JP7061891B2 - Operation estimation device and operation estimation method - Google Patents

Description

The present invention relates to an operation estimation device for estimating an operation set value of a shield excavator, and an operation estimation method.

Conventionally, tunnels and the like have been constructed by the soil pressure shield method (mud pressure shield method). The construction environment (soil quality, water pressure, and other conditions of the ground) at the site excavated by the shield excavator changes from moment to moment depending on the excavation position at the site. For this reason, the operator manually shields the excavator while comparing the pre-planned excavation instructions with the measurement data from various measuring devices measured according to the construction environment and monitoring the measurement data. Perform the operation of.
That is, if the operator does not properly control the shield excavator for each of the sites with different construction environments, the accuracy and safety of the excavated tunnel design will decrease.

In addition, the operator develops the experience of controlling the shield excavator in response to changes in the construction environment and improves the skill level by constructing the tunnel at various sites.
When controlling a shield excavator at a site during excavation, a skilled operator applies the knowledge of control in a similar construction environment in the past to control the current site construction environment. .. However, in the case of an operator with a small number of construction sites, in a construction environment that he has never experienced, it is not possible to control an appropriate shield excavator in the construction environment with poor experience and basic operating knowledge.

For this reason, the accuracy and safety of the design of the tunnel to be excavated vary depending on the operator's skill in controlling each shield excavator.
In order to solve this problem, there is a configuration in which the shield excavator is automatically operated based on the measurement data indicating the rotational state of the cutter of the shield excavator and the propulsion state of the propulsion jack during excavation (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 07-71189

In the above-mentioned Patent Document 1, the operation of the shield excavator by a skilled operator cannot be sufficiently reproduced. That is, since the control is performed by comparing the measured data with the set value, the control is appropriately performed according to the ever-changing construction environment of the site, unlike the control based on the experience of a skilled operator. It is not always the case, and it cannot be said that the accuracy and safety of the excavated tunnel design will be improved.

In addition, using a machine learning model in which human emotion analysis and device operation are learned in advance using teacher data, the same measured values as the teacher data are input, and the analysis results and operation settings are estimated. Is commonly done.
Therefore, by using a machine learning model (hereinafter referred to as a learning model) in which the operation of a shield excavator is learned by machine learning using the operation of a skilled operator as teacher data, a set value that performs an operation closer to that of a more skilled operator can be set. It is possible to estimate.

In the above-mentioned learning model, the set value as the operation amount to be given next to operate the shield excavator is obtained based on a plurality of data including the measured values obtained from the shield excavator.

However, even if the operation is close to the operation of a skilled operator, the geology of the actual ground varies depending on the area, so even if the operation is appropriate in one area, it will be appropriate in another area. Is not always. For this reason, there is a concern that the accuracy and safety of the excavated tunnel design will deteriorate if the shield excavator is not properly operated.

The present invention has been made in view of such circumstances, and an object thereof is to be able to appropriately estimate the set value of the operation to be set for the shield excavator according to the ground to be excavated. It is to provide an operation estimation device and an operation estimation method.

In order to solve the above-mentioned problems, the operation estimation device of the embodiment of the present invention performs an operation on the shield excavator by inputting situation data indicating a situation of excavation by the shield excavator into the operation estimation model. An operation estimation unit that estimates the set value of the operation, an evaluation data acquisition unit that acquires evaluation data indicating evaluation of the operation setting value estimated by the operation estimation unit, and the evaluation acquired by the evaluation data acquisition unit. When the evaluation data indicates that the set value of the operation is negatively evaluated based on the data, the situation data corresponds to the operation result data indicating the actual result of the operation set value for the shield excavator. The operation estimation model has a training data generation unit that generates the attached training data, and the operation estimation model is training data in which the set values of the operations performed on the shield excavator are associated with the data indicating the state of excavation by the shield excavator. It is a model that is created by executing machine learning using the above and is updated by executing additional learning using the training data generated by the training data generation unit.

