Shield tunneling machine tunneling prediction model

Abstract

To provide an excavation prediction model for shield excavator, which improves prediction accuracy in excavation direction by converting data having no meaning by itself included in construction measurement data to be supervised data to meaningful data.SOLUTION: An excavation prediction model predicting deflection of a tip of a shield excavator 20 is generated by machine learning by AI based on construction measurement data transmitted from a shield excavation management system as supervised data. The excavation prediction model is generated by machine learning by AI based on construction measurement data at the time when a plurality of rings N-1 to N-5 on rear side in excavation direction were constructed as explanatory variable and deflection of a tip of the shield excavator 20 when constructing prescribed rings N to N+5 on front side in excavation direction as objective variable. Prescribed data on structure and operation of the shield excavator 20 among the construction measurement data to be explanatory variable is converted by feature amount engineering as synthetic learning data that allows AI to easily perform machine learning.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to an excavation prediction model for a shield machine, and in particular to an excavation prediction model for a shield machine created by machine learning using artificial intelligence in a shield tunneling method. [Background technology]

[0002] The shield tunneling method involves excavating the ground with a rotating cutter while pressing down and stabilizing the face at the tip of a well-known earth pressure or mud water type shield machine using mud soil pressure, mud water pressure, etc., and forming a tunnel in the ground from the starting shaft to the arrival shaft as a segmented tunnel lining is assembled ring by ring behind the shield machine. This method is widely adopted as a method for major tunnel construction in urban areas and plains.

[0003] In addition, shield tunneling machines such as earth pressure type and slurry type used in shield tunneling methods are equipped with a rotating cutter that cuts the face, as well as a partition wall, a cutter drive device, and an earth removal mechanism at the front of a metal outer shell called a skin plate, and a shield jack and an erector device at the rear of the skin plate.The erector device is used to assemble the tunnel lining made of segments ring by ring, and while receiving a reaction force from the assembled tunnel lining, the shield jack pushes out the rotating cutter together with the skin plate, thereby excavating the shield tunnel while cutting the face.

[0004] In this type of shield tunneling method, it is necessary to have the shield machine excavate with precision along the excavation plan extension line set underground, but because construction is carried out underground, various factors such as changes in the geological quality of the surrounding ground, changes in environmental conditions, and the characteristics of each shield machine make it difficult to control its operation so that it excavates with precision along the excavation plan extension line, and generally horizontal, vertical, and directional deviations from the excavation plan extension line occur at the tip of the shield machine. In particular, various factors resulting from the characteristics of each shield machine are called the "habits" of the shield machine at each construction site, and are difficult to grasp accurately, so in the past, the shield machine was excavated by controlling operation based on the experience and intuition of a skilled worker so as to prevent large deviations from the excavation plan extension line.

[0005] Furthermore, in recent years, the use of large amounts of data and AI based on technological innovations in IoT (Internet of Things) and AI (Artificial Intelligence) is expected to bring about the Fourth Industrial Revolution, and in the field of shield tunneling, for example, a shield tunneling planning support system that utilizes AI has been developed (see, for example, Non-Patent Document 1). In the shield tunneling planning support system of Non-Patent Document 1, a reinforcement learning method is used in which the AI ​​self-learns through trial and error to arrive at an optimal solution, making it possible to derive planned values ​​for the operation of the shield machine according to the linear shape of the shield tunnel and a segment placement plan. [Prior art documents] [Non-patent literature]

[0006] [Non-Patent Document 1] Developing a shield tunneling planning support system using AI / Corporate information / Shimizu Corporation, May 25, 2018, [Retrieved June 10, 2019], Internet (URL: https: / / www.shimz.co.jp / company / about / news-release / 2018 / 2018005 / html) [Patent documents]

[0007] [Patent Document 1] Patent Publication No. 2021-14727 Summary of the Invention [Problem to be solved by the invention]

[0008] However, the shield tunneling planning support system in Non-Patent Document 1 merely sets these planned values ​​by performing a pre-simulation of the operation method of the shield machine for the planned linear alignment and the allocation method of multiple segments of different shapes prior to actually carrying out shield tunneling work at the construction site; moreover, the data items (parameters) of the training data for creating the trained model are limited and specific, making it difficult to obtain an accurate predictive model that reflects the characteristics of the shield machine, known as its "habits," which are due to various uncertain factors at each construction site, such as changes in the geology of the surrounding ground, changes in environmental conditions, or the characteristics of each shield machine, and to properly manage the shield machine so that it excavates along the planned extension line set underground.

[0009] Meanwhile, in the case of shield tunneling, shield tunneling management systems are often introduced to grasp the detailed tunneling status according to the progress of the shield machine. Shield tunneling management systems are well-known systems that collect various data, such as survey data during shield tunneling work and measurement data from various sensors installed on the shield machine, and centralize the management of the shield machine. They are capable of predicting the condition of the excavated soil in the ground and the load status of the shield machine based on the results of the time-dependent changes in the measurement data during construction and the statistical processing of the data, and by inputting the survey results, they are also capable of calculating the positions of the shield machine and segments and determining the deviation from the tunneling plan extension line.

