JP2019157346A - Ground evaluation system and ground evaluation method - Google Patents

Abstract

To provide a ground evaluation system and a ground evaluation method allowing adequate evaluation of the ground.SOLUTION: A computer terminal 30 comprises a measurement value storage part 35 and a memory 31a storing an index value showing excavation load in excavation and an N value by associating with each other for every depth, and a control part. The control part comprises a management part 31, an N value learning part 32 and an N value calculation part 33. The N value learning part 32 is used as an input layer using the index value stored in the measurement value storage part 35 as an input element, performs machine learning by using teacher data using the N value at the depth of the index value as an output layer to form an N value calculation model, and stores in a learning result memory part 36. The N value calculation part 33 performs estimation processing estimating the evaluation object N value by using the N value calculation model stored in the learning result memory part 36 and the input element of the evaluation object.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a ground evaluation system and a ground evaluation method for evaluating a support layer of a ground on which piles are installed.

  When constructing a structure, there is a method of driving a plurality of piles into a hard formation in the ground and supporting the load of the structure in the hard formation through the piles. A hard layer suitable for supporting a structure is called a support layer. For this reason, the pile hole which inserts a pile must be reached even to a support layer. However, due to the limitations of the excavation method, confirmation of arrival at the support layer often depends on subjective judgment based on experience, and it is difficult to make an objective judgment.

  Usually, before the construction of a structure, a ground survey is performed to specify the depth (position) of the support layer. And the N value of the parameter | index which shows the hardness of a ground is acquired by the standard penetration test in a ground investigation.

  However, the geological structure is not always the same in the entire construction site. In addition, since ground surveys are costly and troublesome, it is difficult to conduct ground surveys at all pile hole positions.

  Therefore, conventionally, as an index indicating the hardness of the ground during excavation, the value obtained by integrating the current value for outputting drilling torque at each drilling depth (integrated current value) is used to reach the support layer of the pile hole. A technique for determining the above has been studied (see, for example, Patent Document 1). In this patent document 1, the current value at the time of excavation of the auger drive motor that drives the auger of the excavator that drills the ground is detected, and the vertical movement distance of the auger is detected. Then, simultaneously with the start of excavation by the excavator, measurement of the auger current data and the excavation depth is started, and these measured values are displayed on the same screen.

JP-A-5-287721

  As shown in Patent Document 1 described above, the support layer may be determined using a change in the integrated current value. However, there are cases where the change in the integrated current value is not linked to the change in the N value. For this reason, when the N value is specified based on the integrated current value, the N value may not be appropriate.

  The present applicant has invented a method of determining using a speed index value for excavation speed during excavation of a pile hole and frequency analysis of vibration during excavation, and Japanese Patent Application Nos. 2017-003676, 2017-003677 and Patent application No. 2017-153621 has been filed.

  The ground evaluation system for solving the above-mentioned problem is stored in the measurement value storage unit, the measurement value storage unit that stores the index value indicating the excavation load during excavation and the ground evaluation information in association with each depth. Uses the index value as an input layer as an input element, and performs a learning process that generates a ground evaluation model by performing machine learning using teacher data using the ground evaluation information at the depth of the index value as an output layer And a calculation unit that executes an estimation process for estimating the ground evaluation information to be evaluated using the ground evaluation model and the input element to be evaluated.

  According to the present invention, the ground can be accurately evaluated.