Further, in the operation estimation device according to the embodiment of the present invention, the situation data includes data indicating the situation regarding the earth pressure, and the operation estimation unit controls the earth pressure of the excavated soil in the shield excavator. It is characterized by estimating the set value of the operation for the purpose.

Further, in the operation estimation device according to the embodiment of the present invention, the situation data includes data indicating a situation regarding the direction, and the operation estimation unit is an operation for controlling the direction of excavation in the shield excavator. It is characterized by estimating the set value of.

Further, in the operation estimation method according to the embodiment of the present invention, the operation estimation unit performs an operation on the shield excavator by inputting situation data indicating the state of excavation by the shield excavator into the operation estimation model. The process of estimating the set value, the process of acquiring the evaluation data indicating the evaluation of the set value of the operation estimated by the operation estimation unit, and the learning data generation unit are the evaluation data acquisition unit. When it is shown in the evaluation data that the set value of the operation is negatively evaluated based on the evaluation data acquired by the above-mentioned situation data, the actual result of the set value of the operation for the shield excavator is shown in the situation data. It has a process of generating training data associated with the operation record data shown , and in the operation estimation model, the set value of the operation performed on the shield excavator is associated with the data indicating the state of excavation by the shield excavator. It is a model that is created by executing machine learning using the training data and is updated by executing additional training using the training data generated by the training data generation unit. ..

As described above, according to the present invention, it is possible to more appropriately estimate the set value of the operation to be set for the shield excavator according to the ground to be excavated. As a result, even if the geology of the ground to be excavated varies, appropriate estimation can be performed, and operation support and uniform operation can be achieved.

It is a schematic block diagram which shows the structural example of the shield excavator 10 applied to the operation estimation apparatus 30 of embodiment. It is a block diagram which shows the structural example of the operation estimation apparatus 30 of embodiment. It is a figure which shows the structural example of the information which the operation state data storage part 35 of embodiment stores. It is a figure which shows the example of the screen which causes the operation estimation apparatus 30 of embodiment to acquire evaluation data. It is a flowchart which shows the operation example of the operation estimation apparatus 30 of embodiment.

Hereinafter, the operation estimation device of the embodiment will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram showing a configuration example of a shield excavator 10 to which the operation estimation device 30 of the present embodiment is applied. FIG. 1A shows a conceptual diagram of the shield excavator 10 viewed from the side, and FIG. 1B shows a conceptual diagram of the shield jack 20 for propelling the shield excavator 10 as viewed from the front.
As shown in FIG. 1 (a), the shield excavator 10 assembles a segment by an elector (not shown) at the rear portion of the cylindrical skin plate 11 in the negative direction of the y-axis, and performs a primary lining S. At the same time, it is a mechanism for excavating the ground. In the shield excavator 10, a chamber 12 is provided at the rear portion of the annular and face plate type cutter 16 provided with the cutter bit 15 in the negative direction of the y-axis. A plurality of soil pressure gauges D are installed on the side wall in the chamber 12. The soil pressure gauge D measures the pressure of mud in the chamber 12.
The mud-making material 14 is injected into the chamber 12 from the mud-making soil material injection pipe 13. The excavated soil deposited in the chamber 12 is kneaded by stirring with the mud-making material 14 by a kneading blade (not shown) and converted into mud.
The screw conveyor 17 discharges the mud from the chamber 12 to the conveyor 18 via the soil discharge gate G. Then, the conveyor 18 carries out the mud discharged from the screw conveyor 17 to the outside of the tunnel via the conveyor 19. The gantry M supports the screw conveyor 17, the conveyors 18, and 19.

Further, as shown in FIG. 1 (b), a plurality of shield jacks 20 are provided so as to surround the inner circumference of the skin plate 11 and are arranged between the skin plate 11 and the segment. When the shield jack 20 is propelled (extended) by hydraulic operation, the surface of the skin plate 11 is pushed and the shield excavator 10 propels.

FIG. 2 is a block diagram showing a configuration example of the operation estimation device 30 of the embodiment.
The operation estimation device 30 includes an operation result data acquisition unit 31, a situation data acquisition unit 32, an operation estimation data output unit 33, an evaluation data acquisition unit 34, a trained model storage unit 35, and an operation estimation unit 36. It includes an operation status data storage unit 37 and a learning data generation unit 38.

