JP7518371B2 - Determination device, determination program, and learning device - Google Patents

Description

The present invention relates to a determination device, a determination program, and a learning device.

When carrying out pile construction work to install piles into the ground, it is important to understand the geological conditions at the planned construction site. In particular, it is important to understand the depth of the supporting layer that will support the weight of the structure.

When constructing piles at a planned construction site, test excavation is carried out at the planned construction site prior to actual construction. In test excavation, holes are actually excavated and fluctuations in the current value of the excavator's auger motor and soil adhering to the excavation head are collected to confirm whether the pre-surveyed geological information matches the actual geology. It is desirable to carry out test excavation evenly throughout the entire planned construction site, but due to cost and construction time constraints, test excavation is often only carried out in one or a few locations.

As a method for evaluating ground, for example, Patent Document 1 discloses a ground evaluation system that uses an N-value calculation model trained by machine learning using teacher data, in which an input layer has current values and excavation speeds at each depth obtained during test excavation as input elements, and an output layer has N-values obtained at specified depth intervals, to estimate an N-value (hereinafter referred to as estimated N-value) at a specified depth for input elements obtained during actual pile construction.
In addition, for example, in Patent Document 2, whether or not the drilling head has reached the supporting layer is determined based on a change in the current value of the auger motor of the drilling machine, etc.

JP 2019-157346 A Patent No. 4958028

However, in Patent Document 1, N values measured at a predetermined depth interval, typically at 1 m intervals, are used as teacher data, so the number of measurements that can be obtained by drilling one hole is only a few to several tens. Therefore, there is a problem that the accuracy of the generated N value calculation model is low because the number of data obtained by test drilling is small. Therefore, there is a problem that even if an estimated N value is calculated based on the current value, drilling speed, etc. that can be obtained during actual pile construction, a result that reflects the actual geology cannot be obtained.
In addition, in Patent Document 2, since the current value and the N value are not necessarily proportional to each other, the person in charge of the excavation work must visually check the graph of the current value and the overall excavation situation to determine whether the excavation head has reached the supporting layer. Therefore, a considerable amount of experience is required before the person in charge can accurately determine whether the excavation head has reached the supporting layer. For this reason, there is a possibility that variations due to experience and individual differences may occur.
Therefore, an object of the present invention is to assist in determining whether a target audience has been reached.

(1) A determination device according to an aspect of the present invention includes an acquisition unit and a model execution unit. The acquisition unit acquires geological information on a planned construction site and excavation information acquired when excavating the planned construction site. The model execution unit uses geological information on a construction end site and excavation information acquired when excavating the construction end site as input data, and inputs the geological information and the excavation information on the planned construction site to a trained model that has been trained with reach information indicating whether or not the supporting layer has been reached in the excavation situation indicated by the input data as correct answer data, thereby outputting the reach information.
According to the above configuration (1), geological information and excavation information regarding the construction end location are used as input data, and reach information indicating whether or not the supporting layer has been reached in response to the input data is learned as correct answer data, thereby making it possible to generate a highly accurate trained model.
Furthermore, because the trained model has high accuracy, arrival information regarding the planned construction site can be calculated with high accuracy.
Furthermore, since the reach information is highly accurate, the person in charge of excavation work or the manager can accurately determine whether the supporting layer has been reached based on the reach information.

(2) In some embodiments, in the above configuration (1),
The acquisition unit acquires data on test excavation at the planned construction site as the excavation information.
According to the above configuration (2), a trained model trained using excavation information from actual test excavation at the planned construction site can be used, which makes it possible to output arrival information suitable for the geology of the planned construction site.

(3) In some embodiments, in the above configuration (1) or configuration (2),
The excavation information acquired by the acquisition unit includes an integrated current value obtained by integrating an instantaneous current value of the excavation motor, and an expected integrated current value generated from the geological information and the integrated current value. The determination device further includes a determination unit. When arrival information indicating that the supporting layer has been reached is output from the model execution unit, the determination unit determines whether or not a difference between the integrated current value and the expected integrated current value is equal to or less than a threshold value.
According to the above configuration (3), if the difference between the integral current value and the expected integral current value is equal to or less than the threshold, the accuracy of the arrival information is high. If the difference is equal to or less than the threshold, the trained model can be used in the actual construction to accurately determine whether the bearing layer has been reached.

