a teacher data storage unit 221, a preprocessed teacher data storage unit 222, and a learned model storage unit 223.

The teacher data storage unit 221 stores teacher data used in the learning process executed by the learning device 20. The teacher data stored in the teacher data storage unit 221 is combination test data showing the results of past combination tests as shown in FIG.

The preprocessed teacher data storage unit 222 stores preprocessed teacher data. The preprocessed teacher data stores information obtained when the preprocessing control unit 232 performs preprocessing on the teacher data. Preprocessing is a process of converting the training data into a data format for inputting into the intensity estimation model.

The trained model storage unit 223 stores a trained model obtained by a learning process using the teacher data stored in the teacher data storage unit 221 and the preprocessed teacher data stored in the preprocessed teacher data storage unit 222. Remember.

The control unit 23 is configured using a processor such as a CPU and a memory. The control unit 23 functions as an information control unit 231, a preprocessing control unit 232, and a learning control unit 233 when a processor executes a program. Note that all or part of each function of the control unit 23 may be realized using hardware such as an ASIC, a PLD, or an FPGA. The above program may be recorded on a computer-readable recording medium. Computer-readable recording media include, for example, portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, semiconductor storage devices (such as SSDs), and storage devices such as hard disks and semiconductor storage devices built into computer systems. It is a device. The above program may be transmitted via a telecommunications line.

The information control unit 231 controls input and output of information. For example, the information control unit 231 acquires teacher data from another device (an information processing device or a storage medium) and records it in the teacher data storage unit 221. Further, for example, the information control unit 231 transmits the learned model stored in the learned model storage unit 223 to another device (for example, the estimation device 30). The preprocessing control unit 232 converts the teacher data into a data format that can be input to the learned model. For example, if the 7-day strength of a plurality of specimens is included for the same type of combination case, the pretreatment control unit 232 may convert it into an average of the 7-day strength of the specimens. Similarly, if the 28-day strength of a plurality of test specimens is included for the same type of combination case, the pretreatment control unit 232 may convert it into an average of the 28-day strength of the specimens. Furthermore, the preprocessing control unit 232 may delete from the teacher data the 7-day strength of a specimen that is determined to exhibit an abnormal value among the 7-day strengths of a plurality of specimens for the same type of combination case. . Similarly, the preprocessing control unit 232 may delete, from the training data, the 28-day strength of a specimen that is determined to exhibit an abnormal value among the 28-day strengths of multiple specimens for the same type of combination case. good.

The learning control unit 233 executes learning processing using the teacher data stored in the preprocessed teacher data storage unit 222 or the preprocessed teacher data. For example, by performing supervised learning, the learning control unit 233 includes the input cement type, the unit cement amount and water content ratio of each of the nine types of mixing cases, and the 7-day strength of one or more mixing cases. Based on the input information, a trained model is generated to output the estimated value of the 28-day strength for each of the nine combination cases of the input unit cement amount and water content ratio. Below, among the nine types of combination cases, the combination case for which the 7-day strength is input into the estimation model will be referred to as the strength input target combination case. The combination cases for strength input are CASE2-2 for one type, CASE2-2 and 2-3 for two types, CASE2-1, 2-2 and 2-3 for three types, and CASE2-1, 2-2 and 2-3 for four types. can be increased to nine types, such as CASE2-1, 2-2, 2-3, and 3-2. The learning control unit 233 records the generated learned model in the learned model storage unit 223. The trained model obtained by the learning control unit 233 may be transmitted to the estimation device 30 and recorded in the learned model storage unit 321 of the estimation device 30.

For example, ExtraTreeRegressor of scikitlearn in Python can be used as the strength estimation model algorithm (see, for example, Reference 1). ExtraTreeRegressor is characterized by its extremely high randomness and resistance to noise and outliers. In the strength estimation model, ExtraTreeRegressor divided the combination test data into multiple decision trees and averaged the values predicted by each decision tree to perform regression. Specifically, since the objective variable was the 28-day strength results for nine combination cases, nine AI (artificial intelligence) models were created, and cross-validation was performed in seven divisions for this evaluation, so each model I created 7 of them. Therefore, 9×7=63 AI models were created, and these 63 AI models were combined into one strength estimation model. Note that the 28-day strength results for the nine types of combination cases are the measurement results of the 28-day strength for each combination case in the combination test indicated by the teacher data.