The operation record data acquisition unit 31 acquires operation record data indicating the record of the operation set value for the shield excavator 10.
Here, the operation performed on the shield excavator 10 includes an operation performed for earth pressure control and an operation performed for direction control. The earth pressure control is a control for stabilizing the ground by balancing the earth pressure in the chamber 12 with the earth pressure and water pressure of the excavated surface. Further, the direction control is the control of the excavation direction of the shield excavator 10 for actually excavating along the predetermined direction.

The operations performed for earth pressure control include, for example, the operation of the injection amount of the mud soil material 14 added to the excavated soil in the chamber 12, and the operation of the rotation speed of the screw of the screw conveyor 17 for discharging the mud in the chamber 12. , And the operation of the opening / closing degree of the earth removal gate G of the screw conveyor 17. That is, the set values of the operation performed for earth pressure control are, for example, the injection amount of the mud-making material 14 added to the excavated soil in the chamber 12, and the rotation of the screw of the screw conveyor 17 for discharging the mud in the chamber 12. The speed and the degree of opening / closing of the earth removal gate G of the screw conveyor 17 are controlled.

The operations performed for the direction control include, for example, an operation of selecting the shield jack 20, an operation of extending the selected shield jack, and the like. That is, the set value of the operation performed for the direction control is, for example, the position of the force point in the digging direction, the selection of the shield jack 20, the propulsive force set in the selected shield jack, and the like.

The operation record data acquisition unit 31 acquires operation record data indicating the record of the set value of the operation for the shield excavator 10. The operation record data acquisition unit 31 stores the acquired operation record data in the operation status data storage unit 37.

The status data acquisition unit 32 acquires status data indicating the status of excavation by the shield excavator 10. The status data acquisition unit 32 stores the acquired status data in the operation status data storage unit 37.
Here, the situation where the shield excavator 10 excavates includes a situation regarding earth pressure and a situation regarding direction.
The earth pressure status includes the earth pressure in the chamber 12 and the earth pressure / water pressure of the excavated surface. That is, the status data regarding the earth pressure is the data indicating the mud pressure in the chamber 12 measured by the earth pressure gauge D and the pressure (indicated earth pressure) acting on the shield excavator 10 in the direction perpendicular to the excavation surface. .. The indicated earth pressure is presented in advance by the excavation instruction sheet, and is determined by taking into consideration the subsidence of the ground and the like, for example, the earth pressure derived based on the depth and geology of the ground to be excavated.

The situation regarding the direction includes the situation of the excavation direction planned in advance and the situation of the actual excavation direction of the shield excavator 10. That is, the situation data regarding the direction is the data indicating the excavation direction planned in advance and the data indicating the actual excavation direction of the shield excavator 10. The data indicating the excavation direction planned in advance includes a designated direction (horizontal direction) and a designated pitch (vertical direction) indicated for each ring by the plan instruction sheet. Further, the data indicating the actual excavation direction of the shield excavator 10 includes the measurement direction (horizontal direction direction) and the measurement pitch (vertical direction direction) measured by the gyroscope (not shown) of the shield excavator 10. ..

The operation estimation data output unit 33 outputs operation estimation data indicating the set value of the operation estimated by the operation estimation unit 36. The operation estimation data output unit 33 outputs the estimation result to the operation screen of the operation room (not shown) of the shield excavator 10, for example, so that the operator recognizes the set value of the estimated operation.