(4) In some embodiments, in any one of the above configurations (1) to (3),
The input data is data relating to a construction completion site having geology whose similarity to the geology of the planned construction site is equal to or greater than a threshold value.
According to the above configuration (4), a trained model for each geological tendency can be created, thereby obtaining arrival information suitable for the geology of the planned construction site.

(5) In some embodiments, in any one of the above configurations (1) to (3),
The input data is data relating to a construction end location whose distance from the planned construction location is within a threshold value.
According to the above configuration (5), a trained model for each area can be created, thereby obtaining access information suitable for the geology of the planned construction site.

(6) In some embodiments, in any one of the above configurations (1) to (5),
The geological information on the planned construction site includes at least one of a columnar diagram showing the state of the geological condition along the depth direction and ground strength information showing the degree of hardness or softness of the ground,
The excavation information regarding the planned construction site includes at least one of an instantaneous current value of an excavation motor, an integrated current value obtained by integrating the instantaneous current value, and an estimated integrated current value generated from the geological information and the integrated current value.
According to the above configuration (6), since various information related to the planned construction site is included, arrival information of the planned construction site can be obtained with high accuracy. In other words, it is possible to support the judgment of the person in charge of excavation work and the manager with high accuracy.

(7) A determination program according to one aspect of the present invention includes a computer, an acquisition unit for acquiring geological information on a planned construction site and excavation information on an excavation status of the planned construction site,
Geological information regarding the construction end location and excavation information obtained when excavating the construction end location are used as input data, and reach information indicating whether the supporting layer was reached under the excavation conditions indicated by the input data is used as correct answer data.By inputting the geological information and excavation information regarding the planned construction location into a trained model, the model functions as a model execution means that outputs the reach information.
According to the above configuration (7), geological information and excavation information regarding the construction end location are used as input data, and reach information indicating whether or not the supporting layer has been reached in response to the input data is learned as correct answer data, thereby making it possible to generate a highly accurate trained model.
Furthermore, because the trained model has high accuracy, arrival information regarding the planned construction site can be calculated with high accuracy.
Furthermore, since the reach information is highly accurate, the person in charge of excavation work or the manager can accurately determine whether the supporting layer has been reached in a short time based on the reach information.

(8) A learning device according to an aspect of the present invention includes a storage unit and a learning unit. The storage unit stores geological information on a construction end location and excavation information from the start of construction to reaching a supporting layer, the excavation information being acquired when the construction end location is excavated. The learning unit uses the geological information and the excavation information as input data, and trains a model using learning data in which reach information indicating whether the supporting layer has been reached in the excavation situation indicated by the input data is used as correct answer data, thereby generating a trained model.
According to the above configuration (8), geological information and excavation information regarding the construction end location are used as input data, and reach information indicating whether or not the supporting layer has been reached in response to the input data is learned as correct answer data, thereby making it possible to generate a highly accurate trained model.
Furthermore, because the trained model has high accuracy, arrival information regarding the planned construction site can be calculated with high accuracy.

(9) In some embodiments, in the above configuration (8),
The learning device includes an acquisition unit and a selection unit. The acquisition unit acquires geological information on a planned construction site. The selection unit selects, from the storage unit, excavation information on a construction completion site similar to the planned construction site. The learning unit selects, from the storage unit, the excavation information on the construction completion site having geology similar to the geological information acquired by the acquisition unit.
According to the above configuration (9), a trained model can be generated according to the planned construction location, and arrival information suitable for the site can be output.

(10) In some embodiments, in the above configuration (9),
The selection unit selects excavation information of a construction completion site having geology whose similarity to the geology of the planned construction site is equal to or greater than a threshold value.
According to the above configuration (9), a trained model with similar geology can be generated, and arrival information suitable for the planned construction site can be output.

(11) In some embodiments, in the above configuration (8) or configuration (9),
The selection unit selects excavation information of a construction end location whose distance from the planned construction location is within a threshold value.
According to the above configuration (11), a trained model with similar geology can be generated, and arrival information suitable for the planned construction site can be output.

