JP2023003193A - Construction management system and method - Google Patents

Abstract

To provide a construction management system and a method allowing a selection of a rapid and correct construction method in a tunnel excavation work.SOLUTION: A construction management system (100) in a tunnel excavation work comprises a construction scheduler (101) for generating a construction schedule showing a construction method order from a soil distribution of a rock mass, a machine learning model (102) for selecting an optimal construction method on the basis of data showing a physical property or a state of a pit face obtained as a construction result and a controller (103) for updating the construction schedule on the basis of the selected optimal construction method and displaying it at a display part (104).SELECTED DRAWING: Figure 1

Description

The present invention relates to a construction management system and method for excavating tunnels for roads, railways, waterways, and the like.

When excavating a tunnel, a construction method suitable for the condition of the ground will be used. For example, a blasting method using a tunnel drill jumbo is used for hard ground, and a mechanical excavation method using a free cross section excavator is used for soft ground (Patent Documents 1 and 2). In the construction of mountain tunnels, the natural ground is excavated by mechanical excavation or blasting, concrete is sprayed onto the tunnel wall, and rock bolts are driven in from above the concrete. In some cases, the exposed ground surface is closed with shoring (Patent Document 3).

On the other hand, in tunnel construction, it is important to perform an accurate rock mass evaluation in advance so as not to cause large rework at the time of construction, and various rock mass evaluation methods have been proposed. For example, a method of evaluating the ground in the range where support is to be constructed by conducting drilling exploration in front of the excavation direction of the tunnel face (Patent Document 4), and a geological survey that predicts the geological situation in front of the tunnel face using noise vibrations during excavation as the epicenter. A method (Patent Document 5) and the like have been proposed.

JP 2020-094422 A JP-A-10-115187 JP 2019-167678 A JP 2011-052373 A JP 2014-106075 A

However, it is difficult to accurately grasp the condition of the rock ground in preliminary surveys of tunnels, and when the tunnel actually begins to be excavated, it often differs from the soil distribution obtained in advance. In that case, it is necessary to switch to a construction method suitable for the current ground conditions. In the past, this decision to switch construction methods was based on human experience, which made it difficult to make a quick decision. .

The present invention has been devised in view of the above circumstances, and an object of the present invention is to provide a construction management system and method that enable prompt and accurate selection of construction methods in tunnel excavation construction.

According to one embodiment of the present invention to achieve the above object, there is provided a construction management system for tunnel excavation, comprising a construction scheduler for generating a construction schedule indicating the order of construction methods from the soil distribution of natural ground; A machine learning model constructed to select an optimum construction method based on the obtained data indicating physical characteristics or conditions of the face, and updating the construction schedule based on the selected optimum construction method and displaying it on the display unit. and a control unit for
According to one embodiment of the present invention, it is possible to further include a construction result accumulation unit that stores the construction result, and to use the construction result for machine learning of the machine learning model.
According to one embodiment of the present invention, the data indicative of the physical properties or conditions of the face may include at least one of unconfined compressive strength, water content, and seepage rate.
According to one embodiment of the present invention, there is provided a construction management method for tunnel excavation work, in which a construction scheduler generates a construction schedule indicating the order of construction methods from the soil distribution of natural ground, and a machine learning model is obtained as construction results. The optimum construction method is selected based on the data indicating the physical characteristics or condition of the face, and the control unit updates the construction schedule based on the selected optimum construction method and displays it on the display unit.
According to one embodiment of the present invention, the construction result can be accumulated and the construction result can be used for machine learning of the machine learning model.

According to one embodiment of the present invention, the construction schedule is updated and displayed by selecting the optimum construction method based on the data indicating the physical characteristics or state of the face obtained as construction results, thereby enabling quick and accurate construction. It is advantageous for construction.
Moreover, according to one embodiment of the present invention, it is advantageous to build a highly accurate machine learning model by accumulating construction results and using them for machine learning.
Further, according to one embodiment of the present invention, using at least the uniaxial compressive strength as data for selecting the optimum construction method is advantageous in quickly and accurately switching between the blasting method and the mechanical excavation method.

It is a block diagram showing composition of a construction management system by one embodiment of the present invention. It is a flowchart which illustrates the construction method selection method in this embodiment. FIG. 4 is a schematic block diagram illustrating a learning process of artificial intelligence (AI) in the construction management system in this embodiment;

1. System configuration As illustrated in FIG. 1, a construction management system 100 according to one embodiment of the present invention has a construction scheduler 101 and an AI analysis unit 102, a control unit 103 for controlling the entire system, and a selected It has a display unit 104 that displays a construction method and a construction result accumulation unit 105 that accumulates construction results. The construction management system 100 can be constructed on a general-purpose server or a personal computer having processors such as a CPU (Central Processing Unit) and an MPU (Micro Processor Unit).

The construction scheduler 101 inputs the soil distribution data D0 acquired by the preliminary investigation of the natural ground for excavating the tunnel, and generates a construction schedule according to the soil distribution.

The construction schedule is a series of construction methods selected according to the soil distribution D0 from blasting, mechanical excavation, NATM, and shoring. For example, if the soil distribution D0 indicates a hard ground, the blasting method is selected, and if the location indicates a soft ground, the mechanical excavation method is selected.

The AI analysis unit 102 inputs various data indicating the ground condition acquired during construction, and selects the optimum construction method using, for example, a machine learning model (learned optimum construction method selection model) constructed by deep learning.