[0010] In addition, in the shield tunneling management system, when excavation is performed corresponding to each ring of the lining body made of segments that form the shield tunnel, data on many parameters (data items) is collected, for example, at time intervals of about 5 seconds, or for each jack stroke of about 10 mm by the shield jack, and stored as a large amount of data. By analyzing this large amount of data using AI, it is thought that it will be possible to appropriately manage the shield tunneling machine so that it excavates along the excavation plan extension line set underground, reflecting the characteristics of the shield tunneling machine, known as its "habits," which are due to various uncertain factors originating from changes in the geological and environmental conditions of the surrounding ground at each construction site, or the characteristics of each shield tunneling machine.

[0011] For this reason, the applicant of the present application has proposed, for example, in Patent Document 1, a construction management method for a shield machine that uses the results of machine learning by artificial intelligence using construction measurement data sent from a shield excavation management system as teaching data to manage the excavation of the shield machine. In the construction management method for a shield machine in Patent Document 1, a learning area is set on the excavation plan extension line of the shield machine at the construction site, and in the set learning area, a learned model is created by machine learning with construction measurement data when constructing one or more rings on the rear side of the excavation direction as explanatory variables and the deviation of the tip of the shield machine when constructing a specified ring on the front side of the excavation direction as the objective variable, and in the main excavation area forward of the excavation direction from the learning area, the created learned model is used as a prediction model, and construction measurement data when constructing one or more rings on the rear side of the excavation direction that have been constructed is input, and the deviation of the tip of the shield machine when constructing a specified ring on the front side of the excavation direction is output, so that the excavation of the shield machine is managed while predicting the deviation of the tip of the shield machine from the excavation plan extension line.

[0012] However, in the construction management method for a shield tunneling machine in Patent Document 1, the prediction model created by machine learning using artificial intelligence contains data that is difficult to evaluate individually when creating a model for predicting the direction of the shield tunneling machine, such as the shield jack n stroke (see FIG. 4), the selection of No. n shield jack, the copy cutter n stroke, the copy cutter over-digging position m degrees selection, the cutter torque that is a positive value regardless of the direction of rotation, and the cutter rotation direction that is either "1" or "0" and only one of the values ​​is represented by "1". For this reason, it is desirable to combine these difficult-to-evaluate data and convert them into data that is easier to evaluate, thereby improving the prediction accuracy of the excavation direction of the shield tunneling machine.

[0013] The present invention aims to provide an excavation prediction model for a shield machine that can convert difficult-to-evaluate data contained in construction measurement data sent from a shield excavation management system, which serves as teacher data, into data that is easier to evaluate, thereby improving the accuracy of predicting the excavation direction. [Means for solving the problem]

[0014] The present invention relates to a shield tunneling method, and provides an excavation prediction model for predicting the deviation of the tip of a shield machine, which is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as teacher data. The construction measurement data when constructing one or more rings on the rear side of the excavation direction is used as an explanatory variable, and the deviation of the tip of the shield machine when constructing a specified ring on the front side of the excavation direction is used as a target variable. Among the construction measurement data that serves as an explanatory variable, a variable related to the jack stroke of the shield jack is used as a predetermined data related to the structure and operation of the shield machine, and is converted into the horizontal inclination of the cutter face and the vertical inclination of the cutter face by using the n-stroke of the shield jack as an explanatory variable through feature engineering, and the difference in the jack stroke is decomposed into the horizontal and vertical directions to calculate the inclination of the cutter face, which is used as synthetic learning data that is easy to machine learn using artificial intelligence. In a main excavation area further forward in the excavation direction than the learning area, a server including a computer is caused to function to receive as input the construction measurement data which is an explanatory variable when constructing one or more rings on the rear side in the excavation direction, perform calculations based on the input construction measurement data, and output the deviation of the tip of the shield machine when constructing a predetermined ring on the front side in the excavation direction. The above objective is achieved by providing a model for predicting the tunneling performance of a shield machine.

[0016] The present invention also relates to a shield tunneling method, and an excavation prediction model for predicting the deviation of the tip of a shield tunneling machine, which is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as teacher data, the construction measurement data when constructing one or more rings on the rear side of the excavation direction is used as an explanatory variable, and the deviation of the tip of the shield tunneling machine when constructing a specified ring on the front side of the excavation direction is used as a target variable, and the construction measurement data, which is the explanatory variable, is a variable related to the jack number of a selected shield jack, which is specified data related to the structure and operation of the shield tunneling machine, and is converted into horizontal and vertical moments by feature engineering to use as composite learning data that is easy to machine learn using artificial intelligence, by the following procedures: calculating the moment vectors of the jacks, adding the calculated moment vectors of the jacks to obtain a composite moment vector, and decomposing the obtained composite moment vector into horizontal and vertical directions. In a main excavation area further forward in the excavation direction than the learning area, a server including a computer is caused to function to receive as input the construction measurement data which is an explanatory variable when constructing one or more rings on the rear side in the excavation direction, perform calculations based on the input construction measurement data, and output the deviation of the tip of the shield machine when constructing a predetermined ring on the front side in the excavation direction. The above objective is achieved by providing a model for predicting the tunneling performance of a shield machine.

