JP7147160B2 - Mountain anchoring support device and mountain anchoring support model learning device - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an earth-mounting support device and an earth-mounting support model learning device.

Conventionally, a method for supporting earth retaining work is known. For example, a basic process planning support device that plans basic processes using an expert system is known (for example, Patent Document 1). Based on construction condition data, columnar diagram data, and building shape data, this basic process planning support device is capable of designing earth retaining walls, piers, and shorings, calculating side pressure and stress on each wall, designing members, calculating penetration length, Check the workability of retaining walls and plan retaining walls, supports, and piers.

JP-A-8-166983

In the technique described in Patent Literature 1, an expert system is used to support the mountain retaining design.

However, in the technology described in Patent Document 1, since actual record data of earth retaining work is not taken into consideration, design support for earth retaining work may not be appropriate.

In consideration of the above facts, it is an object of the present invention to appropriately support the design of earth retaining work based on actual record data of earth retaining work.

In order to achieve the above object, an earth retaining support device of the present invention includes an acquisition unit that acquires information regarding conditions of an earth retaining work to be estimated, and information related to the conditions of earth retaining work that is an estimation target acquired by the acquiring unit. , the information on the design of the earth retaining work to be estimated based on the information on the conditions of the earth retaining work for learning and the trained model preliminarily learned from the learning data representing the information on the design of the earth retaining work for learning. and an information output unit for outputting. As a result, it is possible to appropriately support the design of the earth retaining work according to the actual data of the earth retaining work.

The information on the conditions of the earth retaining work of the present invention includes at least one of the ground conditions at the construction site of the earth retaining work and the construction conditions of the earth retaining work, and the information on the design of the earth retaining work includes the earth retaining type of the earth retaining work and At least one of the shank specifications can be included. By outputting the earth retaining type and earth retaining specification according to the actual data of earth retaining work, it is possible to appropriately support the design of earth retaining work.

The learned model of the present invention is learned in advance from learning data representing information about the conditions of earth retaining work for learning and observation values during construction of the earth retaining work for learning, and the information output unit performs the acquisition Based on the information about the conditions of the estimation target earth retaining work acquired by the unit and the learned model, a predicted value at the time of execution of the estimation target earth retaining work can be further output. By outputting a predicted value at the time of execution of the earth retaining work to be estimated according to the actual data of the earth retaining work, it is possible to appropriately support the design of the earth retaining work.

The information on the design of the earth retaining work of the present invention includes an index regarding the quality of the earth retaining work, an index regarding the cost of the earth retaining work, an index regarding the construction period of the earth retaining work, an index regarding the safety of the earth retaining work, and an index regarding the environment of the earth retaining work. A choke indicator representing at least one may be included. Accordingly, it is possible to appropriately support the design of earth retaining work in consideration of at least one of quality, cost, construction period, safety, and environment of earth retaining work according to the actual data of earth retaining work.

The earth retaining support model learning device of the present invention designs earth retaining work from information about conditions of earth retaining work based on learning data representing information about conditions of earth retaining work for learning and information about design of earth retaining work for learning. a learning unit for learning a model for outputting information about the earth retaining work, and obtaining a learned model for outputting information about the design of the earth retaining work from the information about the conditions of the earth retaining work. As a result, it is possible to obtain a trained model for outputting information on the design of earth retaining work.

The learning unit of the present invention sets, as the learning data, information on the earth retaining work obtained from the results of the earth retaining support device of the present invention for the earth retaining work to be estimated, and based on the set learning data. can be used to re-learn the learned model of the earth retaining support device of the present invention. By re-learning the learned model using the actual data of earth retaining work, a highly accurate learned model can be obtained.

The earth retaining support model learning device of the present invention generates change data, which is data obtained by changing the information regarding the conditions of the earth retaining work for learning, to the mechanical model constructed when executing the earth retaining work for learning. a learning data generation unit that receives the input data, acquires a predicted value for executing the earth retaining work for the change data, and generates pseudo-learning data representing the change data and the predicted value for the change data. , the learning unit may learn the model based on the pseudo learning data generated by the learning data generation unit to obtain the trained model. By using the pseudo-learning data, it is possible to obtain a highly accurate trained model even when the amount of learning data is small.

