Route prediction device, route prediction method, and route prediction program

Abstract

To provide a device, a method, and a program for route prediction predicting a route of a shield machine with higher accuracy.SOLUTION: A route prediction device of a shield machine is provided with: a difference prediction model generation section generating a learned difference prediction model by making an artificial intelligent mechanically learn difference between a basic model predicted arrival position where the shield machine is predicted to arrive on a basis of a basic model generated to predict the route of the shield machine and an actual arrival position of the shield machine in a learning segment after the start of shielding of the shield machine; a first route prediction section predicting a tip position and a direction of the shield machine using the basic model for a construction segment after the lapse of the learning segment; an error prediction section predicting an error of the tip position and the direction of the shield machine predicted using the difference prediction model; and a second route prediction section predicting a route of the shield machine by applying the error predicted with the error prediction section to the route of the shield machine predicted by the first route prediction section.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a course prediction device, a course prediction method, and a course prediction program. [Background technology]

[0002] In the above technical field, Non-Patent Document 1 discloses that a method for operating a shield machine for a planned alignment including curves and a method for allocating multiple segments of different shapes are simulated in advance to set these planned values. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Shimizu Corporation Develops AI-Based Shield Tunneling Planning Support System, Corporate Information, May 25, 2018, [Retrieved December 17, 2023], Internet (URL: https: / / www.shimz.co.jp / company / about / news-release / 2018 / 2018005 / html) Summary of the Invention [Problem to be solved by the invention]

[0004] However, the technique described in Non-Patent Document 1 above was unable to predict the path of the shield machine with higher accuracy. [Means for solving the problem]

[0005] In order to achieve the above object, a course prediction device according to the present invention comprises: a differential prediction model generation unit that uses artificial intelligence to machine learn the difference between a basic model predicted arrival position that the shield machine is predicted to reach when it is allowed to excavate, based on a basic model generated to predict the path of the shield machine in a learning section set in any section after the shield machine has started excavating, and the actual arrival position of the shield machine, to generate a learned differential prediction model; a first path prediction unit that predicts the tip position and orientation of the shield machine using the basic model in a construction section after the learning section has been passed; an error prediction unit that uses the difference prediction model to predict an error in the predicted tip position and orientation of the shield machine from the difference between the tip position and orientation of the shield machine predicted by the basic model and the actual excavation position; a second path prediction unit that predicts the path of the shield machine by adding the error predicted by the error prediction unit to the tip position and orientation of the shield machine predicted by the first path prediction unit; Equipped with.

[0006] In order to achieve the above object, a course prediction method according to the present invention includes: a differential prediction model generation step in which, based on a basic model generated to predict the path of the shield machine in a learning section set at any section after the shield machine starts excavating, artificial intelligence is trained to learn the difference between the basic model predicted position that the shield machine is expected to reach when the shield machine is allowed to excavate and the actual position that the shield machine actually reaches, thereby generating a learned differential prediction model; a first path prediction step of predicting the tip position and orientation of the shield machine using the basic model in a construction section after the learning section; an error prediction step of using the learned difference prediction model to predict an error in the predicted tip position and orientation of the shield tunneling machine from the difference between the tip position and orientation of the shield tunneling machine predicted by the basic model and the actual excavation position; a second path prediction step of predicting the path of the shield machine by adding the error predicted in the error prediction step to the tip position and orientation of the shield machine predicted in the first path prediction step; Includes.

[0007] Furthermore, in order to achieve the above object, the course prediction program according to the present invention comprises: a differential prediction model generation step in which, based on a basic model generated to predict the path of the shield machine in a learning section set at any section after the shield machine starts excavating, artificial intelligence is trained to learn the difference between the basic model predicted position that the shield machine is expected to reach when the shield machine is allowed to excavate and the actual position that the shield machine actually reaches, thereby generating a learned differential prediction model; a first path prediction step of predicting the tip position and orientation of the shield machine using the basic model in a construction section after the learning section; an error prediction step of using the learned difference prediction model to predict an error in the predicted tip position and orientation of the shield tunneling machine from the difference between the tip position and orientation of the shield tunneling machine predicted by the basic model and the actual excavation position; a second path prediction step of predicting the path of the shield machine by adding the error predicted in the error prediction step to the tip position and orientation of the shield machine predicted in the first path prediction step; to be executed by the computer. [Effects of the Invention]