[0017] Furthermore, the present invention provides a shield tunneling method, an excavation prediction model for predicting the deviation of the tip of a shield tunneling machine, which is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as teaching data, and which predicts the deviation of the tip of a shield tunneling machine when constructing a specified ring on the front side of the excavation direction as the objective variable, and which is created by machine learning using artificial intelligence using the construction measurement data when constructing one or more rings on the rear side of the excavation direction as the explanatory variable, and the deviation of the tip of the shield tunneling machine when constructing a specified ring on the front side of the excavation direction as the objective variable. Among the construction measurement data that serves as the explanatory variable, a variable related to the cutter stroke of the copy cutter is used as specified data related to the structure and operation of the shield tunneling machine, and is converted into the horizontal overexcavation amount and the vertical overexcavation amount as synthetic learning data that is easy to machine learn using artificial intelligence by the following procedures: copy cutter n stroke and copy cutter overexcavation position m degrees selected are used as explanatory variables to calculate the area of ​​the overexcavation amount from the range of the copy cutter [selected] and the [copy cutter stroke], calculate the overexcavation amount vector assuming that all forces are applied to the center of the selected copy cutter, and decompose the calculated overexcavation amount vector in the horizontal and vertical directions. In a main excavation area further forward in the excavation direction than the learning area, a server including a computer is caused to function to receive as input the construction measurement data which is an explanatory variable when constructing one or more rings on the rear side in the excavation direction, perform calculations based on the input construction measurement data, and output the deviation of the tip of the shield machine when constructing a predetermined ring on the front side in the excavation direction. The above objective is achieved by providing a model for predicting the tunneling performance of a shield machine.

[0018] Furthermore, the present invention relates to a shield tunneling method, and a tunneling prediction model for predicting the deviation of the tip of a shield tunneling machine, which is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as teaching data, the construction measurement data when constructing one or more rings on the rear side of the tunneling direction is used as an explanatory variable, and the deviation of the tip of the shield tunneling machine when constructing a specified ring on the front side of the tunneling direction is used as a target variable, and the model was created by machine learning using artificial intelligence, and among the construction measurement data that serves as explanatory variables, three variables, namely, when the cutter is rotating left, when the cutter is rotating right, and cutter torque, are combined into one variable using a specified formula by feature engineering, and the combined variable is converted into directional torque and used as synthetic learning data that is easy to machine learn using artificial intelligence. In a main excavation area further forward in the excavation direction than the learning area, a server including a computer is caused to function to receive as input the construction measurement data which is an explanatory variable when constructing one or more rings on the rear side in the excavation direction, perform calculations based on the input construction measurement data, and output the deviation of the tip of the shield machine when constructing a predetermined ring on the front side in the excavation direction. The above objective is achieved by providing a model for predicting the tunneling performance of a shield machine.

[0019] In addition, in the excavation prediction model for the shield tunneling machine of the present invention, it is preferable that the machine learning is machine learning using a neural network that uses a support vector machine as an algorithm.

[0020] Furthermore, in the shield machine excavation prediction model of the present invention, it is preferable that the deviation of the tip of the shield machine, which is the objective variable, is a horizontal deviation, a vertical deviation, and a directional deviation. Effect of the Invention

[0021] According to the shield tunneling machine excavation prediction model of the present invention, difficult-to-evaluate data contained in the construction measurement data sent from the shield tunneling management system, which serves as teacher data, can be converted into data that is easier to evaluate, thereby improving the prediction accuracy of the excavation direction. [Brief description of the drawings]

[0022] [Figure 1] FIG. 2 is an explanatory diagram of a system network in which an excavation prediction model for a shield machine is created according to a preferred embodiment of the present invention. [Diagram 2] FIG. 1 is an explanatory diagram of the trained model creation process. [Diagram 3] FIG. 13 is an explanatory diagram of a ring report. [Figure 4] FIG. 1 is an explanatory diagram illustrating a plurality of variables (parameters) that serve as training data for creating a trained model. [Diagram 5] FIG. 4 is an explanatory diagram of the positional relationship of construction data. [Figure 6] FIG. 1 is an explanatory diagram of the process of creating a trained model. [Figure 7] FIG. 13 is an explanatory diagram of how to determine the inclination of a cutter face. [Figure 8] This is an explanatory diagram of how to calculate horizontal and vertical moments. [Figure 9]FIG. 13 is an explanatory diagram of how to calculate the amount of over-digging in the horizontal and vertical directions. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] The excavation prediction model of a shield machine according to a preferred embodiment of the present invention is created, for example, in a system network with the configuration shown in Fig. 1. The system network with the configuration shown in Fig. 1 uses a shield excavation management system 10 that manages the construction of a shield machine 20 (see Fig. 2) at a shield construction site, for example, and a preferably cloud server 11 that is connected via a LAN 12 and has a known machine learning tool (software) that can implement a neural network with a support vector machine as an algorithm as AI (Artificial Intelligence) built in, to create a prediction model with, for example, the deviation of the tip of the shield machine 20 as a target variable, thereby enabling appropriate management so that the shield machine 20 excavates along an excavation plan extension line 21 (see Fig. 2) set underground. The excavation prediction model for the shield tunneling machine of this embodiment is preferably created by machine learning using AI, using a large amount of data (construction measurement data) on numerous parameters sent from the shield tunneling management system 10 at the shield construction site and stored in a cloud server 11, in a system network with the configuration shown in Figure 1, and is capable of converting data that is not very meaningful on its own, contained in the construction measurement data that serves as training data, into meaningful data, thereby improving the prediction accuracy of the excavation direction of the shield tunneling machine.