(Reference 1) Pierre Geurts et al., "Extremely randomized trees," Machine Learning 63(1), 3-42, 2006

FIG. 16 is a flowchart showing a specific example of the processing of the learning device 20. First, the information control unit 231 acquires teacher data (step S101). The teacher data may be input by the user, may be obtained through communication from another information device, or may be obtained from a recording medium connected to the learning device 20, for example. The preprocessing control unit 232 performs predetermined preprocessing on the teacher data (step S102). The learning control unit 233 executes a learning process using the preprocessed teacher data, and records the learned model in the learned model storage unit 223 (step S103).

FIG. 17 is a schematic block diagram showing a specific example of the functional configuration of the estimation device 30. The estimation device 30 is configured using, for example, an information processing device such as a personal computer or a server device. The estimation device 30 includes a communication section 31, a storage section 32, and a control section 33.

The communication unit 31 is a communication device. The communication unit 31 may be configured as a network interface, for example. The communication unit 31 performs data communication with other devices via the network 70 under the control of the control unit 33 . The communication unit 31 may be a device that performs wireless communication or may be a device that performs wired communication.

The storage unit 32 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 32 stores data used by the control unit 33. The storage unit 32 may function as a learned model storage unit 321, for example.

The learned model storage unit 321 stores information on a learned model generated in advance through learning processing. Such a learning process may be executed, for example, by another device (for example, the learning device 20) or by the own device (the estimation device 30).

The control unit 33 is configured using a processor such as a CPU and a memory. The control unit 33 functions as an information control unit 331, a preprocessing control unit 332, and an estimation unit 333 when a processor executes a program. Note that all or part of each function of the control unit 33 may be realized using hardware such as an ASIC, a PLD, or an FPGA. The above program may be recorded on a computer-readable recording medium. Computer-readable recording media include, for example, portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, semiconductor storage devices (such as SSDs), and storage devices such as hard disks and semiconductor storage devices built into computer systems. It is a device. The above program may be transmitted via a telecommunications line.

The information control unit 331 acquires input information from other devices such as the terminal device 10. The input information includes information on the type of cement, the unit cement amount and water content ratio for each of the nine mixing cases, and the 7-day strength of one or more specimens in the mixing case for strength input and the mixing case for strength input. . The information on the strength input target combination case is information indicating which of the nine types of combination cases it is, that is, which of CASE1-1 to CASE3-3. The information control unit 331 transmits information indicating the estimation result obtained by the estimation unit 333 to other devices such as the terminal device 10. Such exchange of information between the information control unit 331 and other devices may be performed by communication by the communication unit 31, for example.

The preprocessing control unit 332 performs predetermined preprocessing on the input information acquired by the information control unit 331. That is, the preprocessing control unit 332 converts the input information into a data format that can be input to the learned model through preprocessing similar to that of the preprocessing control unit 232 of the learning device 20. For example, if the 7-day strength of a plurality of test specimens is included for the same type of combination case, the pretreatment control unit 332 may convert it into an average of the 7-day strength of the specimens. Further, the preprocessing control unit 332 may delete from the input information the 7-day strength of the specimen determined to show an abnormal value among the 7-day strengths of multiple specimens for the same type of combination case. .

The estimation unit 333 performs estimation processing using the trained model stored in the trained model storage unit 321 and the input information whose data format has been converted by the preprocessing control unit 332. The estimation unit 333 may perform estimation processing using the trained model stored in the trained model storage unit 321 and the input information acquired by the information control unit 331. The estimation process outputs estimated values of the 28-day intensity for each of the nine combination cases included in the input information.