The various data provided to the AI analysis unit 102 may be data that typically indicates the physical characteristics or state of the face as a result of construction, and includes, for example, at least the uniaxial compressive strength D1 of the face ground, and the water content ratio D2 and Data such as the amount of spring water D3 from the face can be used. Deep learning technology can build a machine learning model with sufficiently small error by providing appropriate training data and adjusting neuron parameters even when a large amount of data influences each other to determine the optimal construction method. By using such a machine learning model, it becomes possible to efficiently and quickly select the optimum construction method from various data during construction.

If the optimum construction method selected by the AI analysis unit 102 is different from the construction method of the construction schedule, the control unit 103 changes the construction schedule and causes the display unit 104 to display the changed construction schedule. The worker can prepare and perform the next construction method according to the construction schedule displayed on the display unit 104 .

The construction result accumulation unit 105 stores the results (various data) when executing the scheduled or selected construction method, and can be used as learning data for the optimal construction method selection model of the AI analysis unit 102.

Each function of the construction scheduler 101, the AI analysis unit 102, and the control unit 103 described above can be realized by executing a program stored in a memory (not shown) on a processor.

2. Operation As illustrated in FIG. 2, the construction management system 100 according to the present embodiment first inputs soil distribution data D0 acquired by a preliminary survey (operation 201). The soil distribution data D0 can be acquired using a well-known exploration method such as boring exploration, elastic wave exploration, or the like.

The construction scheduler 101 holds, for example, a table of optimal construction methods for each predetermined soil quality, and generates a construction schedule that defines construction methods according to the soil distribution of the input soil distribution data D0 (operation 202). For example, if a preliminary survey reveals that hard soil continues behind soft soil, a construction schedule is generated that first uses the mechanical excavation method using a free cross-section excavator and then the blasting method. Construction is started according to the construction schedule thus generated (operation 203).

The AI analysis unit 102 inputs various data as construction results (operation 204), and selects the optimum construction method using the optimum construction method selection model (operation 205). After inputting the selected optimum construction method, the control unit 103 determines whether or not to switch the construction method of the construction schedule (operation 206). In a simplified example, if the uniaxial compressive strength D1 falls below a predetermined value after starting construction by the blasting method, the blasting method scheduled as the next method is switched to the mechanical excavation method according to the optimum method selection model. . Specifically, when the unconfined compressive strength of the ground becomes 49 N/mm ² or 500 kgf/cm ² or less, the method is switched to the mechanical excavation method. Alternatively, when excavation is performed by the NATM construction method and the natural ground deteriorates, the construction method is switched to a construction method that uses shoring. Also, if there is spring water from the face or crown, switch to a construction method that involves chemical injection. In this way, the optimum construction method corresponding to various data is selected according to the constructed learned optimum construction method selection model.

If there is a construction method switching determination (YES in operation 206), the control unit 103 stores various data indicating the construction result and the switching determination in the construction result accumulation unit 105 (operation 207), changes the construction schedule, and displays the display unit 104. to display a construction method switching instruction (operation 208). Thus, construction is started according to the new construction schedule (operation 203).

If the construction method according to the current construction schedule is acceptable, that is, if there is no construction method switching determination (NO in operation 206), the control unit 103 stores various data indicating construction results and the switching unnecessary determination in the construction result accumulation unit 105 ( Operation 209), if construction is not finished (NO in operation 210), the next construction is started according to the current construction schedule (operation 203). The above operations 203 to 210 are repeated until the construction is completed (YES in operation 210), and the construction result is accumulated in the construction result accumulation unit 105 while determining whether the construction method is switched or not to be changed in the construction schedule. In addition, it is also possible to explore the nature of the natural ground using the vibration during the blasting method or the period excavation method, and to reflect it in the construction schedule.

As illustrated in FIG. 3, the optimal construction method selection model for the AI analysis unit 102 can be constructed using past data. In other words, it is possible to accumulate data representative of the physical characteristics or conditions of the working face obtained as a result of construction and the presence or absence of construction method switching, and to construct an optimal construction method selection model using the accumulated data.

100 construction management system 101 construction scheduler 102 AI analysis unit (optimal construction method selection model)
103 control unit 104 display unit 105 construction result accumulation unit

Claims (5)

A construction management system in tunnel excavation work,
a construction scheduler that generates a construction schedule indicating the order of construction methods from the soil distribution of natural ground;
A machine learning model constructed to select the optimum construction method based on data indicating the physical characteristics or condition of the face obtained as a result of construction;
a control unit that updates the construction schedule based on the selected optimum construction method and displays it on a display unit;
A construction management system comprising: 2. The construction management system according to claim 1, further comprising a construction result accumulation unit for storing said construction result, wherein said construction result is used for machine learning of said machine learning model. 3. The construction management system according to claim 1, wherein the data indicating the physical properties or conditions of said face include at least uniaxial compressive strength. A construction management method in tunnel excavation work,
The construction scheduler generates a construction schedule indicating the order of construction methods from the soil distribution of the natural ground,
A machine learning model selects the optimal construction method based on data indicating the physical characteristics or condition of the face acquired as construction results,
The control unit updates the construction schedule based on the selected optimum construction method and displays it on the display unit;
A construction management method characterized by: 5. The construction management method according to claim 4, wherein said construction results are accumulated and said construction results are used for machine learning of said machine learning model.

Images