The present invention can assist in determining whether a target audience has been reached.

FIG. 1 is a block diagram showing a determination device according to the present embodiment. FIG. 2 is a conceptual diagram of the determination device according to this embodiment during model learning. FIG. 3 is a conceptual diagram of the determination device according to this embodiment when using a model. FIG. 4 is a flowchart showing the operation of the determination device according to this embodiment.

The following describes in detail the determination device, determination program, and learning device according to the embodiments of the present invention with reference to the drawings. Note that in the following embodiments, parts with the same numbers perform the same operations, and will not be described again.

The determination device according to this embodiment will be described with reference to the block diagram of FIG.
The determination device 1 according to this embodiment includes a processing circuit 11, a storage unit 12, and a communication interface 13, which are connected to each other via a bus. The determination device 1 according to this embodiment may be implemented as an administrator terminal in a server, a PC, a tablet type information terminal, a dedicated computer, or the like, or may be implemented in a management device mounted on a pile driver, or may be configured as a standalone device. When the determination device 1 is configured as a standalone device, it may include, for example, an input unit such as a keyboard, a mouse, or a touch panel, and a display unit such as a display.

The processing circuit 11 is realized by any one of a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), etc., or a combination of these. The processing circuit 11 includes an acquisition unit 111, a learning unit 112, a model execution unit 113, a display control unit 114, a selection unit 115, and a determination unit 116.

The acquisition unit 111 acquires location information, geological information, and excavation information regarding the planned construction site. In this embodiment, the "planned construction site" is assumed to be a place where a pile is planned to be installed in the ground, but is not limited thereto and can be applied to work that requires reference to actual measurement data when embedding an object such as an anchor in the ground. The location information may be an address or coordinate information of latitude and longitude. The geological information includes at least one of geological data such as soil classification indicating the state of the geology (e.g., columnar diagram, etc.) and ground strength information indicating the degree of hardness of the ground in the depth direction (e.g., N value, compressive strength, etc.). The geological information may also include the date and time of geological survey, depth information determined to be the supporting layer, groundwater level information, etc.
The excavation information includes at least one of an instantaneous current value, an integral current value, and an estimated integral current value, which are measured from the start of excavation. The integral current value is an integral value obtained by integrating a current value supplied to an auger motor of an excavator at a predetermined depth interval, and is calculated at a predetermined depth interval along the depth direction. The estimated integral current value is an estimated value of an integral current value along the depth direction generated from geological information and the integral current value.
The excavation information may further include data such as excavation date and time, excavator model information, excavation speed, and water supply volume.

The learning unit 112 uses geological information and excavation information on the completed construction site as input data, and generates a trained model by training a network model using reach information indicating whether the supporting layer was reached in the excavation conditions indicated by the input data as correct answer data. The reach information may be flag information that indicates "1" when the supporting layer is reached and "0" when the supporting layer is not reached, or it may be text information such as "reached" or "not reached." In other words, any information that can distinguish between when the supporting layer is reached and when it is not reached will suffice. The pair of input data and correct answer data is called learning data.

The model execution unit 113 inputs geological information and excavation information about the planned construction site into the trained model 23, and outputs arrival information.

The display control unit 114 controls the display data including the arrival information to be displayed on a display device (not shown, but for example a screen via a display or projector) via, for example, the communication interface 13 described below. Note that the display control unit 114 may also cause the display device to display the display data via a display interface (not shown) different from the communication interface 13.

The selection unit 115 selects learning data in accordance with given conditions. A method for selecting learning data will be described later with reference to the flowcharts of FIGS.
The determination unit 116 determines whether the reaching information indicating that the support layer has been reached by the model execution unit 113 indicates that the support layer has been reached. Specifically, when the reaching information indicating that the support layer has been reached is output, the determination unit 116 determines whether the difference between the integral current value and the assumed integral current value is equal to or smaller than a threshold value.