[0024] Here, in this embodiment, for example, a product name "Arigataya" (manufactured by Calculation Studio Co., Ltd.) can be preferably used as a known shield excavation management system 10. The cloud server 11 is configured to include, for example, a computer. The computer includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), an I / F (Interface), a HDD (Hard Disk Drive), a storage means, an input means, a display means, an output means, and the like. The CPU controls machine learning by AI while using the RAM as a work area according to various programs incorporated in the ROM. In addition, the CPU, by incorporating various computer programs in the ROM, functions the storage means, the input means, the display means, the output means, and the like, and stores a large amount of data sent from the shield excavation management system 10 and analysis results by AI, for example, in a database unit, or displays them as predetermined information on, for example, a display 14 at the site via a LAN 12 or a personal computer 13 installed at the site, or outputs them from a printer.

[0025] The excavation prediction model for the shield tunneling machine in this embodiment is a prediction model that predicts the deviation of the tip of the shield tunneling machine 20 in a shield tunneling method, created by machine learning using AI using construction measurement data sent from the shield tunneling management system as teacher data, and preferably created by machine learning using AI using construction measurement data when constructing one or more rings N-1 to N-5 (see Figure 2) on the rear side of the excavation direction as explanatory variables and the deviation of the tip of the shield tunneling machine 20 when constructing a specified ring N to N+5 on the front side of the excavation direction as the objective variable, and among the construction measurement data that serves as explanatory variables, certain data related to the structure and operation of the shield tunneling machine 20 is converted into synthetic learning data that is easy to machine learn by AI using feature engineering, and is used.

[0026] In addition, in this embodiment, the specified data regarding the structure and operation of the shield tunneling machine 20 is preferably a variable regarding the jack stroke of the shield jack, and the synthetic learning data converted by feature engineering is the horizontal inclination of the cutter face and the vertical inclination of the cutter face.

[0027] Furthermore, in this embodiment, the predetermined data regarding the structure and operation of the shield tunneling machine 20 is preferably a variable related to the jack number of the selected shield jack, and the composite learning data converted by feature engineering is horizontal moments and vertical moments.

[0028] Furthermore, in this embodiment, the specified data regarding the structure and operation of the shield tunneling machine 20 is preferably a variable regarding the cutter stroke of the copy cutter, and the synthetic learning data converted by feature engineering is the horizontal overexcavation amount and the vertical overexcavation amount.

[0029] In addition, in this embodiment, the specified data regarding the structure and operation of the shield tunneling machine 20 is preferably three variables: cutter rotating left, cutter rotating right, and cutter torque, and the composite learning data converted by feature engineering is directional torque.

[0030] In this embodiment, at the shield construction site, the construction measurement data during tunnel excavation, which serves as teacher data sent from the shield excavation management system 10, is data relating to a large number of parameters, for example, about 187 items shown in Fig. 4, selected as those relating to directional control of the shield machine 20, preferably based on the knowledge of a shield engineer, from a large number of data items that can be output as "ring reports" (see Fig. 3) by the shield excavation management system 10. These data are collected as a set of data for each ring of the lining that constitutes the shield tunnel, and are sent to the cloud server 11 preferably every 5 seconds, and are stored, for example, in a memory unit. That is, in the shield tunneling method, the face is excavated with the rotating cutter 20a at the tip of the shield machine 20 over an excavation length corresponding to the length of one ring, while the shield jack 20b is extended. Then, the excavation work of the shield machine 20 is temporarily stopped and one ring's worth of tunnel lining made of segments 20e is assembled using the erector device 20d at the rear of the skin plate 20c. Since excavation work is carried out ring by ring, the construction measurement data is preferably processed as a consolidated data for each ring.

[0031] In this embodiment, the selected construction measurement data of preferably 187 data items (parameters) is classified into a total of 44 pieces of excavation data (related to the front body, middle body, rear body, gyrocompass, coordinates, deviation, displacement, etc.) 50A related to grasping the position of the shield machine 20, a total of 120 pieces of excavation data (related to the shield jack, center bending jack, copy cutter, etc.) 50B related to the directional control of the shield machine 20, and a total of 23 pieces of excavation data (related to the recording date and time, cutter bit acceleration, tail clearance, etc.) 50C related to soil quality, etc. These classified data items are collected as data that changes over time by being sent to the cloud server 11 via the shield excavation management system 10, for example, every 5 seconds, as described above, and are stored in, for example, a memory unit. In FIG. 4, the horizontal deviation of the tip of the shield machine 20, the vertical deviation of the tip, and the azimuth deviation (directional deviation) indicated by "*" are the objective variables.