Next, estimation of the 28-day intensity using the estimation device 30 will be explained. First, the test person examines the type of cement and nine types of mixing cases as shown in FIG. 1 for the sample (earth and sand) to be tested. The tester decides which combination case from CASE 1-1 to 3-3 should be at least subjected to the combination test, depending on the number of strength input combination cases. The tester performs steps S1 to S8 of the blending test shown in FIG. 2 for the blending cases 1 to 9 that have been determined to be tested, and obtains the 7-day strength.

FIG. 18 is a flowchart showing a specific example of the processing of the estimation device 30. The information control unit 331 acquires input information from the terminal device 10 (step S201). The input information includes the type of cement used in the mix test, the unit cement amount and water content ratio of the nine mix cases considered, the strength input mix case, and the strength input mix case obtained from the mix test. 7-day strength of the sample. The preprocessing control unit 332 performs preprocessing on the input information to convert it into a data format that can be input to the learned model (step S202). The estimation unit 333 performs estimation processing using the input information whose data format has been converted (step S203). The estimation unit 333 transmits information on the estimated value of the 28-day strength of each of the nine combination cases obtained as the estimation result to the terminal device 10 (step S204).

In addition, if the test implementer refers to the estimation results output in step S204 and determines that it is necessary to obtain the 28-day strength of a different mixture case, the test implementer uses the terminal device 10 to calculate the unit cement amount and amount of the mixture case. Generate input information with changed water content ratio. For example, assume that the strength input combination case is CASE2-2. In this case, change the unit cement amount of CASE1 and CASE3 and the water content value of CASE1-1, 1-2, 1-3, 2-1, 2-3, 3-1, 3-2, 3-3. It is possible to generate input information based on Furthermore, there are four types of strength input combination cases: CASE2-1, 2-2, 2-3, and 3-2. In this case, it is possible to generate input information in which the unit cement amount of CASE 1 and the water content values of CASE 1-1, 1-2, 1-3, 3-1, and 3-3 are changed. In addition, if one type of strength input combination case is CASE 2-2, and the 7-day strength of multiple combination cases is measured in the combination test, each of those combination cases is used as the strength input combination case of CASE 2-2. It is also possible to generate input information. The estimation device 30 obtains the generated input information from the terminal device 10 and performs the processing shown in FIG. 18 to calculate the sabo soil with the unit cement amount and water content ratio for which no mixing test was performed, without conducting any further mixing tests. The 28-day strength of cement can be obtained.

FIG. 19 is a diagram showing an example of an input screen displayed on the terminal device 10. In the input screen, the type of cement (E1), the unit cement amount (E2) and water content ratio (E3) of each mixing case, and the 7-day strength (E4) of each specimen in the strength input mixing case are input. Which of the combination cases is the strength input combination case is specified by the combination case number at the beginning of the row in which the 7-day strength (E4) is input. In FIG. 19, three intensity input combination cases, CASE 2-1, CASE 2-2, and CASE 2-3, are shown as examples. For example, if there is one type of strength input combination case, the 7-day strength of CASE 2-2 is input, and the 7-day strength of CASE 2-1 and 2-3 is not input. The information control unit 331 of the estimation device 30 acquires the input information input on the input screen (step S201 in FIG. 18). When the 7-day strengths of a plurality of specimens are input for one strength input combination case, the preprocessing control unit 332 performs preprocessing to obtain the average thereof (step S202 in FIG. 18). In the input screen shown in FIG. 19, the average intensity (E5) for 7 days after pretreatment is displayed. Note that the terminal device 10 may set the average of the 7-day intensity as the input information. The estimation unit 333 transmits the 28-day intensity of each combination case obtained by inputting the preprocessed input information to the learned model to the terminal device 10 (step S203 in FIG. 18). The terminal device 10 displays the 28-day strength (E6) of each received combination case.

Note that in step S201 in FIG. 18, the average 7-day strength (E5) may be input without inputting the 7-day strength (E4). In this case, it is not necessary to perform preprocessing to calculate the average for each intensity input combination case in step S202.