The storage unit 12 is composed of a hard disk drive (HDD), a solid state drive (SSD), etc., and stores construction data that associates location information regarding the construction end location, geological information, and excavation information when excavating the location.
The storage unit 12 also stores location information and geological information related to a planned construction site in association with each other. Furthermore, the storage unit 12 stores learning data, learned models, and the like.

The communication interface 13 is an interface that conforms to a specific communication standard and is used for transmitting and receiving data between the determination device 1 and an external device. Possible communication standards include mobile networks such as 4G and 5G, wireless LANs (Local Area Networks) such as Wi-Fi (registered trademark), wireless networks such as Bluetooth (registered trademark) and NFC (Near Field Communication), and wired networks such as wired LANs and USBs (Universal Serial Bus).

Next, model learning of the determination device 1 according to this embodiment will be described with reference to the conceptual diagram of FIG. 2. The storage unit 12, the learning unit 112, and the selection unit 115 are collectively referred to as the learning device.

The learning unit 112 uses the geological information and excavation information on the construction end location as input data 21 and the reaching information on whether the bearing layer has been reached as correct answer data 22 to train the network model 20 and generate a trained model 23. By training in this manner, the trained model 23 can output a determination result of whether the bearing layer has been reached or not from the value of the excavation information.
The learning data is assumed to be input data 21 containing excavation information when the administrator determines that the support layer has been reached, and reach information (e.g., flag information "1") indicating that the support layer has been reached, as correct answer data 22. This makes it possible to learn what kind of time-series excavation information is obtained when the support layer has been reached, while also learning that the support layer has not been reached in excavation information states other than when the support layer has been reached.
In addition, multiple pieces of excavation information may be prepared in a time series from the start of construction at a predetermined depth or at a predetermined time from the start of excavation as the input data 21. In this case, as the corresponding correct answer data 22, arrival information indicating that the supporting layer has been reached (e.g., flag information "1") may be prepared for the excavation information when it is determined that the supporting layer has been reached, and arrival information indicating that the supporting layer has not been reached (e.g., flag information "0") may be prepared for other excavation information.
In addition, position information regarding the construction end location may also be used as the input data 21.
As a learning method, machine learning may be performed using a general method. For example, a neural network, linear regression, random forest, or the like may be used as a network model. Furthermore, as a neural network, a multi-layer network such as a convolutional neural network (CNN), a multi-layer convolutional neural network (DCNN: Deep CNN), a Residual Network (ResNet), or a DenseNet may be used.

The trained model 23 may be generated for each similar group based on the geological information. For example, the selection unit 115 selects data having similar geology as training data. That is, the selection unit 115 prepares in advance geological information and excavation information for construction end locations having similar geological trends (geological similarity, for example, similarity of columnar charts is equal to or greater than a threshold value) by dividing them into multiple groups. The network model 20 is machine-learned using the geological information and excavation information for each of the multiple groups prepared by the selection unit 115 as input data 21, thereby generating a trained model 23 for each columnar chart trend (geological trend).

The selection unit 115 may also prepare the learning data by dividing it into multiple groups based on similar location information, i.e., geological information and excavation information related to construction end locations with nearby addresses. The network model 20 is similarly trained on machine learning using the learning data for each of the multiple groups grouped by similar location information, thereby generating a trained model 23 for each area.

Of course, the learning data may be prepared in groups according to the combination of the geological information and location information described above. When the learning data is grouped according to the combination of the geological information and location information, the learning data may be selected by weighting the geological information and location information to determine which parameter is to be prioritized. By generating a trained model trained with data having similar geological information and/or location information, it is possible to output arrival information suitable for use.

The learning data may be created according to the planned construction location of a new project. There are two methods for selecting learning data, for example.

As a first selection method, the selection unit 115 selects, as input data 21, geological information and excavation information of the construction end location that is different in location (positional information) but has a similarity to a columnar diagram obtained from the geological information of the planned construction location that is equal to or greater than a threshold, and selects the corresponding arrival information as correct answer data 22. As a result, a trained model with similar geology is generated in accordance with the planned construction location, making it possible to output appropriate arrival information.