It is explanatory drawing of the excavation apparatus which excavates the pile hole in embodiment, Comprising: (a) is a schematic sectional drawing of an excavation apparatus, (b) is a block diagram which shows the structure of a drilling management system. Explanatory drawing of the data structure memorize | stored in the measured value memory | storage part of embodiment. Schematic explanatory drawing explaining the process of the learning part in embodiment. The flowchart explaining the process sequence of the whole process in embodiment. It is explanatory drawing explaining the correlation process with the depth in embodiment, (a) is a measured value graph, (b) is a graph which extracted the drilling time zone, (c) is the measured value of the drilling time zone. The connected graph, (d) is a flowchart of the processing procedure. It is explanatory drawing of the data corresponding to the depth in embodiment, (a) is excavation time, (b) is an electric current value, (c) is a flow rate of water, (d) is a drilling speed, (e) is an integral current. Indicates the value. It is explanatory drawing of the graph of the estimated N value and N value corresponding to the depth in embodiment, (a) is a case where the estimated N value is calculated using the depth, speed, current value, water amount, and integrated current value. ) Indicates a case where the estimated N value is calculated using the speed, current value, water amount, and integrated current value. It is explanatory drawing at the time of considering the vibration in a modified example, (a) is a vibration characteristic value corresponding to depth, (b) is estimated using depth, speed, current value, water volume, integrated current value, and vibration characteristic value. When the N value is calculated, (c) shows a case where the estimated N value is calculated using the speed, current value, water amount, integrated current value, and vibration characteristic value.

  Hereinafter, an embodiment that embodies the ground evaluation system and the ground evaluation method will be described with reference to FIGS. In the present embodiment, it is evaluated whether or not the ground is hard enough to be a support layer. For this evaluation, ground evaluation information for evaluating the ground on which the building is built is used. Here, the N value is used as the ground evaluation information. This ground evaluation information is estimated using an index value indicating the excavation load during excavation. In this embodiment, the index value indicating the excavation load includes an elapsed time corresponding to the depth of excavation, an (instantaneous) current value supplied to the excavator, a flow rate of excavation water supplied to the head of the excavator, and Calculated values (drilling speed and integrated current value) calculated from these measured values (depth, current value, amount of water) are used.

  FIG. 1 (a) shows an excavator 10 as an excavator for excavating a pile hole h0 for installing a building pile. The excavator 10 includes a base machine 11, a mast 14, and an auger machine 16. The base machine 11 includes a lower traveling body including the crawler 12 and an upper swing body including the operation chamber 13.

  The mast 14 is erected on the base machine 11. A wire for measuring the depth / velocity meter is provided in the mast 14. An auger machine 16 is attached to the mast 14 so as to be movable up and down. The auger machine 16 includes a drive motor housed in a box and a drilling rod 17 that is rotationally driven by the drive motor. An excavation head 18 is attached to the tip (lower end) of the excavation rod 17. The excavation head 18 has an excavation blade formed at the tip of a pair of (two) excavation arms that swing. The elevation of the excavation head 18 is controlled by the operator of the operation chamber 13.

  The excavator 10 is connected to a drilling water supply device (not shown) that supplies drilling water to the drilling head 18. The amount of the drilling water is adjusted by the operator of the drilling water supply device (not shown) according to the instruction of the operator of the operation chamber 13 according to the excavation situation.

  As shown in FIG. 1B, the excavator 10 includes a drilling management system 20. The drilling management system 20 includes a computer terminal 30, a drilling depth measuring device 21, a flow rate measuring device 22, a current measuring device 23, an input unit 25, and a display unit 26. Each measuring instrument (21 to 23) always performs measurement, and transmits the measured actual value to the computer terminal 30.

The drilling depth measuring device 21 measures the wire feed amount in the mast 14 and measures the drilling depth (depth) according to the position of the excavation head 18.
The flow rate measuring device 22 measures the injection flow rate (water amount) of drilling water supplied from a drilling water supply device (not shown).
The current measuring instrument 23 measures the load current (instantaneous current value) of the drive motor of the auger machine 16.

The input unit 25 includes a keyboard and a pointing device arranged in the operation room 13, and is used for inputting various data to the computer terminal 30.
The display unit 26 includes a display or the like disposed in the operation room 13 and displays various data.

  The computer terminal 30 acquires each measured value data from each measuring instrument (21 to 23). The computer terminal 30 includes a control unit (CPU, RAM, ROM, etc.), a measured value storage unit 35, and a learning result storage unit 36, and each of the processes described later (management stage, N value learning stage, N value calculation stage, etc.) ). Therefore, the control part of the computer terminal 30 functions as the management part 31, the N value learning part 32, and the N value calculation part 33 by executing the drilling management program stored in the memory.

