Excavation shape prediction device, excavation shape prediction method, and program

Abstract

To provide an excavation shape predicting device, an excavation shape predicting method and a program for improving construction accuracy in construction using a caisson.SOLUTION: An excavation shape predicting device used in construction using a caisson comprises: an excavation shape information acquiring part for acquiring excavation shape information showing the shape of ground after excavation on the basis of sensor information acquired by a sensor device provided to a skeleton of the caisson; an immersion state information acquiring part acquiring the immersion state information showing the information regarding the immersion state of the caisson after immersion by excavation of the ground on the basis of sensor information acquired by the sensor device; and a prediction part predicting the excavation shape of the ground in next excavation from the acquired excavation shape information and the immersion state information, by using a prediction model that learns a relation of the ground excavation with the immersion of the caisson.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] In the past, in construction using caissons, it was necessary to properly manage the state of the caissons during construction in order to achieve more accurate construction. For example, one construction method using caissons is the pneumatic caisson method, in which the caissons are sunk while being supported by excavated soil.

[0003] In the pneumatic caisson construction method, for example, a site supervisor predicts the type of excavation and the type of subsidence that will occur based on measurement data collected during excavation using a remote excavator installed on the underside of the caisson. Specifically, the site supervisor makes predictions based on measurement data, boring surveys, and other data, taking into account the soil quality at the excavation site and the actual excavation condition confirmed visually, and then decides how to excavate. If subsidence occurs due to excavation, the supervisor re-evaluates and predicts the subsidence condition, and then decides how to excavate next, following this cycle of work. This cycle ensures that the caisson's body remains properly submerged. For example, Patent Document 1 below discloses a technology for calculating the shape of excavated soil based on the distance from the slope of the excavated soil to a reference point measured using radar. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2015-229826 Summary of the Invention [Problem to be solved by the invention]

[0005] However, because there are many factors that affect the sinking state of the caisson body, even if the supervisor is an experienced and knowledgeable person, it is difficult to accurately predict what type of excavation will result in what type of settlement. As a result, there is a risk of problems such as the body tilting due to human error.

[0006] In view of the above-mentioned problems, an object of the present invention is to provide an excavation shape prediction device, an excavation shape prediction method, and a program that can improve the accuracy of construction using caissons. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems, an excavation shape prediction device according to one aspect of the present invention comprises an excavation shape information acquisition unit that acquires excavation shape information indicating the shape of the ground after excavation based on sensor information acquired by a sensor device provided on a body of a caisson in construction using a caisson, a sinking state information acquisition unit that acquires sinking state information indicating information about the sinking state of the caisson after it has subsided due to excavation of the ground based on the sensor information acquired by the sensor device, and a prediction unit that predicts the excavation shape of the ground in the next excavation from the acquired excavation shape information and sinking state information using a prediction model that has learned the relationship between the excavation of the ground and the subsidence of the caisson. The prediction model predicts and outputs an excavation shape that can improve the abnormality if there is an abnormality in the submerged state after excavation based on the input excavation shape information and the submerged state information, and predicts and outputs an excavation shape that can maintain the submerged state if there is no abnormality in the submerged state after excavation. .

[0008] An excavation shape prediction method according to one aspect of the present invention includes an excavation shape information acquisition process in which an excavation shape information acquisition unit acquires excavation shape information indicating the shape of the ground after excavation based on sensor information acquired by a sensor device installed on a body of a caisson during construction using a caisson; a sinking state information acquisition process in which a sinking state information acquisition unit acquires sinking state information indicating information about the sinking state of the caisson after it has subsided due to excavation of the ground based on the sensor information acquired by the sensor device; and a prediction process in which a prediction unit predicts the excavation shape of the ground in the next excavation from the acquired excavation shape information and sinking state information using a prediction model that has learned the relationship between the excavation of the ground and the subsidence of the caisson. The prediction model predicts and outputs an excavation shape that can improve the abnormality if there is an abnormality in the submerged state after excavation, based on the input excavation shape information and the submerged state information, and predicts and outputs an excavation shape that can maintain the submerged state if there is no abnormality in the submerged state after excavation. .