[0008] According to the present invention, the path of a shield tunneling machine can be predicted with greater accuracy. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a diagram for explaining an outline of the operation of a course prediction device according to a preferred embodiment of the present invention; [Figure 2]1 is a block diagram illustrating the configuration of a course prediction device according to a preferred embodiment of the present invention. [Figure 3] 1 is a diagram illustrating a hardware configuration of a course prediction device according to a preferred embodiment of the present invention. [Figure 4] 3 is a flowchart illustrating a processing procedure of a course prediction device according to a preferred embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present invention will be described in detail by way of example with reference to the drawings. However, the configurations, numerical values, processing flows, functional elements, etc. described in the following embodiments are merely examples, and are open to modification and alteration, and are not intended to limit the technical scope of the present invention to the following description.

[0011] A path prediction device 100 according to a preferred embodiment of the present invention will be described with reference to Figs. 1 to 5. Fig. 1 is a diagram for explaining an overview of the path prediction device 100 according to this embodiment. The path prediction device 100 is a device that predicts the error between an arrival position and a target arrival position according to the operation details of the operator of the shield tunneling machine 113, and predicts the path of the shield tunneling machine 113 according to the operation details input by the worker. The path prediction device 100 may be mounted on the shield tunneling machine 113, or may be a server, PC, mobile terminal, or the like that can communicate with the shield tunneling machine 113 via wired or wireless communication.

[0012] As shown in Figure 1, a learning section 110 is provided in the initial excavation area on the planned alignment of the shield machine 113. In other words, the learning section 110 is provided in the area immediately after the shield machine 113 starts excavation. In the learning section 110, the path prediction device 100 causes the shield machine 113 to excavate using a basic model 111. The basic model 111 is a general-purpose excavation model used to cause the shield machine 113 to excavate.

[0013] Then, while shield machine 113 is being driven to excavate using basic model 111 in learning section 110, the artificial intelligence is made to learn the quirks of shield machine 113 (characteristics when excavating straight and curved sections) and the specific conditions of the construction site (soil hardness, viscosity, etc.). In other words, even if equal force is applied to all jacks to make shield machine 113 move in a straight line, it may not move in a straight line, but may excavate with a slight deviation in any direction, for example, up, down, left, or right. By making the artificial intelligence learn the quirks specific to shield machine 113 and the characteristics of the site, it is possible to grasp the tendency of shield machine 113's path, and excavation can be carried out using shield machine 113 as planned, even without a veteran engineer.

[0014] For example, consider a case where the path of the shield machine 113 is currently off by 50 mm at the tip of the shield machine 113 and the path of the shield machine 113 needs to be returned to the planned path. In such a case, some engineers think that it is sufficient to return to the planned path two rings ahead, while others think that it is sufficient to return to the planned path five rings ahead, but in reality, this will differ depending on the cause of the deviation and the stock status of segments for fine adjustment. In this way, the parameters to be input to excavate the shield machine 113 vary depending on the condition of the shield machine 113 and the soil quality, so each engineer's judgment may differ.

[0015] Therefore, the path prediction device 100 uses artificial intelligence to learn the difference between the position predicted to be reached by the basic model 111, which is a model for excavation by a general-purpose shield tunneling machine 113, and the position actually reached by the shield tunneling machine 113, and generates a difference prediction model 112.

[0016] Then, in the construction section 120, the path prediction device 100 predicts the path of the shield tunneling machine 113 by adding the generated differential prediction model 112 to the basic model 111. Therefore, it becomes possible to make highly accurate predictions even at sites where accurate predictions cannot be made using only the general-purpose basic model 111.