  The management unit 31 accumulates the start instruction from the input unit 25 and the actually measured values (depth, water amount, current value) from the measuring instruments (21 to 23) in the memory 31a. Furthermore, the management part 31 calculates a calculated value based on the measured value for every predetermined time. Specifically, the management unit 31 calculates a drilling speed and an integrated current value in a time zone (drilling time zone) in which the drilling is actually performed. The excavation head 18 may be temporarily lifted in a hard formation or the like just before digging. For this reason, as shown in FIG. 5A, the actual drilling depth of the excavation head 18 does not necessarily increase monotonously according to the elapsed time. Therefore, the management unit 31 deletes the lifting / stopping period of the excavation head 18 (the shaded time zone in FIG. 5B) from the entire work time, and the cutting part substantially used for drilling. Specify the measurement value of the hole time zone. The management unit 31 generates a graph shown in FIG. 5C by connecting the measured values in the specified drilling time zone.

Furthermore, the management unit 31 controls the N value learning unit 32 and the N value calculation unit 33.
The N value learning unit 32 generates an N value calculation model as a ground evaluation model using the teacher data. The N value learning unit 32 estimates an N value calculation model by deep learning (machine learning) using an index value indicating the excavation load and a depth as an input layer and using the N value as an output layer. Here, as the index value indicating the excavation load, the flow rate of water used for excavation, the (instantaneous) current value during excavation, the drilling speed, and the integrated current value are used.

  The N value calculation unit 33 performs a process of calculating the N value by using an N value calculation model and an input element (an index value or a depth indicating the excavation load) associated with the depth (depth).

  As shown in FIG. 2, the measurement value storage unit 35 stores measurement value information 350. The measurement value information 350 includes a columnar diagram 351, N value data 352 corresponding to depth, a pile hole identifier 353, measurement value data 355, and calculation value data 356 in association with the site identifier.

The site identifier is an identifier for specifying each site.
The columnar diagram 351 is created based on the soil sample acquired in the boring survey at the construction site where the pile hole is excavated. The N value data 352 corresponding to the depth is created based on the standard penetration test data acquired in this boring survey.

The pile hole identifier 353 is an identifier for specifying the drilled pile hole.
The measured value data 355 and the calculated value data 356 are index values indicating the excavation load. The measured value data 355 is an index value measured at the time of excavation. The measurement value data 355 is generated and recorded when the measurement value is acquired. The measurement value data 355 includes elapsed time data R1 corresponding to the depth, current value data R2 corresponding to the depth, and water amount (flow rate) data R3 per unit time corresponding to the depth.

  FIGS. 6A to 6C show examples of elapsed time data R1 according to depth, (instantaneous) current value data R2 according to depth, and water amount (flow rate) data R3 per unit time according to depth. Show.

  The calculated value data 356 is an index value calculated using the measured value data 355. In the present embodiment, the calculated value data 356 includes drilling speed data R11 corresponding to the depth and integrated current value data R21 corresponding to the depth.

FIGS. 6D and 6E show examples of drilling speed data R11 corresponding to the depth and integrated current value data R21 corresponding to the depth.
The drilling speed data R11 is calculated by using the elapsed time data R1 to divide every fixed excavation range (for example, depth 0.5 m) by the time required for drilling in this fixed excavation range.
The integrated current value data R21 is calculated by summing current values for the time required to excavate each constant excavation range (for example, depth 0.5 m) using the current value data R2.

  As illustrated in FIG. 3, the learning result storage unit 36 stores a learning result (N value calculation model) generated by the N value learning unit 32 through learning. This learning result is generated by machine learning (deep learning) using an index value indicating excavation load and an input layer having depth as an input element and using an N value as an output layer.

(Drilling)
Next, according to FIGS. 4-7, the excavation process of a pile hole is demonstrated using the excavator 10 comprised as mentioned above.