As a second method, the selection unit 115 selects the geological information of the construction end location within a certain distance range from the planned construction location as the input data 21, and the corresponding arrival information as the correct answer data 22. When the second method is adopted, a threshold may be set stepwise based on a certain distance range from the planned construction location. Specifically, the selection unit 115 collects geological information and excavation information of the construction end location within a radius of 100 m from the planned construction location as input data, but determines that the number of data pieces of geological information within the radius of 100 m is insufficient (i.e., the number of data pieces is less than a predetermined number). In this case, the selection unit 115 may expand the data collection range of the construction end location to be selected until the number of data pieces becomes equal to or exceeds a predetermined number, such as geological information and excavation information of the construction end location within a radius of 200 m, geological information and excavation information of the construction end location within a radius of 500 m, and so on.
This allows a trained model similar to the planned construction location to be generated, allowing appropriate arrival information to be output.

Next, the determination device 1 according to this embodiment when using a model will be described with reference to the conceptual diagram of FIG.
When in use, the model execution unit 113 of the judgment device 1 inputs geological information of the planned construction site and excavation information during excavation (hereinafter referred to as processing target data) as input data 31 to the learned model 23, and generates reach information indicating whether the supporting layer has been reached under the excavation conditions of the input processing target data as output data 32.

Next, the operation of the determination device 1 according to this embodiment will be described with reference to the flowchart of Fig. 4. Here, it is assumed that the data to be processed are geological information of the planned construction site and excavation information during test excavation.
In step S401, during test excavation, the acquisition unit 111 acquires geological information and excavation information of the planned construction site as data to be processed.

In step S402, the model execution unit 113 inputs the data to be processed into the trained model and outputs the arrival information. At this time, the display control unit 114 may graph the expected integral current value and the integral current value obtained during excavation of the planned construction site, and display the arrival information on a display or the like.
In step S403, the determination unit 116 determines whether the arrival information indicates that the support layer has been reached. If the arrival information indicates that the support layer has been reached, the process proceeds to step S404, and if the arrival information indicates that the support layer has not been reached, the process returns to step S401 and the same process is repeated.
In step S404, the determination unit 116 determines whether the difference between the integral current value included in the excavation information of the data to be processed and the assumed integral current value is equal to or less than a threshold value. If the difference is equal to or less than the threshold value, the process proceeds to step S405. If the difference is greater than the threshold value, the process proceeds to step S406.
In step S405, since it can be determined that the test excavation has acquired data similar to the excavation information of the actual construction, the construction parameters set in the test excavation are adopted as the parameters for the actual construction. The parameters include the length and volume of the enlarged excavation section and the foot protection section. This allows efficient construction management.
In step S406, the learning unit 112 uses the processing target data as learning data for a trained model, and applies it when updating (or re-training) the trained model, or when training a new model.

If a trained model is generated for each similar group, after the processing of step S401, the selection unit 115 may select a trained model corresponding to the data to be processed. For example, a trained model of a group similar to the location information and/or geological information of the data to be processed may be selected.

Specifically, if a trained model 23 is generated for each trend of columnar charts, a trained model 23 trained with training data of a group whose similarity of columnar charts is equal to or greater than a threshold value can be selected based on the geological information (columnar charts) included in the data to be processed. Also, if a trained model is generated for each area, a trained model trained with training data of a group within a certain range from the position indicated by the location information included in the input data 21 can be selected based on the location information. Also, if a trained model is generated based on a combination of location information and geological information, a trained model based on a combination of similar location information and geological information of the data to be processed can be selected.

As another example of applying a trained model, instead of preparing a trained model in advance, a trained model corresponding to the data to be processed may be generated after the data to be processed is acquired.

According to the embodiment described above, a trained model is generated in which geological information and excavation information regarding the construction end location are used as input data, and reach information indicating whether the supporting layer has been reached is used as correct answer data. By inputting geological information and excavation information regarding the planned construction location into the trained model, reach information indicating whether the supporting layer has been reached can be output. This allows an appropriate decision to be made regarding whether the supporting layer has been reached, regardless of individual differences such as the inexperience of the manager, and can assist the user in making decisions.