[0017] In this way, by setting up a learning section 110 in an area immediately after the start of actual excavation or in an area during excavation and having the shield tunneling machine 113 learn the characteristics of the site, it becomes possible to perform highly accurate excavation even at sites where accurate excavation is not possible using only the general-purpose basic model 111.

[0018] Next, the configuration of the path prediction device 100 will be described with reference to Fig. 2. The path prediction device 100 has a differential prediction model generation unit 201, a first path prediction unit 202, an error prediction unit 203, and a second path prediction unit 204.

[0019] The differential prediction model generation unit 201 uses artificial intelligence to machine-learn the difference between the predicted basic model arrival position predicted by the basic model 111 generated to cause the shield tunneling machine 113 to excavate in a learning section 110 set at any section after the shield tunneling machine 113 starts excavating, and the actual arrival position of the shield tunneling machine 113 when the same operations as input to the basic model are performed, thereby generating a learned differential prediction model 112.

[0020] Basic model 111 predicts the path of the shield machine using a calculation method devised based on conventional methods, with input data including parameters such as the shield jack pressure input to the propulsion jack and articulating jack of the shield machine and the articulating angle caused by the extension and contraction of the articulating jack. Based on this, basic model 111 can be made into a model (basic model 111) that can predict the path of shield machine 113 with a certain level of accuracy or higher by correcting the theoretical formula, for example by adding a coefficient to the calculation method, with reference to the operation of shield machine 113 by a veteran engineer.

[0021] In this embodiment, the basic model 111 uses the above calculation method to predict the path of the shield machine 113 based on parameters (rotational moment and articulation angle) that have a significant effect on the excavation distance and direction of the shield machine 113. Here, the rotational moment is calculated as a moment from the thrust direction and thrust force based on the arrangement and operating status of the thrust jacks (which jacks are in which positions and with what pressure they are pushing the segments). The articulation angle is calculated by accumulating the thrust direction due to the articulation angle between the front and rear bodies, starting from the point where the path changes, and the effective articulation angle is calculated from the articulation angle caused by the extension of the articulation jacks and the delay distance (the distance from when the articulation jacks are extended until the direction actually changes), and this is compared with the excavation vector (D n-1 ) is used as input data, and the current excavation vector (D n ) is calculated. The amount of overcutting (extension of the copy cutter) also has a large effect on the excavation direction, but since this is a process carried out before bending, it is difficult to determine its effect, and so it is not used as a parameter. However, if a method is established that can determine the effect of overcutting, this does not prevent it from being used as a parameter.

[0022] To ensure that the model can reliably learn the characteristics of the shield machine 113 and the characteristics specific to the site, the input data for the rotational moment and bending angle, such as the shield jack placement and operating status, bending angle, and delay distance, which were input parameters in the basic model 111, are given lower weighting coefficients in advance than the other major parameters in the differential prediction model 112. In other words, this is to take into account the influence of parameters other than the rotational moment and bending angle as a cause of error. Note that the input data for the amount of overexcavation, such as the position and length for extending the copy cutter, which was excluded from the basic model 111, is given a weighting coefficient through learning in the same way as the other parameters.

[0023] The basic model predicted arrival position is the position that the shield machine 113 is predicted to arrive at if it is caused to excavate according to the basic model 111. The actual arrival position is the position that the shield machine 113 actually arrives at if it is caused to advance according to the basic model 111.

[0024] Furthermore, because basic model 111 is used to predict the path of shield machine 113 in learning section 110, predictions can be made with a certain degree of accuracy even in learning section 110. As a result, extreme outliers are less likely to occur in error values ​​and they can be kept within a certain range, improving the accuracy of differential prediction model 112.

[0025] The basic model 111 is not trained, and the same model is used at the next site. That is, if the basic model 111 used changes, the prediction results will also change, so it is desirable to use it as a set with the generated differential prediction model 112. Repeated training of the differential prediction model 112 can improve the error prediction accuracy.