<Boring investigation process>
First, as shown in FIG. 4, a boring investigation process is executed before excavation. In this boring survey process, as is well known, a geological survey is conducted on the site of the construction site. A columnar map is generated according to the type of soil sample acquired during the geological survey. Moreover, N value for every predetermined predetermined depth is acquired, and N value data is produced | generated.

  And the registration process of a columnar figure and N value is performed using the management part 31 of the computer terminal 30 (step S1-1). Specifically, the management unit 31 outputs a ground information registration screen to the display unit 26. This ground information registration screen includes an input field for registering data relating to the columnar diagram and N-value data corresponding to the depth. The manager of the construction site inputs an N-value graph corresponding to the site identifier, the columnar diagram, and the depth on the ground information registration screen. The management unit 31 of the computer terminal 30 generates measurement value information 350 including the input site identifier, the columnar diagram 351, and N value data 352 corresponding to the depth, and records the measurement value information 350 in the measurement value storage unit 35.

<Testing process>
Thereafter, a trial digging process is performed. In this trial excavation process, for example, a pile hole near the point where the boring survey was performed is excavated, and an index value indicating the excavation load at this time is acquired.

  Here, first, the control unit of the computer terminal 30 of the drilling management system 20 acquires the excavation start input using the input unit 25. The control unit 31 of the control unit starts rotation of the drive motor of the auger machine 16, inserts the excavation head 18 into the ground, and starts drilling. In this case, the management unit 31 displays an output screen including the columnar diagram 351 and the N value data 352 corresponding to the depth on the display of the display unit 26. This output screen includes a display area for displaying the elapsed time, current value, water volume, drilling speed, and integrated current value according to the depth.

  And the management part 31 of the computer terminal 30 performs the acquisition process of measured value during drilling (step S2-1). Specifically, the management part 31 acquires the actual value measured in each measuring device (21-23) for every predetermined time.

  Next, the control unit of the computer terminal 30 executes an association process with the depth (step S2-2). Specifically, the management unit 31 of the control unit, among the actual measurement values obtained from the respective measuring instruments (21 to 23), the actual measurement value in the drilling time zone excluding the time when the actual drilling is not performed, Identified as a measurement value to be evaluated. Details of this processing will be described later.

  Next, the control unit of the computer terminal 30 executes a calculation process of the drilling speed and the integrated current value (step S2-3). Specifically, the management unit 31 of the control unit uses the excavation distance (depth) corresponding to the time (excavation time) in the drilling time zone specified in step S2-2 to make this constant excavation range for each constant excavation range. Divide by the time required for drilling in the excavation range to calculate the drilling speed. Furthermore, the management unit 31 uses the depth and the current value according to the time in the drilling time zone, and sums the current value for the time required to excavate each fixed excavation range in the drilling time zone, Calculate the integrated current value. Then, the management unit 31 records the calculated drilling speed and integrated current value in the memory 31a.

  Then, whenever the measured value is acquired at any time during the drilling (each time the process of step S2-1 is executed), the control unit of the computer terminal 30 executes the process of steps S2-2 and S2-3. To do.

  Thereafter, when it is determined that the hole has been drilled to a predetermined depth using the columnar diagram 351 and the N value data 352, the control unit of the computer terminal 30 raises the excavation head 18 to be pulled out from the pile hole. Here, the management unit 31 of the control unit determines that the drilling has been completed when an instruction to raise the excavation head 18 is acquired via the input unit 25. When it is determined that the drilling is finished, the elapsed time, current value, water amount, drilling speed, and integrated current value corresponding to the depth stored in the memory 31a are set to each data (R1, R2, R3, R11, R21) is recorded in the measured value storage unit 35. It should be noted that the end of the drilling may be determined at another timing. For example, it may be just before the auger machine 16 is stopped after the excavation head 18 has been pulled up.

[Association with depth]
Next, the details of the above-described depth association process (step S2-2) will be described with reference to FIG.