On the input screen shown in Fig. 19, by changing the cement amount of the cement case to which the strength input mix case does not belong and the water content ratio of the cement case other than the strength input mix case, you can change the 28-day strength of other unit cement amounts and water content ratios. It is possible to obtain an estimate of . In addition, when measuring the 7-day strength of multiple blending cases in a blending test, the unit cement and The water content ratio may be set and the input information may be changed.

In the estimation system 100 configured as described above, it is possible to estimate the 28-day strength of multiple types of combination cases by using information on the 7-day strength of at least one specimen of at least one type of combination case. becomes. Specifically, the details are as follows. The estimation system 100 executes a learning process by machine learning using the combination test data of 7-day strength and 28-day strength of a plurality of combination cases obtained as a result of the combination test, and obtains a learned model. Then, the estimation system 100 acquires new input information regarding the 7-day strength of the combination test conducted on at least some of the combination cases of which the 28-day strength is to be estimated, and applies the acquired input information to the Based on this, estimated information on the 28-day strength of each of the plurality of combination cases to be estimated is acquired.

In the above description, the case where there are nine types of combination cases has been described, but the types of combination cases may be other than nine. Further, the training data generated based on the input information input to the estimation device 30 in the process of FIG. 18 and the estimation result obtained using the input information may be used in the learning process of FIG. .

FIG. 20 is a diagram schematically showing an example of the hardware configuration of the information processing device 90 applied to this embodiment. The information processing device 90 includes a processor 91 , a main memory 92 , a communication interface 93 , an auxiliary memory 94 , an input/output interface 95 , and an internal bus 96 . The processor 91 , main memory 92 , communication interface 93 , auxiliary memory 94 , and input/output interface 95 are communicably connected to each other via an internal bus 96 . The information processing device 90 may be applied to the learning device 20 and the estimation device 30, for example. In this case, for example, the communication unit 21 and the communication unit 31 may be configured using the communication interface 93. For example, the storage unit 22 and the storage unit 32 may be configured using an auxiliary storage device 94. Further, the control unit 23 and the control unit 33 may be configured using the processor 91 and the main storage device 92.

(Modified example)
In this embodiment, the terminal device 10 and the estimation device 30 are configured as different devices, but they may be configured as an integrated device. FIG. 21 is a diagram showing a modification of the estimation device 30 configured in this manner. The estimation device 30 shown in FIG. 21 includes an input section 34 and an output section 35. The input unit 34 and output unit 35 of the estimation device 30 shown in FIG. 21 function similarly to the input unit 12 and output unit 13 of the terminal device 10, respectively. The control unit 33 operates in response to an operation on the input unit 34, performs estimation processing using input information, and outputs information using the output unit 35.

In this embodiment, the learning device 20 and the estimation device 30 are configured as different devices, but they may be configured as an integrated device. FIG. 22 is a diagram showing a modification of the estimation device 30 configured in this way. The storage unit 32 of the estimation device 30 shown in FIG. 22 also functions as a teacher data storage unit 322 and a preprocessed teacher data storage unit 323. The control unit 33 of the estimation device 30 shown in FIG. 22 also functions as a learning control unit 334. The teacher data storage section 322 and the preprocessed teacher data storage section 323 function similarly to the teacher data storage section 221 and the preprocessed teacher data storage section 222 of the learning device 20, respectively. The preprocessing control unit 332 executes not only the preprocessing of the estimation device 30 (preprocessing on input information) but also the preprocessing of the learning device 20 (preprocessing on teacher data). The learning control unit 334 functions similarly to the learning control unit 233 of the learning device 20.

Furthermore, even when the learning device 20 and the estimation device 30 are configured as different devices, the estimation device 30 may use the learned model storage section 223 of the learning device 20 as the learned model storage section 321.