ADVANTAGE OF THE INVENTION According to this invention, the effect that the design support of earth retaining work can be appropriately performed based on the performance data of actual earth retaining work is acquired.

1 is a block diagram showing a schematic configuration of an earth retaining support device according to a first embodiment; FIG. It is a figure which shows an example of the data set for learning. It is a figure which shows an example of the trained model used by 1st Embodiment. It is a figure which shows an example of the learning processing routine of this embodiment. It is a figure which shows an example of the mountain stop support processing routine of this embodiment. It is a figure which shows an example of the trained model used by 2nd Embodiment. FIG. 13 is a diagram showing an example of a trained model used in the third embodiment; FIG. FIG. 11 is a block diagram showing a schematic configuration of an earth retaining support device according to a fourth embodiment; It is a figure which shows an example of the data generation processing routine for pseudo-learning of this embodiment.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail.

<System configuration of mountain retaining support device according to first embodiment>

In this embodiment, an earth retaining support device for supporting the design of earth retaining work in construction work will be described.

In the first embodiment, at the design stage of earth retaining work, a trained model created in advance from learning data is used to determine the earth retaining type (type of earth retaining method and type of earth retaining wall) and earth retaining specification (e.g., earth retaining wall). (size of H steel used for walls, etc.).

In the earth retaining work, it can be said that there is a non-linear relationship between the ground conditions and construction conditions of the earth retaining work and the earth retaining type and earth retaining specifications of the earth retaining work. For example, if there is earth retaining work A and earth retaining work B, even if the construction conditions for earth retaining work A and earth retaining work B are the same, if the ground conditions differ, the earth retaining method, etc. may differ. Therefore, in the present embodiment, information on past earth retaining work is set as learning data, and a learned model is generated based on the learning data. Since the relationship between ground conditions, construction conditions, earth retaining types, and earth retaining specifications is reflected in the learned model, it is possible to appropriately support the design work of earth retaining work.

FIG. 1 is a block diagram showing an example of the configuration of an earth retaining support device 100 according to the first embodiment. The mountain retaining support device 100 can be functionally represented by a configuration including a data receiving section 10, a computer 20, and an output section 30, as shown in FIG.

The data receiving unit 10 receives a plurality of learning data representing a combination of information regarding the conditions of earth retaining work for learning and information regarding the design of the earth retaining work for learning. In addition, the acquisition unit 22 acquires information about the conditions of the earth retaining work to be estimated.

In the present embodiment, actual data of earth retaining work that has already been completed is used as learning data. The information on the conditions of earth retaining work includes the ground conditions at the construction site of earth retaining work (e.g. stratum composition, depth, soil layer name, N value, site address, groundwater level, etc.) and the construction conditions of earth retaining work (e.g. area and root cutting depth, etc.).

The information on the design of earth retaining work includes the earth retaining type (for example, earth retaining method and earth retaining wall) and earth retaining specifications (size of H steel, thickness of earth retaining wall, etc.).

The data reception unit 10 is implemented by, for example, a keyboard, a mouse, or an input/output device that receives input from an external device.

The computer 20 includes a CPU (Central Processing Unit), a ROM (Read Only Memory) that stores programs and the like for realizing each processing routine, a RAM (Random Access Memory) that temporarily stores data, and a memory as storage means. , a network interface, etc. The computer 20 functionally includes an acquisition unit 22 , a learning data storage unit 24 , a learning unit 26 , a trained model storage unit 28 , and an information output unit 29 .

Acquisition unit 22 acquires a plurality of pieces of learning data received by data reception unit 10 . The acquiring unit 22 also acquires the ground conditions and construction conditions of the earth retaining work to be estimated, which are received by the data receiving unit 10 .