[0032] These construction measurement data are preferably collected with the ring number of the assembly position of the segment 20e as the title. That is, as shown in Fig. 5, depending on the data item, the position where the data is collected is, for example, the position of the face at the tip of the shield tunneling machine 20 where the rotary cutter 20a is placed, or the position where the backfill material is injected behind the skin plate 20c, and the assembly position of each ring by the segment 20e placed in the expansion and contraction part of the shield jack 20b, which is more than the length of one ring. The data of the positions more than the length of one ring are also sent to the shield tunneling management system 10 and collected as data during the excavation work on the face to ensure the ring width at the assembly position of the ring number, and data during the assembly work on the ring at the assembly position of the ring number, with the ring number of the assembly position of the segment as the title.

[0033] Furthermore, in this embodiment, prerequisite data on the structure of the shield machine or the performance of the machinery and equipment is input and registered in advance in the shield tunneling management system 10 or the cloud server 11. Examples of prerequisite data on the structure of the shield machine include the diameter of the cutter face plate, the number and arrangement of shield jacks and bending jacks, etc. These prerequisite data on the structure of the shield machine are not directly input as explanatory variables, but can be combined with construction measurement data during tunnel excavation sent from the shield tunneling management system 10 and normalized by performing feature engineering before being input. Examples of prerequisite data on the performance of the machinery and equipment of the shield machine include the maximum pressure, maximum speed of extension and contraction, and maximum stroke of each shield jack and bending jack, for example. These prerequisite data on the performance of the machinery and equipment of the shield machine are input with the maximum value converted to 1.0, and can be used to normalize and evaluate the construction measurement data during tunnel excavation sent from the shield tunneling management system 10 with values ​​from 0.0 to 1.0.

[0034] Furthermore, in this embodiment, among the construction measurement data during tunnel excavation sent from the shield tunneling management system 10, the measurement data that is processed by feature engineering and normalization and then input can include, for example, data related to the operation of the worker, such as data related to the pressure and stroke of the shield jack and articulating jack, and data related to cutter rotation. Also, for example, data related to the position and direction of the shield tunneling machine can include data related to the horizontal displacement amount, vertical displacement amount, directional displacement amount, etc. Furthermore, for example, data related to the surrounding environment affecting the shield tunneling machine can include data related to the face water pressure, tail clearance, etc.

[0035] In this embodiment, the construction measurement data during tunnel excavation sent from the shield excavation management system 10 includes data that takes positive and negative values. Examples of such data that takes positive and negative values ​​include data related to the pitching angle and rolling angle of the front body, the pitching angle and rolling angle of the rear body, cutter rotation, cutter torque, etc. For example, in the case of cutter torque, where there is a positive side (right rotation) and a negative side (left rotation), the minimum value after normalization can be -0.5 and the maximum value can be +0.5 (the difference between the minimum and maximum is 1.0). When the maximum value of the cutter torque is 300 kN·m and the right rotation is 150 kN·m, the normalized value can be +0.25.

[0036] In this embodiment, at a shield construction site where the above-mentioned shield excavation management system 10 is installed, a highly accurate excavation prediction model for the shield machine is created by machine learning using AI.

[0037] That is, at a given shield construction site, an initial construction section set on the excavation plan extension line 21 of the shield tunneling machine 20, for example, with an extension of 70 rings (1 ring = 1.5 m), is set as the learning area for the first week (for example, 125 m), and as shown in Fig. 2, construction measurement data including synthetic learning data converted by feature engineering when constructing preferably five rings N-1 to N-5 on the rear side of the excavation direction is used as explanatory variables, and a trained model is created in which the deviation (horizontal deviation, vertical deviation, directional deviation) of the tip of the shield tunneling machine 20 when constructing each of six rings N to N+5 on the front side of the excavation direction is used as the objective variable. Here, of the construction measurement data including synthetic learning data serving as explanatory variables sent from the shield tunneling management system 10 when constructing preferably five rings N-1 to N-5 on the rear side of the excavation direction, the predetermined data related to the structure of the shield tunneling machine 20 or the performance of the mechanical equipment is normalized. Furthermore, the deviation of the tip of the shield tunneling machine 20, which is the objective variable, can be preferably the value of the horizontal deviation, vertical deviation or directional deviation of the cutter center at the tip of the shield tunneling machine 20 from the planned excavation extension line 21 of the shield tunnel, calculated according to a predetermined formula from data related to specified surveying data, such as measurement data from a gyrocompass attached to the shield tunneling machine 20, based on construction measurement data obtained when constructing each of the six rings N to N+5 at the front side in the excavation direction.

[0038] Here, when normalizing data regarding the performance of the mechanical equipment of a shield tunneling machine, as mentioned above, it is preferable to normalize the data using values ​​that change between 0 and 1, with 0 representing a state in which the mechanical equipment is not being used and 1 representing a state in which the mechanical equipment is being used at maximum performance.

[0039] For example, it is preferable to set the maximum pressure of the shield jack to 1.0, the pressure at the time of stopping to 0.0, and the pressure at the time of tunneling to be input as a numerical value between 0.0 and 1.0. In addition, for example, when the maximum pressure of the shield jack is 100 kN, 100 kN can be normalized as 1.0 and a non-dimensional value between 0.0 and 1.0, and when the maximum pressure of the shield jack is 5000 kN, 5000 kN can be normalized as 1.0 and a non-dimensional value between 0.0 and 1.0. In this case, it is preferable that the maximum performance of the machinery and equipment, such as the maximum pressure, is not the maximum value in the range of performance that is considered to be actually used, but the maximum capacity (maximum performance) according to the specifications of the machinery and equipment. This makes it possible to properly evaluate the capacity of the machinery and equipment at other shield construction sites, regardless of the construction conditions of the shield tunneling machine when the general-purpose trained model was created.