The learning device 20 may be implemented using a plurality of information processing devices. For example, the learning device 20 may be implemented using a device such as a cloud. For example, in the learning device 20, the storage section 22 and the control section 23 may be implemented in different information processing devices. For example, the storage unit 22 of the learning device 20 may be distributed and implemented in a plurality of information processing devices. The estimation device 30 may be implemented using a plurality of information processing devices. For example, the estimation device 30 may be implemented using a device such as a cloud. For example, in the estimation device 30, the storage section 32 and the control section 33 may be implemented in different information processing devices. For example, the storage unit 32 of the estimation device 30 may be distributed and implemented in a plurality of information processing devices.

Further, the learning control unit 233 of the learning device 20 and the learning control unit 334 of the estimation device 30 may create a strength estimation model for each type of cement. In this case, the estimation unit 333 of the estimation device 30 selects a trained model based on the cement type information included in the input information, and inputs the input information excluding the cement type information into the selected trained model. , obtain estimates of the 28-day strength for each of the multiple combination cases.

A prediction accuracy evaluation experiment using the trained model of this embodiment will be described. ExtraTreeRegressor was used as the trained model. The learning device 20 learned two trained models. One used 3034 pieces of teacher data, and the other used 2178 pieces of teacher data. Based on the data of the evaluation test conducted according to the procedure shown in FIG. 2, input information was generated in which the number of combination cases for strength input was increased from one to nine. The estimation device 30 inputted the input information to each learned model and obtained the estimation results of the 28-day strength for each of the nine combination cases. The coefficient of determination ^R2 was determined for each estimation result obtained. For all trained models, the coefficient of determination ^R2 is 0.8 or more for all combination cases from 1 to 9 for strength input, and the greater the number of combination cases for strength input, the higher the accuracy. Rose.

FIG. 23 is a diagram showing the estimated results of the 28-day strength of the conventional blending test actually conducted and the 28-day strength using the present embodiment. Predictions according to this embodiment were performed for two patterns: a case in which there is one type of intensity input combination case (hereinafter referred to as "one case prediction result") and a case in which there are three types of intensity input combination cases (hereinafter referred to as "three case prediction result"). Ta. Symbols F1, F2, and F3 are actually measured 28-day strengths obtained by conventional combination tests, symbols G1, G2, and G3 are 1-case prediction results (AI_1), and symbols H1, H2, and H3 are 3-case prediction results. (AI_3). In addition, codes F1, G1, and H1 have a cement amount of 50 (C-50), codes F2, G2, and H2 have a cement amount of 100 (C-100), and codes F3, G3, and H3 have a cement amount of 150 (C-50). 150).

The coefficient of determination (R ² value) for the 1-case prediction result was 0.83, and the determination coefficient (R ² value) for the 3-case prediction result was 0.90. Although the 3-case prediction result is more accurate, the deviation in the 1-case prediction result is very small if the variations in natural soil are taken into consideration. As mentioned above, the results of estimating the 28-day strength of nine combination cases were obtained with high estimation accuracy.

According to the embodiment described above, the minimum number of specimens to be produced for a physical blending test is changed from the conventional 54 specimens (3 specimens in each of 9 cases) to 3 specimens in 1 case, or 1 specimen in 1 case. It can be done. Additionally, the amount of soil collected for blending tests is significantly reduced. Additionally, the time it takes to obtain the results of a blending test is shortened from the conventional 28 days to 7 days, reducing construction time. In addition, even if the required quality is not satisfied in the initial blending case, by changing the blending case settings in the input information, it is possible to obtain useful information for determining the recommended blending without conducting additional blending tests. . Therefore, the risks of collecting additional samples, increasing the construction period, and increasing costs associated with these are reduced. In addition, if the compounding test facility is far from the sample collection site or if it is difficult to transport the sample, JIS standard tests can be conducted near the sampling site and three or one specimen of one compounding case can be carried out on-site. By creating a 7-day strength information and using it as input data, it becomes possible to study the recommended combination. For example, a mixture test kit containing one mixture case worth of specimens is easy to transport, so even if the site is overseas, for example, approximately 15 kg of soil can be collected at the site and its strength measured for 7 days. By doing so, it becomes possible to obtain the data necessary for the specific formulation in a short period of time. Although the above description has been made using sabo soil cement as an example, the present invention can be applied to soil cement that mixes earth and sand, a solidification material, and water.