A plurality of pieces of learning data acquired by the acquisition unit 22 are stored in the learning data storage unit 24 . For example, as shown in FIG. 2, information on earth retaining construction conditions for learning includes stratum structure, groundwater level, etc., which are examples of ground conditions, and flat area, root cutting depth, etc., which are examples of construction conditions. stored as

In addition, the mountain retaining construction method and the type of mountain retaining wall, which are examples of the mountain retaining type, and the size of the H steel and the thickness of the mountain retaining wall, which are examples of the mountain retaining specifications, are stored as information related to the design of the mountain retaining work for learning.

For example, as shown in FIG. 2, the data for the performance ID "00001" include stratum composition "A1", groundwater level "B1", flat area "C1", root cutting depth "D1", earth retaining method "E1". , the retaining wall “F1”, the size of the H steel “G1”, and the thickness of the retaining wall “H1” are used as one learning data. In this case, for example, stratum structure, earth retaining construction method, earth retaining wall type, and the like are represented by numerical values. For example, the mountain retaining method x1 is expressed as "1", and the mountain retaining method x2 is expressed as "2".

In the data with the actual ID "00001", in the already completed construction work, the stratum composition as information on the conditions of the earth retaining work is "A1", the groundwater level is "B1", the flat area is "C1", and the excavation depth is is "D1", the earth retaining method is "E1", the earth retaining wall is "F1", the size of the H steel is "G1", and the thickness of the earth retaining wall is "H1" as information on the conditions of earth retaining work It is shown that it was

The learning unit 26 learns a model for outputting the earth retaining type and earth retaining specification of earth retaining work from the ground conditions and construction conditions of the earth retaining work based on the plurality of learning data stored in the learning data storage unit 24. to obtain a trained model. Since the learning data is data corresponding to already completed earth retaining work, information on past earth retaining work is reflected in the learned model.

In this embodiment, a trained model as shown in FIG. 3 is generated. For example, as shown in FIG. 3, a neural network can be used as an example of a model, and deep learning can be used as an example of a learning algorithm.

A trained model obtained by the learning section 26 is stored in the trained model storage section 28 .

The information output unit 29 outputs the estimation target earth retaining work based on the ground conditions and construction conditions of the estimation target earth retaining work acquired by the acquisition unit 22 and the learned model stored in the learned model storage unit 28. Outputs the type of mountain fastening and the specifications of the mountain fastening.

As shown in FIG. 3, when the ground conditions and construction conditions of the earth retaining work to be estimated are input to the learned model, the earth retaining type and earth retaining specifications are output. As a result, as the design contents of the earth retaining work, the information about what earth retaining type and earth retaining specification are appropriate is output from the learned model in which the information on the earth retaining work in the past is reflected.

The trained model outputs information in the form of, for example, a 60% probability of the mountain retaining method A, a 30% probability of the mountain retaining method B, and a 10% probability of the mountain retaining method C as the mountain retaining type. Also, as the mountain retaining specification, for example, probability values corresponding to the actual size of the H steel or a plurality of different sizes of the H steel are output.

A construction engineer, who is a user of the earth retaining support device, performs design work for the earth retaining work of interest with reference to the earth retaining type and earth retaining specification output from the trained model.

The output unit 30 outputs, as a result, the earth retaining type and earth retaining specification of the earth retaining work to be estimated, which are output by the information output unit 29 . For example, the output unit 30 is realized by a display.

The construction engineer who performs the design work of the earth retaining work checks the earth retaining type and the earth retaining specification output by the output unit 30, and uses them as a reference for the design work of the earth retaining work. The construction engineer then makes the final design decisions. After the construction engineer completes the design work, the actual earth retaining work will be carried out.

<Action of the mountain retaining support device>

Next, the action of the mountain retaining support device 100 will be described. The mountain retaining support device 100 executes a learning processing routine and a mountain retaining support processing routine.