The instructions shown in the processing procedures shown in the above-described embodiments can be executed based on a software program. A general-purpose computer system can store this program in advance on a recording medium, and by reading the stored program, it is possible to obtain the same effect as that of the above-described identification device. Furthermore, the recording medium in this embodiment is not limited to a medium independent of the computer or embedded system, but also includes a recording medium that stores or temporarily stores a program downloaded from a LAN, the Internet, etc.

1: Determination device, 11: Processing circuit, 12: Storage unit, 13: Communication interface, 20: Network model, 21, 31: Input data, 22: Correct answer data, 23: Trained model, 32: Output data, 111: Acquisition unit, 112: Learning unit, 113: Model execution unit, 114: Display control unit, 115: Selection unit, 116: Determination unit.

Claims (9)

An acquisition unit that acquires geological information regarding a planned construction site and excavation information regarding an excavation status of the planned construction site;
a model execution unit that inputs geological information about a construction end location and excavation information acquired when excavating the construction end location as input data, and outputs reach information indicating whether the supporting layer was reached in the excavation situation indicated by the input data into a trained model that has been trained as correct answer data, by inputting the geological information and the excavation information about the planned construction location;
Equipped with
The geological information includes at least one of a columnar diagram showing a geological state along a depth direction and ground strength information showing a degree of hardness or softness of the ground,
The excavation information includes at least one of an instantaneous current value of an excavation motor and an integrated current value obtained by integrating the instantaneous current value . The determination device according to claim 1, wherein the acquisition unit acquires data during test excavation at the planned construction site as the excavation information. The excavation information acquired by the acquisition unit includes the integrated current value and an estimated integrated current value, which is an estimated value of an integrated current value along a depth direction based on a past integrated current value when assuming that the ground has a geology indicated in the geological information ,
The determination device according to claim 1 or claim 2, further comprising a determination unit that, when arrival information indicating that the support layer has been reached is output from the model execution unit, determines whether or not the difference between the integrated current value and the assumed integrated current value is less than or equal to a threshold value. The determination device according to claim 1 , wherein the input data is data regarding the construction end location when the similarity between the geological information regarding the planned construction location and the geological information regarding the construction end location is equal to or greater than a threshold value. The determination device according to any one of claims 1 to 3, wherein the input data is data regarding a construction end location that is within a threshold distance from the planned construction location. Computer,
An acquisition means for acquiring geological information on a planned construction site and excavation information on an excavation status of the planned construction site;
A judgment program for causing a trained model to function as a model execution means for outputting the reaching information by inputting geological information about a construction end location and excavation information obtained when excavating the construction end location as input data, and reaching information indicating whether or not a supporting layer has been reached in the excavation situation indicated by the input data as correct answer data, by inputting the geological information and the excavation information about the planned construction location ,
The geological information includes at least one of a columnar diagram showing a geological state along a depth direction and ground strength information showing a degree of hardness or softness of the ground,
A judgment program , wherein the excavation information includes at least one of an instantaneous current value of an excavation motor and an integrated current value obtained by integrating the instantaneous current value. A storage unit for storing geological information on a construction end location and excavation information acquired when excavating the construction end location from the start of construction to the reaching of a supporting layer;
a learning unit that uses the geological information and the excavation information as input data, and uses reach information indicating whether or not a bearing layer has been reached in the excavation state indicated by the input data as correct answer data to learn a model using learning data, thereby generating a learned model,
The geological information includes at least one of a columnar diagram showing a geological state along a depth direction and ground strength information showing a degree of hardness or softness of the ground,
A learning device, wherein the excavation information includes at least one of an instantaneous current value of an excavation motor and an integrated current value obtained by integrating the instantaneous current value. An acquisition unit that acquires geological information regarding a planned construction site;
A selection unit that selects the excavation information of the construction end location from the storage unit when the similarity between the geological information on the construction planned location acquired by the acquisition unit and the geological information on the construction end location is equal to or greater than a threshold value,
The learning device according to claim 7 , wherein the learning unit trains the model using the excavation information selected by the selection unit as the input data, and generates the trained model. The learning device according to claim 8 , wherein the selection unit selects the excavation information of the construction end location whose distance from the planned construction location is within a threshold value.