According to the embodiment described above, the estimation system includes a learning control section and an estimation section. The learning device may have a learning control section, and the estimating device may have an estimating section. The learning control unit calculates the amount of solidified material and water content of each of the plurality of types of mixture cases of soil cement, the first period strength which is the strength after the first period of the soil cement of each of the plurality of types of mixture cases, and the first period. Using training data indicating the second period strength, which is the strength after a second period longer than A strength estimation model is learned by machine learning, which takes the one-period strength as an input and outputs the second-period strength of soil cement for each of a plurality of combination cases of the input solidified material amount and water content. The estimation unit adds the amount of solidified material and water content of each of the plurality of combination cases targeted for second period strength estimation, and one or more of the plurality of combination cases targeted for second period strength estimation to the learned strength estimation model. By inputting the first period strength of the soil cement for the combination case of , the second period strength of the soil cement for each of the plurality of combination cases of the input solidification material amount and water content is obtained. For example, the first period is 7 days in the embodiment described above, and the second period is 28 days in the embodiment described above. The first period is a period during which the hydration reaction of cement is not progressing, and the second period is a period during which it can be confirmed whether the hydration reaction of cement has reached a strength that can guarantee that the material has hardened. .

The training data may further include information on the type of solidifying material. The strength estimation model further inputs the type of solidifying material. The estimation unit inputs the type of solidification material, the amount of solidification material and water content of each of the plurality of combination cases, and the first period strength of the soil cement of one or more combination cases into the strength estimation model. Obtain the output of the second period strength of soil cement for each of the plurality of combination cases using the input type of solidification material.

In addition, in the above embodiment, an example was explained where the second period strength is higher than the first period strength, but this also applies to the case of soil cement where the second period strength is lower than the first period strength. It is possible.

Although the embodiments of the present invention have been described above in detail with reference to the drawings, the specific configuration is not limited to these embodiments, and includes designs within the scope of the gist of the present invention.

100... Estimation system, 10... Terminal device, 11... Communication unit, 12... Input unit, 13... Output unit, 14... Storage unit, 15... Control unit, 20... Learning device, 21... Communication unit, 22... Storage unit, 221... Teacher data storage section, 222... Preprocessed teacher data storage section, 223... Learned model storage section, 23... Control section, 231... Information control section, 232... Preprocessing control section, 233... Learning control section, 30 ...Estimation device, 31...Communication section, 32...Storage section, 321...Learned model storage section, 322...Teacher data storage section, 323...Preprocessed teacher data storage section, 33...Control section, 331...Information control section, 332... Preprocessing control section, 333... Estimation section, 334... Learning control section, 34... Input section, 35... Output section

Claims (14)