<Learning processing routine>

When the data reception unit 10 of the earth retaining support device 100 receives input of a plurality of learning data, the data are stored in the learning data storage unit 24 . Then, when the computer 20 of the earth retaining support device 100 receives the instruction signal for execution of the learning process, it executes the learning process routine shown in FIG.

In step S<b>100 , the learning section 26 acquires a plurality of learning data stored in the learning data storage section 24 .

In step S102, the learning unit 26 learns a model for outputting the earth retaining type and earth retaining specification of earth retaining work from the ground conditions and construction conditions of earth retaining work based on the plurality of learning data acquired in step S100. to generate a trained model.

In step S104, the learning unit 26 stores the learned model generated in step S102 in the learned model storage unit 28, and ends the learning processing routine.

<Mounting Support Processing Routine>

When the learned model is stored in the learned model storage unit 28 and the ground conditions and construction conditions of the earth retaining work to be estimated are received by the data receiving unit 10, the earth retaining support device 100 executes the earth retaining support processing routine shown in FIG. Run.

In step S<b>200 , the information output unit 29 acquires the ground conditions and construction conditions of the earth retaining work to be estimated, received by the data receiving unit 10 .

In step S<b>202 , the information output unit 29 reads the learned model stored in the learned model storage unit 28 .

In step S204, the information output unit 29 outputs the estimation target based on the ground conditions and construction conditions of the estimation target earth retaining work acquired in step S200 and the learned model read out in step S202. Outputs the type of earth retaining work and earth retaining specifications.

The output unit 30 outputs, as a result, the earth retaining type and earth retaining specification of the earth retaining work to be estimated, which are output by the information output unit 29 . The construction engineer confirms the type of earth retaining wall and the specification of earth retaining wall output by the output unit 30, and uses it as a reference for the design work of the target earth retaining work.

As described in detail above, in the first embodiment, the estimation target earth retaining work is determined based on the ground conditions and construction conditions of the estimation target earth retaining work and the learned model that has been learned in advance from the learning data. Outputs the type of retaining wall and the specification of retaining wall. As a result, it is possible to appropriately support the design of the earth retaining work based on the actual record data of the earth retaining work.

<System configuration of mountain retaining support device according to second embodiment>

Next, a second embodiment will be described. The second embodiment differs from the first embodiment in that a predicted value for the construction of the earth retaining work to be estimated is further output. The configuration of the second embodiment is the same as that of the first embodiment, so the same reference numerals are given and the description is omitted.

When designing earth retaining works, it is necessary to take into account the predicted values for earth retaining works. A construction engineer usually uses a dynamic model calculated according to construction work to estimate predicted values for earth retaining work (for example, predicted values for earth retaining displacement, strut axial force, anchor axial force, etc.).

A mechanical model is a model of an object for earth retaining work (for example, the ground or earth retaining frame), and is constructed by a construction engineer. For example, an earth retaining wall is regarded as a beam of a structural member, and a mechanical model is constructed to determine the amount of stress and deformation that occurs in the beam when earth and water pressure is applied to the beam.

Construction engineers need to construct a mechanical model according to the target earth retaining work. However, when the ground conditions and construction conditions for earth retaining work are determined, there are cases where it is desired to refer to the design work for earth retaining work by using approximate predicted values calculated from actual data of past earth retaining work.

Therefore, in the second embodiment, a learned model is constructed so as to output the type of earth retaining wall, the specifications of earth retaining wall, and the predicted value relating to the earth retaining work.

Specifically, the learning unit 26 of the second embodiment generates a learned model as shown in FIG. 6 by using actual data of already completed earth retaining work as learning data.

The learning data of the second embodiment is data representing a combination of the ground conditions and construction conditions of the earth retaining work for learning and the observed values during the construction of the earth retaining work for learning. The observed values during the execution of the earth retaining work for learning are the observed values when the earth retaining work was executed in the past.

As shown in FIG. 6, the trained model of the second embodiment, when the ground conditions and construction conditions are input, the type of earth retaining, the specifications of earth retaining, and the predicted values related to earth retaining work (e.g., earth retaining displacement, strut Predicted values of axial force, anchor axial force, etc.) are output.