  First, as shown in FIG.5 (d), the management part 31 performs the acquisition process of the drilling depth of a digging head (step S5-1). Specifically, as shown in FIG. 5A, the management unit 31 acquires the drilling depth measured from the drilling depth measuring instrument 21, and temporarily stores it in the memory 31a together with the acquired measurement time.

  And the management part 31 of a control part performs the extraction process of a drilling time slot | zone (step S5-2). Specifically, as shown in FIG. 5 (b), the management unit 31 uses the drilling depth temporarily stored in the memory 31a in this drilling to a depth not less than the maximum value in the past drilling depths. Then, the drilling time zone in which the depth monotonously increases (region other than the shaded portion in FIG. 5B) is specified. In addition, when the excavation head 18 is pulled up, the excavation head 18 is added to the drilling time zone again from the time when the excavation head 18 reaches the hole bottom.

  Next, the management part 31 of a control part performs the specific process of the measured value of a drilling time slot | zone (step S5-3). Specifically, as illustrated in FIG. 5C, the management unit 31 calculates the elapsed time according to the drilling depth by connecting the drilling depths in the drilling time zone. And the management part 31 memorize | stores a depth, an electric current value, and the amount of water in the memory 31a in association with the time (digging time) in a drilling time zone.

<Learning process>
Next, as shown in FIG. 4, a learning process for learning the N-value calculation model is executed using the measured value and the calculated value acquired in the trial drilling process as teacher data.

  Here, first, the control unit of the computer terminal 30 executes a teacher data acquisition process (step S3-1). Specifically, the model generation execution screen is output to the display unit 26. This model generation execution screen includes an input field for inputting a site identifier for specifying a site for calculating the N value, and an execution button. The manager of the construction site inputs the site identifier in the input field of the model generation execution screen and selects the execution button. The management part 31 of the computer terminal 30 acquires the execution instruction | indication of a learning process with the input field identifier.

  In this case, the N value learning unit 32 extracts N value data 352, measurement value data 355, and calculated value data 356 associated with the site identifier in the measurement value storage unit 35. Then, the N value learning unit 32 includes the elapsed time data R1, the current value data R2 and the water amount data R3 of the measured value data 355, the drilling speed data R11 and the integrated current value data R21 of the calculated value data 356 (and the depth). Are specified as input elements of the input layer. Each of these data (R1, R2, R3, R11, R21) is associated with depth. Further, the N value learning unit 32 specifies N value data 352 associated with the depth of each data (R1, R2, R3, R11, R21) specified as the input layer as an output layer.

  Next, the control part of the computer terminal 30 performs the learning process of N value calculation (step S3-2). Specifically, the N value learning unit 32 of the control unit performs machine learning (deep learning) using the input layer and the output layer to generate an N value calculation model.

  Next, the control unit of the computer terminal 30 executes a learning result storing process (step S3-3). Specifically, the N value learning unit 32 of the control unit stores the generated N value calculation model in the learning result storage unit 36.

<Drilling process>
Next, the excavation process will be described with reference to FIG. In this excavation process, the N value of the ground located at the tip of the excavation head 18 during excavation is estimated using the calculated N value calculation model. The computer terminal 30 of the drilling management system 20 starts the drilling by starting the rotation of the drive motor of the auger machine 16 and inserting the excavation head 18 into the ground, similarly to the trial drilling process. In this case, an output screen including the estimation result display area is displayed on the display unit 26. This estimation result display area is an area in which the N value calculated in the estimation process described later is displayed.

  And the control part of the computer terminal 30 is the same as steps S2-1 to S2-3, the actual value output process (step S4-1), the depth association process (step S4-2), the drilling speed, An integral current value calculation process (step S4-3) is executed.

  Next, the control part of the computer terminal 30 performs the estimation process of N value (step S4-4). Specifically, the N value calculation unit 33 of the control unit calculates the N value using the N value calculation model stored in the learning result storage unit 36, the acquired measurement value, and the calculated calculation value. Here, an input element including a current value, a water amount, a drilling speed, and an integrated current value corresponding to the depth (evaluation target) stored in the memory 31a is used as an input layer, and an output layer at the depth (evaluation target) is used. Estimate the N value.