The evaluation data acquisition unit 34 acquires evaluation data indicating evaluation of the set value of the operation estimated by the operation estimation unit 36. The evaluation data acquisition unit 34 stores the acquired evaluation data in the operation status data storage unit 37.
The evaluation data acquisition unit 34 acquires evaluation data indicating the content of evaluation input by the operator via, for example, an input device (for example, a touch panel) in the operation room of the shield excavator 10. For example, when the operator visually recognizes the set value (estimated value) of the operation displayed on the operation screen of the operation room of the shield excavator 10, and the value is the same as the set value that the skilled operator tried to operate. Makes a positive evaluation. Further, for example, the operator visually recognizes the set value (estimated value) of the operation displayed on the operation screen, and if the value is different from the set value that the skilled operator tried to operate, a negative evaluation is made. I do. For example, if it is predicted that the difference between the earth pressure in the chamber 12 and the indicated earth pressure will increase when the set value of the estimated operation is set as it is, the operator gives a negative evaluation. conduct.
The evaluation data acquisition unit 34 outputs the acquired evaluation data to the learning data generation unit 38.

The trained model storage unit 35 stores the operation estimation model. The operation estimation model uses data related to the excavation status of the shield excavator 10 as input data, and executes machine learning using the training data in which the input data is associated with the set value of the operation of the shield excavator 10. It is a model created by this. Here, the data related to the excavation status in the shield excavator 10 as the input data in the learning data is, for example, the data corresponding to the status data acquired by the status data acquisition unit 32 in a general shield machine. In this case, the output data in the training data is, for example, a set value of an operation performed corresponding to the data indicating the excavation situation in a general shield machine.
As the machine learning technique for creating the estimation model, any of commonly used techniques such as decision tree learning, neural network, genetic programming, and support vector machine may be used.

The operation estimation unit 36 estimates the operation set value of the shield excavator 10 by using the operation estimation model stored in the learned model storage unit 35. The operation estimation unit 36 stores the estimated operation setting value of the shield excavator 10 in the operation status data storage unit 37 as operation estimation data.
The operation estimation unit 36 inputs the situation data acquired by the situation data acquisition unit 32 into the operation estimation model, and sets the operation setting value output from the operation estimation model as the operation setting value of the shield excavator 10.

For example, the operation estimation unit 36 inputs data indicating the situation regarding the earth pressure acquired by the situation data acquisition unit 32 into the operation estimation model, and sets the operation setting value for the control of the earth pressure estimated by the operation estimation model. It is a set value of the operation related to the control of earth pressure in the shield excavator 10. In this case, the data indicating the situation regarding the earth pressure are the mud pressure in the chamber 12 measured by the earth pressure gauge D, the indicated earth pressure, the current injection amount of the mud soil material 14, the rotation speed of the screw of the screw conveyor 17, and the earth pressure. The degree of opening and closing of the earth removal gate G and the like. Further, the set values of the operation related to the control of the estimated earth pressure are the set values such as the injection amount of the mud soil material 14 to be set in the future, the rotation speed of the screw of the screw conveyor 17, and the opening / closing degree of the earth removal gate G. Is.

For example, the operation estimation unit 36 inputs data indicating the situation regarding the direction acquired by the situation data acquisition unit 32 into the operation estimation model, and shield excavates the set value of the operation related to the direction control output from the operation estimation model. It is estimated as a set value of an operation related to direction control in the machine 10. In this case, the data indicating the situation regarding the direction is the designated direction, the designated pitch, the current measurement direction measured by the gyroscope of the shield excavator 10, the measurement pitch, and the like, which are indicated by the plan instruction sheet. Further, the set value of the operation related to the control of the estimated direction is the position of the force point in the excavation direction of the shield excavator 10 to be set in the future, the shield jack 20 to be selected, the propulsive force thereof, and the like.

The operation status data storage unit 37 includes status data acquired by the status data acquisition unit 32, operation estimation data estimated by the operation estimation unit 36, evaluation data acquired by the evaluation data acquisition unit 34, and operation performance data acquisition unit. The operation record data acquired by 31 is stored respectively.

The learning data generation unit 38 generates learning data. In the learning data generated by the learning data generation unit 38, the set value (operation record data) of the actually performed operation is associated with the data (situation data) indicating the situation actually excavated by the shield excavator 10. Data.
The learning data generation unit 38 updates the operation estimation model using the generated learning data. Specifically, the learning data generation unit 38 executes machine learning using the generated learning data, and corresponds to the situation of the ground actually excavated by the shield excavator 10 and the situation in the operation estimation model. Learn (additional learning) the setting value of the operation. As a result, the operation estimation model can more appropriately estimate the set value of the operation according to the condition such as the geology of the ground actually excavated by the shield excavator 10.