Therefore, the information output unit 29 of the second embodiment, based on the ground conditions and construction conditions of the earth retaining work to be estimated acquired by the acquisition unit 22 and the learned model shown in FIG. It also outputs the predicted value during the execution of the earth retaining work.

The construction engineer who designs the earth retaining work completes the design work of the earth retaining work by referring to the predicted values at the time of the earth retaining work.

As described in detail above, in the second embodiment, the learned model is prepared in advance from learning data representing ground conditions and construction conditions for learning earth retaining work and observed values during construction of earth retaining work for learning. be learned. Then, in the second embodiment, based on the information about the conditions of the estimation target earth retaining work and the learned model, a predicted value at the time of execution of the estimation target earth retaining work is further output. As a result, by outputting the predicted value for the execution of the earth retaining work to be estimated according to the actual data of the earth retaining work, it is possible to appropriately support the design of the earth retaining work.

<System configuration of mountain retaining support device according to third embodiment>

Next, a third embodiment will be described. In the third embodiment, the first and second points are that after the construction for which design support is performed using the learned model is completed, the learned model is re-learned (feedback) using the actual data of the earth retaining work. 2 embodiment. The configuration of the third embodiment is the same as that of the first embodiment, so the same reference numerals are used and the description thereof is omitted.

In the third embodiment, QCDSE (Quality Cost Delivery Safety Environment) is further learned as additional information. Also, a comprehensive evaluation of the construction work in consideration of QCDSE may be used.

Specifically, the person in charge of earth retaining work evaluates the QCDSE of the earth retaining work that has already been completed, and the score of each item of QCDSE (quality, cost, construction period, safety, and environment of earth retaining work) is , is given to the performance data of the earth retaining work. Then, the person in charge of the earth retaining work inputs the result data of the construction work and the information about the QCDSE into the data reception unit 10 as learning data.

The information on QCDSE represents at least one of the following: an index regarding the quality of earth retaining work, an index regarding the cost of earth retaining work, an index regarding the construction period of earth retaining work, an index regarding the safety of earth retaining work, and an index regarding the environment of earth retaining work. It is an example of an index.

The data reception unit 10 of the third embodiment stores the learning data input by the person in charge of the earth retaining work in the learning data storage unit 24 .

The learning unit 26 of the third embodiment re-learns the trained model based on the learning data containing information on QCDSE stored in the learning data storage unit 24 .

For example, as shown in FIG. 7, the trained model of the third embodiment is configured to also output information on QCDSE in earth retaining work. Therefore, as shown in FIG. 7, when the ground conditions and construction conditions are input to the learned model, the type of earth retaining wall, the specifications of earth retaining wall, the predicted value for earth retaining work, and the information on QCDSE are output.

The information output unit 29 of the third embodiment further outputs information about QCDSE based on the ground conditions and construction conditions of the earth retaining work to be estimated acquired by the acquisition unit 22 and the learned model shown in FIG. Output. As a result, the information on the QCDSE is also output, so that the construction engineer can also consider the information on the quality, cost, construction period, safety, and environment of the earth retaining work to be estimated.

As described in detail above, in the third embodiment, information obtained from the results of the earth retaining work to be estimated is set as learning data, and the trained model is re-learned based on the set learning data. Let Also, the trained model of the third embodiment further outputs information on QCDSE. Accordingly, it is possible to appropriately support the design of earth retaining work in consideration of at least one of quality, cost, construction period, safety, and environment of earth retaining work according to the actual data of earth retaining work. In addition, by re-learning the learned model using actual data of earth retaining work, a highly accurate learned model can be obtained.

<System Configuration of Erection Support Device According to Fourth Embodiment>

Next, a fourth embodiment will be described. The fourth embodiment is different from the first to third embodiments in that a simulation is performed using the dynamic model used in the earth retaining work that has already been completed, and a large amount of learning data is generated. different.