  And the control part of the computer terminal 30 performs the output process of N value (step S4-5). Specifically, the management unit 31 of the control unit outputs the estimated N value to the estimation result display area of the output screen.

  Then, it is determined whether the operator has reached the support layer using the columnar diagram 351 displayed on the output screen, the depth, and the calculated N value. Note that the processes of steps S4-2 to S4-5 are executed every time the actual value acquisition process (step S4-1) is executed.

  Next, the control part of the computer terminal 30 performs the arrival determination process to a support layer (step S4-6). Specifically, the management unit 31 of the control unit receives an instruction to stop drilling by an operator who has determined that the support layer has been reached. Furthermore, the management part 31 receives the raise instruction | indication of the excavation head 18, in order to pull out the excavation head 18 from a pile hole similarly to a trial excavation process. Then, the management unit 31 records each data according to the depth stored in the memory in the measurement value storage unit 35.

[N value calculation model and estimated N value]
FIG. 7A shows an N value estimated using an N value calculation model generated from each value of depth, speed, current value, water amount, and integrated current value as an input layer. Moreover, FIG.7 (b) shows the N value estimated using the N value calculation model produced | generated by each value of speed, an electric current value, water amount, and an integral electric current value without using a depth value in Fig.7 (a). ing. In FIG. 7A and FIG. 7B, the broken line indicates an actually measured N value, and the solid line indicates an evaluation created using a statistical method (least square method or steepest descent method) in order to evaluate the estimated N value. It is a curve. In this embodiment, a B-spline curve is used as the evaluation curve. The evaluation curve is not limited to the B-spline curve, and other curves that can complement the N value (Bézier curve, moving average curve, etc.) may be used.
Comparing the evaluation curve and the distribution of N values, as shown in FIG. 7A, the N value estimated using the N value calculation model including the depth in the input layer has less variation and is stable. The support layer can be determined. Note that the N value calculation model using a plurality of measurement values or measurement values is closer to the actual N value than the N value calculation model using any one of the measurement values and calculation values as the input layer. Was able to be calculated. In FIGS. 7A and 7B, the N value calculated for a depth shallower than 24 m is not displayed.

According to this embodiment, the following effects can be obtained.
(1) In the present embodiment, the N value calculation unit 33 of the computer terminal 30 uses the measurement value of each measuring instrument (21 to 23) at the time of drilling as an input layer using an index value indicating the excavation load as an input element The N value is calculated using the N value calculation model stored in the learning result storage unit 36. Thereby, N value can be estimated and the hardness of the ground can be evaluated.

  (2) In the present embodiment, the N value learning unit 32 of the computer terminal 30 uses the measured value data 355 and the calculated value data 356 acquired in the trial drilling process and the corresponding N value data as teacher data, Machine learning is performed to generate an N value calculation model. Thereby, the N value calculation model can be generated using the numerical value obtained by processing the measurement value.

  (3) In this embodiment, the N value learning unit 32 of the computer terminal 30 stores the depth, current value data R2, water amount data R3, drilling speed data R11, and integrated current value data R21 specified from the elapsed time data R1. An N value calculation model is generated as an input element. Thereby, the exact N value can be estimated using the value which measured excavation load from many sides.

  (4) In the present embodiment, the N value learning unit 32 of the computer terminal 30 generates an N value calculation model using the measurement value information 350 having the same site identifier. As a result, an accurate N value calculation model can be generated using actually measured values at the same site as the ground to be evaluated.

  (5) In the present embodiment, in the excavation process, the N value calculation unit 33 of the computer terminal 30 executes an actual value acquisition process (step S4-1) and an N value estimation process (step S4-4). Thereby, the arrival to the support layer can be determined during drilling.