[0040] Furthermore, in this embodiment, the construction measurement data sent from the shield tunneling management system 10 is preferably converted in the cloud server 11, where predetermined data relating to the structure and operation of the shield tunneling machine 20 is converted into synthetic learning data that is easy for AI to machine learn, by feature engineering. In this embodiment, as described above, preferably one or more of data relating to the horizontal inclination of the cutter face, the vertical inclination of the cutter face, the horizontal moment, the vertical moment, the horizontal overexcavation amount, the vertical overexcavation amount, or the directional torque is converted into synthetic learning data by feature engineering.

[0041] Here, the inclination of the cutter face in the horizontal direction and the inclination of the cutter face in the vertical direction are values ​​calculated from the difference in the horizontal and vertical jack strokes as synthetic learning data that can be easily evaluated by AI machine learning through feature engineering. Specifically, for example, as shown in Figure 7, a total of six variables of n strokes of the shield jack (n: 1, 2, . . . , 6) shown in Table 1 can be used as explanatory variables to calculate the inclination of the cutter face that will be synthetic learning data by decomposing it into the horizontal and vertical directions from the difference in the jack stroke.

[0042] [Table 1]

[0043] In addition, as synthetic learning data that can be easily evaluated by AI machine learning through feature engineering, the horizontal and vertical moments are values ​​calculated by generating a composite moment by detecting whether the jack is on or off, and decomposing the composite vector into the horizontal and vertical directions. Specifically, as shown in Figure 8, for example, a total of 18 variables, No. n shield jack selection (n: 1, 2, . . . , 18) shown in Table 2, are used as explanatory variables to calculate the horizontal and vertical moments that will become synthetic learning data through the following steps: 1) calculate the moment vector of the [selected] jack, 2) add up the calculated moment vectors of the jacks to obtain a composite moment vector, and 3) decompose the obtained composite moment vector into the horizontal and vertical directions.

[0044] [Table 2]

[0045] For example, if 16 shield jacks are installed at equal intervals and the two on the top right are pressed with 100 kN, pressing the two on the top right will have equal weight as a cause of deviation, and not pressing the other 14 will have equal weight as a cause of deviation. However, by converting the position where a circle connecting the centers of the shield jacks on the cross section of the shield tunneling machine 20 intersects with a line 33.75° diagonally to the right from that center into synthetic learning data that represents the moment that occurs when a total of 200 kN is pressed, it becomes possible to obtain meaningful data as a cause of deviation of the shield tunneling machine 20.

[0046] In addition, as synthetic learning data that is easier to evaluate through feature engineering using AI machine learning, the horizontal and vertical over-excavation amounts are values ​​calculated by calculating the over-excavation amount vector after calculating the total over-excavation amount, and then decomposing it into the horizontal and vertical directions. More specifically, as shown in Figure 9, for example, two variables, the n strokes of the copy cutter (n: 1, 2) shown in Table 3, and a total of 24 variables, the m degree selection of the copy cutter over-excavation position (m: 0, 15, 30, . . . , 345), are used as explanatory variables to calculate the horizontal and vertical over-excavation amounts that will become synthetic learning data by the following steps: 1) calculate the area of ​​the over-excavation amount from the range of the copy cutter [selected] and the [copy cutter stroke], 2) calculate the over-excavation amount vector assuming that all forces are applied to the center of the selected copy cutter, and 3) decompose the calculated over-excavation amount vector into the horizontal and vertical directions.

[0047] [Table 3]

[0048] In addition, as synthetic learning data that is meaningful data for AI machine learning through feature engineering, the directional torque is a value obtained by calculating the direction of torque depending on the rotation direction of the cutter face plate. More specifically, using a total of three variables shown in Table 4, namely, cutter rotating left, cutter rotating right, and cutter torque, as explanatory variables, the directional torque that is synthetic learning data can be calculated and summarized into one variable using the following (Equation 1). In (Equation 1), Tdir is the directional torque, Rr is the cutter rotating right, Rl is the cutter rotating left, and T is the cutter torque. Furthermore, the cutter torque takes a positive value regardless of the rotation direction of the cutter. When the cutter is rotating right or left, it is 1 when the cutter is rotating, and 0 when it is stopped. Only one of these two variables is 1 according to the rotation direction of the cutter. The directional torque takes a positive value when the cutter is rotating right, and a negative value when the cutter is rotating left.

[0049] [Table 4]

[0050]

number

[0051] For example, if the cutter face plate rotates to the right and the torque value is 100 kN·m, if the data is entered as is, then right rotation and the torque value of 100 kN·m will have equal weight as causes of the deviation, but in reality it only becomes meaningful as a factor of deviation when it is "rotated right with a torque value of 100 kN·m." For this reason, right rotation is + and left rotation is -, and the torque value of the cutter torque, which is a positive value regardless of the direction of rotation, is converted into synthetic learning data as a directional torque of "+100 kN·m," making it possible to make the data meaningful as a factor of deviation in the shield tunneling machine 20.