A large amount of training data is required to obtain a highly accurate trained model. However, there are cases in which learning is not properly performed with only the actual data of past earth retaining work.

Therefore, in the fourth embodiment, a simulation is performed based on the dynamic model used in the already completed earth retaining work to generate a large amount of learning data. Specifically, all condition data is input to a dynamic model constructed by a construction engineer, and the condition data and its results are generated as learning data.

FIG. 8 is a block diagram showing an example of the configuration of an earth retaining support device 400 according to the fourth embodiment. The mountain retaining support device 400 can be functionally represented by a configuration including a data reception unit 10, a computer 420, and an output unit 30, as shown in FIG.

The computer 420 includes a CPU, a ROM that stores programs and the like for implementing each processing routine, a RAM that temporarily stores data, a memory as storage means, a network interface, and the like. The computer 420 functionally includes an acquisition unit 422, a learning data generation unit 423, a learning data storage unit 424, a learning unit 426, a trained model storage unit 28, and an information output unit 29. ing.

When acquiring the learning data, the acquiring unit 422 further acquires the dynamic model constructed when performing the earth retaining work for learning.

The learning data generation unit 423 acquires the dynamic model constructed when the earth retaining work for learning is executed. In addition, the learning data generation unit 423 generates change data, which is data obtained by changing the ground conditions and construction conditions of the earth retaining work for learning, by simulation. Then, the learning data generation unit 423 inputs the change data to the acquired dynamic model, acquires the predicted value for the earth retaining work for the change data, and compares the change data with the prediction value for the change data. Generate pseudo-learning data representing

It should be noted that information on QCDSE given by the construction engineer, who is the user, can be added to the simulation results and used as learning data (in the case of inappropriate condition data, the information on QCDSE has a low score). , and training data that produces inappropriate results is generated). As a result, a large amount of learning data can be obtained. In addition, since it is a mechanical model based on actual results, it is possible to obtain learning data whose ground conditions and constants are substantially in line with actual conditions.

The learning data storage unit 424 stores the pseudo learning data generated by the learning data generation unit 423 .

The learning unit 426 generates a trained model based on the learning data and the pseudo-learning data stored in the learning data storage unit 424 .

<Action of the mountain retaining support device>

Next, a pseudo-learning data generation processing routine executed by the mountain retaining support device 400 of the fourth embodiment will be described.

<Learning processing routine>

The data reception unit 10 of the earth retaining support device 400 receives the input of the learning data and the dynamic model of the earth retaining work that has already been completed. Then, the computer 420 of the earth retaining support device 100 executes the pseudo-learning data generation processing routine shown in FIG.

In step S400, the acquiring unit 422 acquires the learning data received by the data receiving unit 10 and the dynamic model constructed when the earth retaining work corresponding to the learning data is executed.

In step S402, the learning data generation unit 423 generates change data, which is data obtained by changing the ground conditions and construction conditions of the learning data acquired in step S400, by simulation.

In step S404, the learning data generation unit 423 inputs the change data generated in step S404 to the dynamic model acquired in step S400. Then, the learning data generation unit 423 acquires the predicted value for the earth retaining work for the change data, which is output from the dynamic model.

In step S406, the learning data generation unit 423 stores the change data and the predicted value output from the dynamic model obtained in step S404 in the learning data storage unit 424 as pseudo learning data.

As described in detail above, in the fourth embodiment, the ground conditions and construction conditions for the earth retaining work corresponding to the learning data are changed with respect to the dynamic model constructed when constructing the earth retaining work for learning. Change data, which is data that has been caused to change, is input, predicted values for construction of earth retaining work for the changed data are obtained, and pseudo-learning data representing the changed data and predicted values for the changed data are generated. Then, in the fourth embodiment, a trained model is obtained based on the generated pseudo-learning data. As a result, by using the pseudo-learning data, it is possible to obtain a highly accurate trained model even when the amount of learning data is small.