[0026] The differential prediction model generation unit 201 then uses the artificial intelligence to further machine-learn weighting coefficients for each of the parameters input to cause the shield machine 113 to excavate in the learning section 110, thereby generating a learned differential prediction model 112. By learning the weighting coefficients, the artificial intelligence can learn the parameters that are important for causing the shield machine 113 to excavate. In this way, the differential prediction model generation unit 201 can generate a differential prediction model for the shield machine 113 by assigning importance to the parameters.

[0027] The learning section 110 may be set to the section immediately after the shield machine 113 starts excavating, or may be set to the section after the shield machine 113 has excavated a predetermined distance after starting excavation. In other words, the learning section 110 can be set to any section between the departure shaft and the arrival shaft.

[0028] First path prediction unit 202 uses basic model 111 to predict the tip position and orientation of shield machine 113 in the construction section after learning section 110 has been completed. As described above, after learning section 110 has been completed, basic model 111 can be used to predict the basic path of shield machine 113, so first path prediction unit 202 predicts the path of shield machine 113 using basic model 111. The predicted path of shield machine 113 is, for example, the tip position and orientation (excavation direction) of shield machine 113. In other words, first path prediction unit 202 predicts which direction shield machine 113 is facing and how far it has traveled.

[0029] Error prediction unit 203 uses differential prediction model 112 to predict errors in the predicted tip position and orientation of shield machine 113 from the difference between the tip position and orientation of shield machine 113 predicted by basic model 111 and the actual excavation position. In other words, error prediction unit 203 predicts errors in the excavation of shield machine 113 based on basic model 111 from the deviation (difference) between the path of shield machine 113 predicted by basic model 111 and the target excavation position predicted by differential prediction model 112. In other words, error prediction unit 203 predicts the path of shield machine 113, which changes depending on the excavation habits of shield machine 113 and the soil quality (characteristics) of the earth and sand being excavated, and predicts errors.

[0030] Second path prediction unit 204 predicts the tip position and orientation of the shield machine by adding the error predicted by error prediction unit 203 to the tip position and orientation of the shield machine predicted by first path prediction unit 202, and predicts the path of the shield machine in construction section 120. In other words, second path prediction unit 204 predicts the path of shield machine 113 by taking into account the deviation (error) from the actual destination position to the path that would be taken if shield machine 113 were made to excavate using basic model 111 (basic model destination position). Second path prediction unit 204 more accurately predicts, for example, the tip position of shield machine 113 several rings ahead and the orientation (excavation direction) of shield machine 113 at that time as a path prediction for shield machine 113. Then, based on this course prediction, the operator of the shield tunneling machine 113 selects the operation details (such as the pressure of each jack and the extension length) for the shield tunneling machine 113 to excavate, and causes the shield tunneling machine 113 to excavate.

[0031] Next, the hardware configuration of the trajectory prediction device 100 will be described with reference to FIG. 4. The CPU (Central Processing Unit) 410 is a processor for arithmetic control, and by executing programs, it realizes the various functional components of the trajectory prediction device 100 shown in FIG. 2. The CPU 410 may have multiple processors and execute different programs, modules, tasks, threads, etc. in parallel. The ROM (Read Only Memory) 420 stores fixed data such as initial data and programs, as well as other programs. The network interface 430 communicates with other devices via a network. The CPU 410 is not limited to a single CPU, and may include multiple CPUs or a GPU (Graphics Processing Unit) for image processing. The network interface 430 preferably has a CPU independent of the CPU 410 and writes and reads transmitted and received data to and from an area in the RAM (Random Access Memory) 440. It is also preferable to provide a DMAC (Direct Memory Access Controller) (not shown) for transferring data between the RAM 440 and the storage 450. The CPU 410 recognizes that data has been received or transferred to the RAM 440 and processes the data accordingly. The CPU 410 also prepares the processing results in the RAM 440, and leaves the subsequent transmission or transfer to the network interface 430 or DMAC.

[0032] The RAM 440 is a random access memory used by the CPU 410 as a work area for temporary storage. The RAM 440 has a storage area reserved for storing data necessary for implementing this embodiment. The basic model predicted arrival position data 441 is data on the position predicted to be reached when the shield tunneling machine 113 is caused to excavate using the basic model 111 in the learning section 110. The error data 442 is data relating to the error between the target arrival position and the position predicted to be reached using the basic model 111. The predicted arrival position data 443 is data on the predicted arrival position of the shield tunneling machine 113 in the construction section 120 after the learning section, with the error due to the basic model 111 added.