[0052] In this embodiment, at a specific shield construction site, machine learning is performed using construction measurement data including synthetic learning data converted by feature engineering as explanatory variables, preferably using a neural network with a support vector machine as an algorithm, to create a trained model in which construction measurement data including synthetic learning data converted by feature engineering for five rings N-1 to N-5 on the rear side in the excavation direction is used as input data, and the deviation of the tip of the shield machine 20 when constructing each of six rings N to N+5 on the front side in the excavation direction is used as output data. In the learning area for the first week, six of these trained models can be created as trained general-purpose models in which the deviation of the tip of the shield machine 20 when constructing each of six rings N to N+5 on the front side in the excavation direction is used as output data (see FIG. 6).

[0053] In addition, in this embodiment, following the first week's learning area, a section similar to the first week's learning area, for example an extension of 70 rings (for example 125 m), is set as the second to fourth week's learning area on the excavation plan extension line 21 of the shield tunneling machine 20, and, similar to the first week's learning area, six trained models are created for each of the second to fourth week's learning areas, as shown in Figure 2, in the same way as the first week's learning area. That is, in each of the learning areas for the second to fourth weeks, as shown in Figure 2, the construction measurement data including the synthetic learning data converted by feature engineering when constructing preferably five rings N-1 to N-5 on the rear side in the excavation direction is used as the explanatory variable, and the deviation of the tip of the shield machine 20 when constructing each of the six rings N to N+5 on the front side in the excavation direction is used as the objective variable, and machine learning is performed preferably by a neural network using a support vector machine as an algorithm to create six trained general-purpose models in which the construction measurement data including the synthetic learning data converted by feature engineering for the five rings N-1 to N-5 on the rear side in the excavation direction is used as input data, and the deviation of the tip of the shield machine 20 when constructing each of the six rings N to N+5 on the front side in the excavation direction is used as output data.

[0054] As a result, in this embodiment, as shown in Figure 6, the prediction period is the first week to the fourth week, the prediction interval is the future six rings (N to N+5) in each prediction period, and a total of 24 trained models are created for outputting deviations, with the tip of each shield machine 20 being the predicted destination point when constructing six rings N to N+5.

[0055] Furthermore, in this embodiment, it is preferable to select the trained model with the most accurate prediction period and prediction interval from these total of 24 trained models and use it as the excavation prediction model for the shield tunneling machine. Here, the selection of the trained model with the most accurate prediction period and prediction interval to be used as the excavation prediction model for the shield tunneling machine can be easily performed, for example, by comparing the output data when construction measurement data including synthetic training data converted by feature engineering, obtained in a learning area in a week other than the week in which one trained model was obtained, is input to the one trained general-purpose model with a known value related to the deviation of the tip of the shield tunneling machine in the construction measurement data obtained in the learning area in the other week.

[0056] Furthermore, according to the excavation prediction model for the shield tunneling machine of this embodiment, which is created by AI machine learning at a specific shield construction site as described above, it is possible to convert difficult-to-evaluate data contained in the construction measurement data sent from the shield tunneling management system, which serves as training data, into data that is easier to evaluate, thereby improving the accuracy of predicting the excavation direction.

[0057] In other words, according to the shield tunneling machine excavation prediction model of this embodiment, certain data related to the structure and operation of the shield tunneling machine 20, which is part of the construction measurement data that serves as explanatory variables, is converted into synthetic learning data that is easy for AI to use in machine learning by feature engineering, and then used.Therefore, by converting data that is difficult to evaluate as explanatory variables into explanatory variables that can be direct factors in the target variable, it becomes easier for the AI ​​to find and evaluate rules, and also makes it easier for machine learning to be used, thereby improving the accuracy of predicting the excavation direction.

[0058] The present invention is not limited to the above embodiment and can be modified in various ways. For example, the construction measurement data that is the explanatory variable for creating the trained model does not necessarily have to be obtained from the five rings immediately preceding the rear side in the excavation direction, and one or more rings on the rear side in the excavation direction can be appropriately selected and used as the construction measurement data obtained from these rings. The ring in which the deviation of the tip of the shield machine, which is the objective variable, is measured does not necessarily have to be one of the six rings immediately following the front side in the excavation direction, and may be the deviation of the tip of the shield machine in a ring seven or more rings forward. Machine learning for creating the trained model, parameter confirmation model, and prediction model does not necessarily have to be based on a neural network that uses a support vector machine as an algorithm, and may be based on various other algorithms that are capable of analyzing large amounts of construction measurement data. [Explanation of symbols]

[0059] 10. Shield tunneling management system 11 Cloud Server 12 LAN 13. Personal Computers 14 Display 20 Shield tunneling machine 20a Rotary Cutter 20b Shield Jack 20c skin plate 20d Erector device 20e Segment 21 Excavation plan extension line 22 Learning Areas N~N+5,N-1~N-5 Ring