The solidified material amount and water content of each of the plurality of types of mixing cases of soil cement, the first period strength which is the strength after the first period of the soil cement of each of the plurality of types of mixing cases, and the strength than the first period Using the training data indicating the second period strength, which is the strength after a long second period, the mixture case for the second period strength estimation for the soil cement that is the target soil cement for the second period strength estimation. Input the amount of solidified material and water content of each of a plurality of types of estimation target mixing cases, and the first period strength of the estimation target soil cement of some of the estimation target mixing cases among the plurality of types of estimation target mixing cases. , a learning control unit that uses machine learning to learn a strength estimation model that outputs an estimated value of the second period strength of the estimation target soil cement of the estimation target mixture case in which the first period strength is not input;
A learning device equipped with. The amount of solidified material and water content of each of the plurality of combination cases of soil cement in which the mixing test was conducted , and the strength after the first period of the soil cement of each of the plurality of combination cases obtained as a result of the mixing test. Using training data showing the first period strength, which is the strength after the second period which is longer than the first period, and the second period strength which is the strength after the elapse of the second period which is longer than the first period, are soil cements using different earth and sand. The amount of solidified material and water content of each of the multiple types of estimation target mixture cases, which are the mixture cases for the second period strength estimation of the soil cements to be estimated, and the soil cements to be estimated . The measured first period strength of one or more of the estimation target mixture cases is input, and an estimated value of the second period strength of the estimation target soil cement for each of the plurality of input estimation target mixture cases is output. a learning control unit that learns the intensity estimation model by machine learning;
A learning device equipped with. The teacher data further includes information on the type of solidification material,
The learning control unit further receives the type of solidification material as input and learns a strength estimation model that outputs the estimated value by machine learning.
The learning device according to claim 1 or claim 2 . The solidified material amount and water content of each of the plurality of types of blending cases of soil cement, the first period strength which is the strength after the first period of the soil cement of each of the plurality of blending cases, and the strength than the first period. The strength estimation model trained using the training data indicating the second period strength, which is the strength after a long second period , is added to the second period for the soil cement to be estimated, which is the soil cement to be estimated for the second period strength. The amount of solidified material and water content of each of the plurality of types of estimation target combination cases that are the combination cases of strength estimation, and the amount of solidified material and water content of the estimation target soil cement of some of the estimation target combination cases among the plurality of types of estimation target combination cases. an estimating unit that obtains an output of an estimated value of the second period strength of the estimation target soil cement of the estimation target mixture case in which the first period strength is not input by inputting the first period strength;
An estimation device comprising: The amount of solidified material and water content of each of the plurality of combination cases of soil cement in which the mixing test was conducted , and the strength after the first period of the soil cement of each of the plurality of combination cases obtained as a result of the mixing test. The strength estimation model learned using training data indicating the first period strength, which is the strength after the second period which is longer than the first period, and the second period strength, which is the strength after the elapse of the second period, which is longer than the first period. The amount of solidified material and water content for each of multiple types of estimation target mixture cases, which are the mixture cases for the second period strength estimation, for the estimation target soil cement, which is a soil cement that uses earth and sand different from the earth and sand used for cement. , the first period strength of one or more of the estimation target mixture cases measured for the estimation target soil cement. an estimator that obtains an output of the estimated value of the two-period intensity;
An estimation device comprising: The teacher data further includes information on the type of solidification material,
The estimating unit obtains an output of the estimated value by further inputting a type of solidifying material to the strength estimating model.
The estimation device according to claim 4 or claim 5 . The solidified material amount and water content of each of the plurality of types of blending cases of soil cement, the first period strength which is the strength after the first period of the soil cement of each of the plurality of blending cases, and the strength than the first period. Using training data indicating the second period strength, which is the strength after a long second period, the mixture case of the second period strength estimation target for the estimation target soil cement, which is the second period strength estimation target soil cement, is calculated. Input the amount of solidified material and water content of each of the plurality of estimation target mixing cases, and the first period strength of the estimation target soil cement of some of the estimation target mixing cases among the plurality of estimation target mixing cases. , a learning control unit that uses machine learning to learn a strength estimation model that outputs an estimated value of the second period strength of the estimation target soil cement of the estimation target mixture case in which the first period strength is not input ;
Estimation of the amount of solidified material and water content of each of the plurality of types of estimation target combination cases for the second period strength estimation, and of some of the estimation target combination cases among the plurality of types of estimation target combination cases for the second period strength estimation. By acquiring input information including the first period strength of the target soil cement and inputting the acquired input information into the strength estimation model , the first period strength included in the input information and the input information an estimating unit that outputs an estimated value of the second period strength of the estimation target soil cement of the estimation target mixture case in which no period strength is input;