This embodiment can be implemented with the following modifications. The present embodiment and the following modifications can be implemented in combination with each other within a technically consistent range.
In the above embodiment, the N value learning unit 32 of the computer terminal 30 generates the N value calculation model using the depth, speed, (instantaneous) current value, water amount, and integrated current value as input layers. . The index values indicating the excavation load used as the input layer are not limited to these. For example, an N value calculation model using a plurality of (at least two or more) such as depth of measurement value, current value, water volume, drilling speed calculated from these, integrated current value, excavation resistance value as input elements is generated. Also good.

  Furthermore, vibration information at the time of hole formation may be used as an index value indicating the excavation load used as an input element. In this case, a vibration measuring instrument attached to the operation chamber 13, the roof of the operation chamber 13, or the operation lever in the operation chamber 13 is used. This vibration measuring instrument measures vibration (for example, horizontal vibration) at the mounting location.

  And the management part 31 of the computer terminal 30 performs the analysis process of a vibration characteristic by performing the frequency analysis of the vibration acquired from the vibration measuring device. Specifically, the management unit 31 calculates the magnitude of vibration (for example, maximum amplitude or the like) for each measured frequency band of horizontal vibration.

And the management part 31 produces | generates the vibration characteristic value data which linked | related the magnitude | size of the calculated vibration, and N value and depth at this time. In this case, a vibration characteristic value having a frequency having a strong correlation with the magnitude and N value of vibration calculated by vibration analysis is used.
Specifically, as shown in FIG. 8A, the management unit 31 generates vibration characteristic value data in which the depth is associated with the maximum amplitude (vibration frequency: 5 Hz), and measures the vibration characteristic value data. The value data 355 is stored in the measured value storage unit 35.
In FIG. 8A, the solid line is the vibration characteristic value (maximum amplitude at a frequency of 5 Hz), and the two-dot chain line is the N value.

The N value learning unit 32 learns using the vibration characteristic value corresponding to each depth as one of the teacher data, and stores the learning result (N value calculation model) in the learning result storage unit 36.
And the N value calculation part 33 performs the estimation process of N value (step S4-4). In this case, the N value calculation unit 33 estimates the N value using the vibration characteristic value calculated by the management unit 31 as an index value indicating the excavation load.

For example, FIG. 8B shows an N value estimated using an N value calculation model generated by each value of depth, speed, current value, water amount, integrated current value, and vibration characteristic value in the input layer. In addition, FIG. 8C is estimated using an N value calculation model generated by each value of speed, current value, water amount, integrated current value, and vibration characteristic value without using the depth value in FIG. 8B. N value is shown. In these drawings, the broken line is an actually measured N value, and the solid line is an evaluation curve created using a statistical method in order to evaluate the estimated N value. Note that the N value is not displayed at a depth shallower than 24 m.
The N value calculated by the N value calculation model used in FIGS. 8B and 8C is more than the N value calculated by the N value calculation model used in FIGS. 7A and 7B. The variation is smaller and the support layer can be determined stably.

  Further, the vibration characteristic value used for the input layer for calculating the N value is not limited to the maximum amplitude at the vibration frequency of 5 Hz in the horizontal direction. For example, vibrations in the vertical direction may be used, or statistical values (for example, average values) of the magnitudes of vibrations at all frequencies may be used. In this case, any value related to vibration (vibration characteristic value) determined to have a strong correlation with the N value may be used.

  In the above embodiment, the N value learning unit 32 of the computer terminal 30 uses the N value as the ground evaluation information for the output layer. The ground evaluation information to be output is not limited to the N value for evaluating the hardness of the soil, but may be information on the estimated N value and the geology estimated from the columnar diagram 351 (information for specifying the geology). Moreover, the information which shows that it is the ground of the hardness used as the support layer of a pile instead of estimating N value may be sufficient. In this case, for example, a determination condition for determining that the support layer is a pile is stored. Then, it is determined whether or not the stored determination condition is satisfied based on the index value indicating the excavation load, and the determination result is displayed on the display unit 26.