When generating the learning data, the learning data generation unit 38 refers to the operation status data storage unit 37 and acquires the status data to be used as the learning data and the operation record data. The learning data generation unit 38 may generate learning data using all of the situation data stored in the operation status data storage unit 37, or one of the situation data stored in the operation status data storage unit 37. You may select a part and generate training data.
When the learning data generation unit 38 selects a part of the situation data stored in the operation situation data storage unit 37 to generate the learning data, for example, the situation data is selected based on the evaluation data. The learning data generation unit 38 extracts, for example, the situation data corresponding to the evaluation data showing a negative evaluation, and generates the learning data associated with the operation performance data corresponding to the extracted situation data. That is, the learning data generation unit 38 generates learning data associated with the operation performance data, which is a more ideal operation setting value, for the operation estimation data estimated in the past that has been negatively evaluated. do. By doing so, it is possible to generate training data that can be retrained by the operation estimation model for those that have received a negative evaluation for which the estimation by the operation estimation model is not appropriate.

FIG. 3 is a diagram showing a configuration example of data stored by the operation status data storage unit 37 of the embodiment.
As shown in FIG. 3, for example, as the data stored in the operation status data storage unit 37, the data for each time stamp is stored for each data type.
In the data type, status data, operation estimation data, evaluation data, and operation record data are stored as data types. One or more data items are stored in association with each data stored in the data type.
In this example, N data items of data items # 1 to #N (where N is an arbitrary natural number) are stored in association with the situation data. For example, the data item # 1 of the situation data is the mud pressure in the chamber 12, the # 2 is the indicated earth pressure, and so on.
Further, M data items of # 1 to #M (where M is an arbitrary natural number) are stored in association with the operation estimation data. For example, the data item # 1 of the operation estimation data is the injection amount of the mud soil material 14, # 2 is the rotation speed of the screw of the screw conveyor 17, and so on.
Further, M data items of # 1 to #M (where M is an arbitrary natural number) are stored in association with the evaluation data. The data item # 1 of the evaluation data is an evaluation for the data item # 1 of the operation estimation data.
Further, M data items of # 1 to #M (where M is an arbitrary natural number) are stored in association with the evaluation actual data. The data item # 1 of the operation performance data is data indicating the performance of the operation corresponding to the data item # 1 of the operation estimation data.
In the time stamp, the value measured for the data corresponding to the data item or the value acquired for the data corresponding to the data item is stored at each time T1 to T4.

FIG. 4 is a diagram showing an example of a screen (evaluation screen) for causing the operation estimation device 30 of the embodiment to acquire evaluation data.
As shown in FIG. 4, for example, the evaluation screen includes an image G-1 showing a data type, an image G-2 showing a time-series change for each data type, and an image G-3 of a touch panel for inputting an evaluation. .. For example, the operator visually recognizes the images G-1 and G-2, monitors the degree of separation between the operation estimation data and the operation record data, and the time-series change of the situation data, and obtains an appropriate estimation value for the operation estimation data. Determine if it is shown. For example, when an appropriate estimated value is shown in the operation estimation data, the operator selects a part showing a positive evaluation of the image G-3 (a touch panel displayed as "Good" in the example of FIG. 4). Enter evaluation data showing a positive evaluation by touching. In addition, when the operation estimation data does not show an appropriate estimated value, the operator selects a part showing a negative evaluation of the image G-3 (a touch panel displayed as "Bad" in the example of FIG. 4). Enter evaluation data showing a negative evaluation by touching.