The present invention is not limited to the above-described embodiments, and various modifications and applications are possible without departing from the gist of the present invention.

For example, in the above embodiment, a case where a neural network model as an example of a model is learned by deep learning has been described as an example, but the present invention is not limited to this. For example, a model different from the neural network model may be learned by a learning method different from deep learning.

Further, in the above embodiment, the case where the mountain-mounting support device performs the learning process and the mountain-mounting support process has been described as an example, but it is not limited to this. For example, the system may be composed of an earth anchoring support model learning device that performs learning processing and an earth anchoring support device that performs earth anchoring support processing.

Further, in the above embodiment, as an example of the information about the conditions of the earth retaining work, the ground conditions at the construction site of the earth retaining work and the construction conditions of the earth retaining work are used, and as an example of the information about the design of the earth retaining work, the earth retaining type Although the case of using the mountain retaining specification has been described as an example, it is not limited to this. Any information may be used as long as it relates to the conditions of the earth retaining work, and any information may be used as long as it relates to the design of the earth retaining work.

In addition, the ground conditions are input to the learned model of the above embodiment. At this time, if the address of the construction work is input, the boring data stored in the database is read and the stratum structure of the site is output. You can configure your database like this:

Alternatively, learning data may be selected to generate a trained model. For example, in consideration of water level fluctuations in recent years, the age of the learning data may be limited.

Also, the output of the learned model may be restricted. For example, the learning data may be limited to those with a mountain retaining displacement of 3 cm or less so that only mountain retaining types and retaining specifications with a mountain retaining displacement of 3 cm or less are output. Alternatively, it is also possible to sort out only the types and specifications of mountain clamps with a mountain clamp displacement of 3 cm or less from the output from the trained model.

Further, the trained model may be configured so as to feed back the amount of earth retaining displacement observed during earth retaining work and recommend future countermeasures. In this case, observation data and countermeasures actually taken at that time are used as learning data.

In the above description, the program is pre-stored (installed) in a storage unit (not shown), but the program may be recorded on any recording medium such as a CD-ROM, DVD-ROM, micro SD card, or the like. It is also possible to provide it in the form provided.

10 Data reception units 20, 420 Computers 22, 422 Acquisition units 24, 424 Learning data storage units 26, 426 Learning unit 28 Trained model storage unit 29 Information output unit 30 Output units 100, 400 Erection support device 423 Learning data generation Department

Claims (2)

Input data representing at least one of the ground conditions at the construction site of the earth retaining work for learning, the flat area of the earth retaining work, and the root cutting depth of the earth retaining work, and at least one of the earth retaining type and the earth retaining specification of the earth retaining work for learning. Based on the learning data associated with the output data representing , learning by learning the model so that when the input data of the earth retaining work to be estimated is input, the output data of the earth retaining work to be estimated is output a learning unit that obtains a finished model ;
Changed data, which is data obtained by changing the input data of the earth retaining work for learning, is generated by simulation, and the changed data is input to the dynamic model constructed when the earth retaining work for learning is executed. , obtaining a predicted value of at least one of an earth retaining displacement, a strut axial force, and an anchor axial force when executing the earth retaining work for the change data, and representing the change data and the predicted value for the change data a learning data generation unit that generates pseudo learning data ,
The learning unit further learns the model based on the pseudo learning data generated by the learning data generation unit to obtain the trained model.
A model learning device for retaining support. an acquisition unit that acquires input data representing at least one of a ground condition at a construction site of an earth retaining work to be estimated, a flat area of the earth retaining work, and a root cutting depth of the earth retaining work;
By inputting the input data of the earth retaining work to be estimated obtained by the obtaining unit into the learned model trained in advance by the earth retaining support model learning device according to claim 1, the learned model Output data representing at least one of the type of earth retaining work and specification of the earth retaining work of the estimation target earth retaining work, and at least one of earth retaining displacement, strut axial force, and anchor axial force when executing the earth retaining work to be estimated. an information output unit that outputs a predicted value for
An earth retaining support device including.

Images