[0033] The transmitted / received data 444 is data transmitted and received via the network interface 430. The RAM 440 also has an application execution area 445 for executing various application modules.

[0034] The storage 450 stores databases and various parameters, as well as the following data or programs required to implement this embodiment.

[0035] Storage 450 further stores a differential prediction model generation module 451, a first path prediction module 452, an error prediction module 453, and a second path prediction module 454. Differential prediction model generation module 451 is a module that uses artificial intelligence to machine-learn the difference between a basic model arrival position reached by shield machine 113 as the shield machine 113 excavates and a target arrival position of shield machine 113, based on basic model 111 generated for excavating shield machine 113 in learning section 110 set in an arbitrary section after shield machine 113 starts excavating, to generate trained differential prediction model 112. Differential prediction model generation module 451 also uses artificial intelligence to machine-learn weighting coefficients for each of the parameters input for excavating shield machine 113 in learning section 110, to generate trained differential prediction model 112.

[0036] First path prediction module 452 is a module that uses basic model 111 to predict the tip position and orientation of shield tunneling machine 113 in construction section 120 after passing learning section 110. Error prediction module 453 is a module that uses difference prediction model 112 to predict errors in the predicted tip position and orientation of shield tunneling machine 113 from the difference between the tip position and orientation of shield tunneling machine 113 predicted by basic model 111 and the target excavation position.

[0037] Second path prediction module 454 is a module that predicts the tip position and orientation of shield machine 113 by adding the error predicted by error prediction module 453 to the tip position and orientation of shield machine 113 predicted by first path prediction module 452, thereby predicting the path of shield machine 113 in construction section 120. These modules 451 to 454 are read by CPU 410 into application execution area 445 of RAM 440 and executed. Control program 455 is a program for controlling path prediction device 100 as a whole.

[0038] The input / output interface 460 interfaces input / output data with input / output devices. A display unit 461 and an operation unit 462 are connected to the input / output interface 460. A storage medium 464 may also be connected to the input / output interface 460. A speaker 463 serving as an audio output unit, a microphone (not shown) serving as an audio input unit, or a GPS position determination unit may also be connected. Note that the RAM 440 and storage 450 shown in FIG. 4 do not include programs or data relating to the general-purpose functions of the course prediction device 100 or other feasible functions.

[0039] Next, the processing procedure of the path prediction device 100 will be described with reference to the flowchart shown in Fig. 5. This flowchart is executed by the CPU 410 in Fig. 4 using the RAM 440, and realizes each functional configuration of the path prediction device 100 in Fig. 2.

[0040] In step S501, differential prediction model generation unit 201 uses basic model 111 to machine-learn the difference between the basic-model predicted position that shield machine 113 will reach by excavating in learning section 110, which is set at an arbitrary section after shield machine 113 starts excavating, and the actual position that shield machine 113 will reach, using artificial intelligence to generate trained differential prediction model 112. In step S503, first path prediction unit 202 uses basic model 111 to predict the tip position and orientation of shield machine 113 in construction section 120 after passing through learning section 110. In step S505, error prediction unit 203 uses differential prediction model 112 to predict an error in the predicted tip position and orientation of shield machine 113 from the difference between the tip position and orientation of shield machine 113 predicted by basic model 111 and the target excavation position. In step S507, the second path prediction unit 204 predicts the tip position and orientation of the shield tunneling machine 113 by adding the error predicted in step S505 to the tip position and orientation of the shield tunneling machine 113 predicted in step S503, and predicts the path of the shield tunneling machine 113 in the construction section 120.