FIG. 5 is a flowchart showing an operation example of the operation estimation device 30 of the embodiment.
First, the operation estimation device 30 outputs operation estimation data based on the operation estimation model (step S10). The operation estimation unit 36 of the operation estimation device 30 inputs the situation data into the operation estimation model, sets the operation set value output from the operation estimation model as the operation estimation data, and outputs the operation estimation data via the operation estimation data output unit 33.
Next, the evaluation data acquisition unit 34 acquires evaluation data indicating evaluation of the operation estimation data input by the operator or the like (step S11). The evaluation data acquisition unit 34 acquires the evaluation data input by operating the operation screen of the touch panel displayed in the operation room (not shown) of the shield excavator 10, for example.
Next, the learning data generation unit 38 determines whether or not the evaluation data shows a negative evaluation (“Bad”) (step S12).
When the evaluation data indicates a negative evaluation, the training data generation unit 38 actually applies the situation data input to the operation estimation model when generating the estimated value that received the negative evaluation by the operator. The training data associated with the operated operation result data is generated (step S13).

As described above, the operation estimation device 30 of the embodiment sets the setting value of the operation to be performed on the shield excavator 10 by inputting the situation data indicating the state of excavation by the shield excavator 10 into the operation estimation model. Based on the operation estimation unit 36 to estimate, the evaluation data acquisition unit 34 that acquires evaluation data indicating the evaluation of the operation set value estimated by the operation estimation unit 36, and the evaluation data acquired by the evaluation data acquisition unit 34. The operation estimation model has a training data generation unit 38 that generates training data, and the operation estimation model is used for the shield excavator with data indicating the excavation situation of the shield excavator (for example, the shield excavator 10 or other shield machine). It is created by executing machine learning using the training data associated with the set values of the operation to be performed, and is updated by executing additional learning using the training data generated by the training data generation unit 38. It is a model to be done. As a result, the operation estimation device 30 of the embodiment can additionally learn based on the situation data when the shield excavator 10 actually excavates, so that the shield excavator can be subjected to additional learning according to the ground to be excavated. It is possible to estimate the set value of the operation to be set.

Further, in the operation estimation device 30 of the embodiment, when the learning data generation unit 38 indicates that the evaluation data negatively evaluates the operation setting value, the operation setting for the shield excavator 10 is set in the situation data. Generate training data associated with operation performance data showing the performance of the value. As a result, the operation estimation device 30 of the embodiment can generate learning data for additional learning in consideration of the content of evaluation, and by focusing on additional learning for those that have received a negative evaluation. , The accuracy of the operation estimation model can be improved.

Further, in the operation estimation device 30 of the embodiment, the evaluation data acquisition unit 34 may acquire the evaluation data by the input operation of the operator. By being evaluated by the operator, data classification for additional learning by statistical analysis or the like becomes unnecessary, and the burden on additional learning is reduced. Further, by being evaluated by the operator, it becomes possible to create learning data for additional learning for each ring and learning data for additional learning for each day.

In the above-described embodiment, the case where the two-step evaluation of "positive evaluation" and "negative evaluation" is performed has been described as an example, but the evaluation does not necessarily have to be two-step evaluation, for example. It may be four stages (for example, four stages in which "Very Good" and "Very Bad" are added in the image G-3 of FIG. 4).

Further, in the above-described embodiment, the case where the operation status data storage unit 37 stores the content of the evaluation (whether it is a “positive evaluation” or a “negative evaluation”) as the evaluation data has been described as an example. , Not limited to this. The operation status data storage unit 37 may store not only the content of the evaluation but also the timing at which the set value of the actual operation corresponding to the evaluation is set when the evaluation data is negative. .. There is a delay between the setting value of the operation and the actual change of the condition such as earth pressure. Therefore, by generating training data using the timing when the actual operation setting value is set, additional learning considering the time from the operation setting value setting to the actual change of the situation is performed. It can be carried out.

Further, the evaluation data may be acquired at a predetermined time interval, or may be acquired at an arbitrary timing. For example, the operator may evaluate at regular time intervals (1 minute, 10 minutes, 1 ring, etc.), or may evaluate at any timing of the operator.