Claims

1. In a shield tunneling method, a tunneling prediction model is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as training data, and predicts the deviation of the tip of a shield tunneling machine, The construction measurement data when constructing one or more rings on the rear side of the excavation direction is used as an explanatory variable, and the deviation of the tip of the shield machine when constructing a specific ring on the front side of the excavation direction is used as an objective variable, and the construction measurement data is created by machine learning using artificial intelligence. Among the construction measurement data serving as explanatory variables, a variable relating to the jack stroke of the shield jack is used as a specific data relating to the structure and operation of the shield tunneling machine, and the n-stroke of the shield jack is used as an explanatory variable through feature engineering. The difference in the jack stroke is decomposed into the horizontal and vertical directions to calculate the inclination of the cutter face, and the data is converted into the horizontal inclination and vertical inclination of the cutter face as synthetic learning data that is easy to machine-learn using artificial intelligence. A server including a computer is provided in a main excavation area ahead of the learning area in the excavation direction. This shield tunneling machine excavation prediction model accepts as input the construction measurement data that serves as an explanatory variable when constructing one or more rings on the rear side of the excavation direction, performs calculations based on the input construction measurement data, and functions to output the deviation of the tip of the shield tunneling machine when constructing a specified ring on the front side of the excavation direction.

2. In a shield tunneling method, a tunneling prediction model is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as training data, and predicts the deviation of the tip of a shield tunneling machine, The construction measurement data when constructing one or more rings on the rear side of the excavation direction is used as an explanatory variable, and the deviation of the tip of the shield machine when constructing a specific ring on the front side of the excavation direction is used as an objective variable, and the construction measurement data is created by machine learning using artificial intelligence. Among the construction measurement data serving as explanatory variables, a variable related to the jack number of a selected shield jack is used as predetermined data related to the structure and operation of the shield tunneling machine. The variable is converted into horizontal and vertical moments as synthetic learning data that is easy to machine-learn using artificial intelligence by using the following procedures: calculating the moment vectors of the jacks using feature engineering; adding the calculated moment vectors of the jacks to obtain a composite moment vector; and decomposing the obtained composite moment vector into horizontal and vertical directions. A server including a computer is provided in a main excavation area ahead of the learning area in the excavation direction. This shield tunneling machine excavation prediction model accepts as input the construction measurement data that serves as an explanatory variable when constructing one or more rings on the rear side of the excavation direction, performs calculations based on the input construction measurement data, and functions to output the deviation of the tip of the shield tunneling machine when constructing a specified ring on the front side of the excavation direction.

3. In a shield tunneling method, a tunneling prediction model is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as training data, and predicts the deviation of the tip of a shield tunneling machine, The construction measurement data when constructing one or more rings on the rear side of the excavation direction is used as an explanatory variable, and the deviation of the tip of the shield machine when constructing a specific ring on the front side of the excavation direction is used as an objective variable, and the construction measurement data is created by machine learning using artificial intelligence. Among the construction measurement data serving as explanatory variables, a variable relating to the cutter stroke of the copy cutter is used as a specific data relating to the structure and operation of the shield tunneling machine, and is converted into the horizontal and vertical overexcavation amounts as synthetic learning data that is easy to machine-learn using artificial intelligence by using, through feature engineering, the copy cutter n stroke and the copy cutter overexcavation position m degrees selected as explanatory variables to calculate the area of ​​the overexcavation amount from the range of the copy cutter [selected] and the [copy cutter stroke], and assuming that all forces are applied to the center of the selected copy cutter, the overexcavation amount vector is calculated, and the calculated overexcavation amount vector is decomposed in the horizontal and vertical directions. A server including a computer is provided in a main excavation area ahead of the learning area in the excavation direction. This shield tunneling machine excavation prediction model accepts as input the construction measurement data that serves as an explanatory variable when constructing one or more rings on the rear side of the excavation direction, performs calculations based on the input construction measurement data, and functions to output the deviation of the tip of the shield tunneling machine when constructing a specified ring on the front side of the excavation direction.

4. In a shield tunneling method, a tunneling prediction model is created by machine learning using artificial intelligence using construction measurement data sent from a shield tunneling management system as teacher data, and predicts the deviation of the tip of a shield tunneling machine, The construction measurement data when constructing one or more rings on the rear side of the excavation direction is used as an explanatory variable, and the deviation of the tip of the shield machine when constructing a specific ring on the front side of the excavation direction is used as an objective variable, and the construction measurement data is created by machine learning using artificial intelligence. Among the construction measurement data serving as explanatory variables, three variables, i.e., cutter left rotation, cutter right rotation, and cutter torque, are combined into one variable using a predetermined formula by feature engineering as predetermined data related to the structure and operation of the shield tunneling machine, and are converted into directional torque and used as synthetic learning data that is easy to machine-learn using artificial intelligence; A server including a computer is provided in a main excavation area ahead of the learning area in the excavation direction. This shield tunneling machine excavation prediction model accepts as input the construction measurement data that serves as an explanatory variable when constructing one or more rings on the rear side of the excavation direction, performs calculations based on the input construction measurement data, and functions to output the deviation of the tip of the shield tunneling machine when constructing a specified ring on the front side of the excavation direction.

5. The excavation prediction model for a shield machine according to any one of claims 1 to 4, wherein the machine learning is machine learning using a neural network with a support vector machine as an algorithm.

6. 6. A shield machine excavation prediction model according to any one of claims 1 to 5, wherein the deviation of the tip of the shield machine which is the objective variable is a horizontal deviation, a vertical deviation, and a directional deviation.

Images