  In the above embodiment, the N value learning unit 32 of the computer terminal 30 generates the N value calculation model by machine learning using the measurement value data 355 and the calculation value data 356 corresponding to the depth. In this case, the N value calculation model may be generated by machine learning using only the index value in a region deeper than a predetermined depth. Index values may not be stable in shallow areas. For example, in the drilling speed shown in FIG. 6D, the fluctuation is large in a portion shallower than 23 m. Therefore, a more accurate N value calculation model can be generated by performing machine learning using only numerical values in a portion deeper than a reference depth (for example, 24 m) with small fluctuation.

  -In the above-mentioned embodiment, the actual measurement value was acquired from each measuring instrument (21-23) at the time of drilling in an excavation process, and N value was computed. The timing for calculating the N value is not limited to drilling. For example, the N value may be calculated for ground evaluation after the end of drilling. Specifically, the N value corresponding to the depth is estimated using the N value calculation model using the measured value data 355 and the calculated value data 356 acquired at the time of drilling as input layers.

In the above embodiment, the N value calculation model is generated using the measurement value data acquired in the trial digging as teacher data. The measurement value data used for generating the N value calculation model is not limited to the value acquired at the time of trial digging. It may be index value data indicating the excavation load when a hole is formed in the ground including the support layer on which the pile is provided. For example, data acquired when a hole is formed in a boring survey may be used.
In the above embodiment, the computer terminal 30 includes the N value learning unit 32 and the N value calculation unit 33. The learning unit that generates the hardness calculation model and the calculation unit that executes the process of estimating the ground evaluation information may be provided in different computer terminals.

  h0 ... Pile hole, R1 ... Elapsed time data, R2 ... Current value data, R3 ... Water volume data, R11 ... Drilling speed data, R21 ... Integral current value data, 10 ... Excavator, 11 ... Base machine, 12 ... Crawler, DESCRIPTION OF SYMBOLS 13 ... Operation room, 14 ... Mast, 16 ... Auger machine, 17 ... Drilling rod, 18 ... Drilling head, 20 ... Drilling control system, 21 ... Drilling depth measuring instrument, 22 ... Flow measuring instrument, 23 ... Current measuring instrument , 25 ... input unit, 26 ... display unit, 30 ... computer terminal, 31 ... management unit as control unit, 31a ... memory, 32 ... N value learning unit as control unit, 33 ... N value calculation unit as control unit 35 ... Measured value storage unit, 36 ... Learning result storage unit, 350 ... Measured value information, 352 ... N value data, 353 ... Pile hole identifier, 355 ... Measured value data, 356 ... Calculated value data.

Claims (3)

For each depth, an index value indicating the excavation load at the time of excavation, and a measured value storage unit that stores the ground evaluation information in association with each other,
Using the index value stored in the measurement value storage unit as an input layer, and performing machine learning using teacher data using the ground evaluation information at the depth of the index value as an output layer, the ground evaluation A learning unit that executes a learning process for generating a model;
A ground evaluation system comprising: a calculation unit that executes an estimation process for estimating the ground evaluation information to be evaluated using the ground evaluation model and an input element to be evaluated.   In the learning process and the estimation process, the depth, the flow rate of water used for excavation, the current value supplied to the excavator used for excavation, the excavation speed calculated from the excavation time and excavation distance, and the integrated current value of the current value The ground evaluation system according to claim 1, wherein a plurality of input elements among vibration characteristic values for vibrations generated during excavation are used. For each depth, an index value indicating the excavation load at the time of excavation, and a measured value storage unit that stores the ground evaluation information in association with each other,
A learning unit that generates a ground evaluation model using the index value;
A method for evaluating the ground using a ground evaluation system comprising a calculation unit that estimates the ground evaluation information using the ground evaluation model,
The learning unit uses the index value stored in the measurement value storage unit as an input layer, and performs machine learning using teacher data using the ground evaluation information at the depth of the index value as an output layer. And execute the learning process to generate the ground evaluation model,
The said evaluation part performs the estimation process which estimates the ground evaluation information of the said evaluation object using the said ground evaluation model and the input element of evaluation object, The ground evaluation method characterized by the above-mentioned.