Further, in the above-described embodiment, the case where the learning data generation unit 38 generates learning data for additional learning with respect to the operation estimation data that has received a negative evaluation has been described as an example, but the present invention is not limited to this. For example, the learning data generation unit 38 receives a negative evaluation for the ring when the operation estimation data that has received a negative evaluation has more floors than the operation estimation data that has received a positive evaluation. Learning data for additional learning may be generated for the estimated data.

Further, the learning data generation unit 38 may generate training data for additional learning for all operation estimation data regardless of whether it has received a positive evaluation or a negative evaluation. In this case, the learning data generation unit 38 may generate training data for additional learning by weighting according to whether it has received a positive evaluation or a negative evaluation.

A computer may realize all or part of the processing performed by the operation estimation device 30 in the above-described embodiment. In that case, a program for realizing this function may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read by a computer system and executed. The term "computer system" as used herein includes hardware such as an OS and peripheral devices. Further, the "computer-readable recording medium" refers to a portable medium such as a flexible disk, a magneto-optical disk, a ROM, or a CD-ROM, and a storage device such as a hard disk built in a computer system. Further, a "computer-readable recording medium" is a communication line for transmitting a program via a network such as the Internet or a communication line such as a telephone line, and dynamically holds the program for a short period of time. It may also include a program that holds a program for a certain period of time, such as a volatile memory inside a computer system that is a server or a client in that case. Further, the above program may be for realizing a part of the above-mentioned functions, and may be further realized for realizing the above-mentioned functions in combination with a program already recorded in the computer system. It may be realized by using a programmable logic device such as FPGA.

Although the embodiments of the present invention have been described in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs and the like within a range that does not deviate from the gist of the present invention.

10 ... Shield excavator, 20 ... Shield jack, 30 ... Operation estimation device, 31 ... Operation record data acquisition unit, 32 ... Situation data acquisition unit, 33 ... Operation estimation data output unit, 34 ... Evaluation data acquisition unit, 35 ... Learning Completed model storage unit, 36 ... operation estimation unit, 37 ... operation status data storage unit, 38 ... learning data generation unit.

Claims (4)

An operation estimation unit that estimates the set value of the operation to be performed on the shield excavator by inputting the situation data indicating the excavation situation of the shield excavator into the operation estimation model.
An evaluation data acquisition unit that acquires evaluation data indicating evaluation of the set value of the operation estimated by the operation estimation unit, and an evaluation data acquisition unit.
When the evaluation data indicates that the set value of the operation is negatively evaluated based on the evaluation data acquired by the evaluation data acquisition unit, the situation data indicates that the operation for the shield excavator is performed. It has a training data generation unit that generates training data associated with operation performance data that shows the performance of set values .
The operation estimation model is created by executing machine learning using learning data in which the data indicating the state of excavation by the shield excavator is associated with the set value of the operation performed on the shield excavator. , An operation estimation device characterized by being a model updated by executing additional learning using the training data generated by the training data generation unit. The situation data includes data indicating the situation regarding earth pressure.
The operation estimation device according to claim 1 , wherein the operation estimation unit estimates a set value of an operation for controlling the earth pressure of excavated soil in the shield excavator. The situation data includes data indicating the situation regarding the direction.
The operation estimation device according to claim 1 or 2 , wherein the operation estimation unit estimates a set value of an operation for controlling the direction of excavation in the shield excavator. The process of estimating the set value of the operation performed on the shield excavator by inputting the situation data indicating the excavation situation of the shield excavator into the operation estimation model by the operation estimation unit.
When the evaluation data acquisition unit acquires the evaluation data indicating the evaluation for the set value of the operation estimated by the operation estimation unit, the process,
When the learning data generation unit is shown to negatively evaluate the set value of the operation in the evaluation data based on the evaluation data acquired by the evaluation data acquisition unit, the situation data is described as described above. It has a process of generating training data associated with operation record data showing the record of operation set values for the shield excavator .
The operation estimation model is created by executing machine learning using learning data in which the data indicating the state of excavation by the shield excavator is associated with the set value of the operation to be performed on the shield excavator. , An operation estimation method characterized in that the model is updated by executing additional learning using the training data generated by the training data generation unit.

Images