An estimation system comprising: The amount of solidified material and water content of each of the plurality of combination cases of soil cement in which the mixing test was conducted , and the strength after the first period of the soil cement of each of the plurality of combination cases obtained as a result of the mixing test. Using training data showing the first period strength, which is the strength after the second period which is longer than the first period, and the second period strength which is the strength after the elapse of the second period which is longer than the first period, are soil cements using different earth and sand. The amount of solidified material and water content of each of the multiple types of estimation target mixture cases, which are the mixture cases for the second period strength estimation of the soil cements to be estimated, and the soil cements to be estimated . The measured first period strength of one or more of the estimation target mixture cases is input, and an estimated value of the second period strength of the estimation target soil cement for each of the plurality of input estimation target mixture cases is output. a learning control unit that learns the intensity estimation model by machine learning;
Estimation of the amount of solidified material and water content of each of the plurality of types of estimation target combination cases for second period strength estimation, and one or more of the plurality of estimation target combination cases for second period strength estimation. By acquiring input information including the first period strength of the target soil cement and inputting the acquired input information into the strength estimation model , each of the plurality of types of the estimation target mixture cases included in the input information is calculated. an estimator that outputs an estimated value of the second period strength of the soil cement to be estimated;
An estimation system comprising: The solidified material amount and water content of each of the plurality of types of blending cases of soil cement, the first period strength which is the strength after the first period of the soil cement of each of the plurality of blending cases, and the strength than the first period. Using training data indicating the second period strength, which is the strength after a long second period, the mixture case of the second period strength estimation target for the estimation target soil cement, which is the second period strength estimation target soil cement, is calculated. Input the amount of solidified material and water content of each of the plurality of estimation target mixing cases, and the first period strength of the estimation target soil cement of some of the estimation target mixing cases among the plurality of estimation target mixing cases. , a learning step of learning by machine learning a strength estimation model that outputs an estimated value of the second period strength of the estimation target soil cement of the estimation target mixture case in which the first period strength is not input ;
A learning method with The amount of solidified material and water content of each of the plurality of combination cases of soil cement in which the mixing test was conducted , and the strength after the first period of the soil cement of each of the plurality of combination cases obtained as a result of the mixing test. Using training data showing the first period strength, which is the strength after the second period which is longer than the first period, and the second period strength which is the strength after the elapse of the second period which is longer than the first period, are soil cements using different earth and sand. The amount of solidified material and water content of each of the multiple types of estimation target mixture cases, which are the mixture cases for the second period strength estimation of the soil cements to be estimated, and the soil cements to be estimated . The measured first period strength of one or more of the estimation target mixture cases is input, and an estimated value of the second period strength of the estimation target soil cement for each of the plurality of input estimation target mixture cases is output. a learning step of learning the strength estimation model by machine learning;
A learning method with The solidified material amount and water content of each of the plurality of types of blending cases of soil cement, the first period strength which is the strength after the first period of the soil cement of each of the plurality of blending cases, and the strength than the first period. The strength estimation model trained using the training data indicating the second period strength, which is the strength after a long second period , is added to the second period for the soil cement to be estimated, which is the soil cement to be estimated for the second period strength. The amount of solidified material and water content of each of the plurality of types of estimation target combination cases that are the combination cases of strength estimation, and the amount of solidified material and water content of the estimation target soil cement of some of the estimation target combination cases among the plurality of types of estimation target combination cases. an estimation step of obtaining an output of an estimated value of the second period strength of the estimation target soil cement of the estimation target mixture case in which the first period strength is not input by inputting the first period strength;
An estimation method with The amount of solidified material and water content of each of the plurality of combination cases of soil cement in which the mixing test was conducted , and the strength after the first period of the soil cement of each of the plurality of combination cases obtained as a result of the mixing test. The strength estimation model learned using training data indicating the first period strength, which is the strength after the second period which is longer than the first period, and the second period strength, which is the strength after the elapse of the second period, which is longer than the first period. The amount of solidified material and water content for each of multiple types of estimation target mixture cases, which are the mixture cases for the second period strength estimation, for the estimation target soil cement, which is a soil cement that uses earth and sand different from the earth and sand used for cement. , the first period strength of one or more of the estimation target mixture cases measured for the estimation target soil cement. an estimation step to obtain the output of the two-period intensity estimate ;
An estimation method with computer,
A program for functioning as the learning device according to claim 1 or claim 2. computer,
A program for functioning as the estimation device according to claim 4 or claim 5 .

Images