[0041] According to this embodiment, the shield machine is caused to excavate using the basic model in the learning section, so it can excavate with a certain degree of accuracy even in the learning section, and a highly accurate differential prediction model can be generated. This makes it possible to more accurately predict the path of the shield machine in the construction section. Furthermore, because the shield machine is caused to excavate using the basic model in the learning section, the distance of the learning section required to generate the differential prediction model can be shortened, and a highly accurate differential prediction model can be generated in a short amount of time.

[0042] The present invention has been described above with reference to the embodiments, but the present invention is not limited to the above-described embodiments and can be modified as appropriate. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention. Furthermore, systems or devices that combine the individual features included in each embodiment in any way are also included in the scope of the present invention. Furthermore, the calculation method used in the basic model is not limited to the above-described method, and the amount of overexcavation may be included as a parameter, or other calculation methods may be used.

[0043] The present invention may also be applied to a system consisting of multiple devices or to a single device. Furthermore, the present invention may also be applied when an information processing program that realizes the functions of the embodiments is supplied to a system or device and executed by a built-in processor. Therefore, the technical scope of the present invention also includes a program installed on a computer to realize the functions of the present invention, a medium storing the program, a WWW (World Wide Web) server from which the program is downloaded, and a processor that executes the program. In particular, the technical scope of the present invention also includes a non-transitory computer-readable medium storing a program that causes a computer to execute at least the processing steps included in the above-described embodiments.

Claims

1. a differential prediction model generation unit that uses artificial intelligence to machine learn the difference between a basic model predicted arrival position that the shield machine is predicted to reach when it is allowed to excavate, based on a basic model generated to predict the path of the shield machine in a learning section set in any section after the shield machine has started excavating, and the actual arrival position of the shield machine, to generate a learned differential prediction model; a first path prediction unit that predicts the tip position and orientation of the shield machine using the basic model in a construction section after the learning section has been passed; an error prediction unit that uses the difference prediction model to predict an error in the predicted tip position and orientation of the shield machine from the difference between the tip position and orientation of the shield machine predicted by the basic model and the actual excavation position; a second path prediction unit that predicts the path of the shield machine by adding the error predicted by the error prediction unit to the tip position and orientation of the shield machine predicted by the first path prediction unit; A course prediction device comprising:

2. The path prediction device described in claim 1, wherein the differential prediction model generation unit further uses artificial intelligence to machine-learn weighting coefficients for each of the parameters input to excavate the shield tunneling machine in the learning section, thereby generating the learned differential prediction model.

3. a differential prediction model generation step in which, based on a basic model generated to predict the path of the shield machine in a learning section set at any section after the shield machine starts excavating, artificial intelligence is trained to learn the difference between the basic model predicted position that the shield machine is expected to reach when the shield machine is allowed to excavate and the actual position that the shield machine actually reaches, thereby generating a learned differential prediction model; a first path prediction step of predicting the tip position and orientation of the shield machine using the basic model in a construction section after the learning section has been passed; an error prediction step of using the learned difference prediction model to predict an error in the predicted tip position and orientation of the shield tunneling machine from the difference between the tip position and orientation of the shield tunneling machine predicted by the basic model and the actual excavation position; a second path prediction step of predicting the path of the shield machine by adding the error predicted in the error prediction step to the tip position and orientation of the shield machine predicted in the first path prediction step; A career prediction method including:

4. a differential prediction model generation step in which, based on a basic model generated to predict the path of the shield machine in a learning section set at any section after the shield machine starts excavating, artificial intelligence is trained to learn the difference between the basic model predicted position that the shield machine is expected to reach when the shield machine is allowed to excavate and the actual position that the shield machine actually reaches, thereby generating a learned differential prediction model; a first path prediction step of predicting the tip position and orientation of the shield machine using the basic model in a construction section after the learning section has been passed; an error prediction step of using the learned difference prediction model to predict an error in the predicted tip position and orientation of the shield tunneling machine from the difference between the tip position and orientation of the shield tunneling machine predicted by the basic model and the actual excavation position; a second path prediction step of predicting the path of the shield machine by adding the error predicted in the error prediction step to the tip position and orientation of the shield machine predicted in the first path prediction step; A career prediction program that runs on a computer.

Images