[0009] A program according to one aspect of the present invention causes a computer to function as: excavation shape information acquisition means for acquiring excavation shape information indicating the shape of the ground after excavation based on sensor information acquired by a sensor device provided on the body of a caisson in construction using a caisson; sinking state information acquisition means for acquiring sinking state information indicating information regarding the sinking state of the caisson after it has subsided due to excavation of the ground based on the sensor information acquired by the sensor device; and prediction means for predicting the excavation shape of the ground in the next excavation from the acquired excavation shape information and sinking state information using a prediction model that has learned the relationship between the excavation of the ground and the subsidence of the caisson. The prediction model predicts and outputs an excavation shape that can improve the abnormality if there is an abnormality in the submerged state after excavation based on the input excavation shape information and the submerged state information, and predicts and outputs an excavation shape that can maintain the submerged state if there is no abnormality in the submerged state after excavation. . [Effects of the Invention]

[0010] According to the present invention, the accuracy of construction using caissons can be improved. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a block diagram showing an example of the configuration of an excavation shape prediction system according to a first embodiment. [Figure 2] FIG. 2 is a diagram illustrating an example of equipment layout according to the first embodiment. [Figure 3] 5 is a flowchart showing an example of a processing flow in the excavation shape prediction device according to the first embodiment. [Figure 4] FIG. 10 is a block diagram showing an example of the configuration of an excavation shape prediction system according to a second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0013] <<1. First Embodiment>> <1-1. Configuration of excavation shape prediction system> First, the configuration of the excavation shape prediction system according to the first embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the excavation shape prediction system according to the first embodiment. As shown in Fig. 1, the excavation shape prediction system 1 includes an excavation shape prediction device 10, a friction sensor 11, an air pressure sensor 12, an inclination sensor 13, a cutting edge reaction force sensor 14, and a range sensor 15. The friction sensor 11, the air pressure sensor 12, the inclination sensor 13, the cutting edge reaction force sensor 14, and the range sensor 15 are examples of sensor devices provided in the body of the caisson.

[0014] The excavation shape prediction device 10 is a device that predicts the excavation shape of ground. In the first embodiment, the excavation shape prediction device 10 predicts the excavation shape of the remaining soil near the cutting edge of a caisson in the ground to be excavated. Here, the remaining soil refers to the remaining soil near the cutting edge for the purpose of reducing the ground reaction force acting on the cutting edge and limiting the settlement of the caisson. When the remaining soil reaches the underside of the base slab, the underside of the base slab receives the ground reaction force, thereby preventing the settlement of the caisson. In particular, in soft ground, increasing the remaining soil size reduces the ground reaction force received by the cutting edge, thereby preventing the settlement of the caisson. Note that the excavation shape predicted by the excavation shape prediction device 10 may be the shape of the portion to be excavated from the remaining soil, or it may be the shape of the portion of the remaining soil (the remaining soil shape) after excavation.

[0015] The excavation shape prediction device 10 is realized by, for example, a PC (personal computer), a server device, etc. The excavation shape prediction device 10 is communicably connected to each of the sensor devices, namely, a friction sensor 11, an air pressure sensor 12, an inclination sensor 13, a cutting edge reaction force sensor 14, and a range sensor 15. The excavation shape prediction device 10 receives information acquired by each of the sensor devices (hereinafter also referred to as "sensor information") from each of the sensor devices.

[0016] The friction sensor 11 measures the friction force of the ground acting on the circumferential surface of the caisson when the caisson sinks. The friction sensor 11 acquires friction information indicating the measured friction force acting on the circumferential surface of the caisson as sensor information and transmits it to the excavation shape prediction device 10. The air pressure sensor 12 measures the air pressure inside the work chamber. The work chamber is a space provided below the body of the caisson. The air pressure sensor 12 acquires air pressure information indicating the measured air pressure inside the work chamber as sensor information and transmits it to the excavation shape prediction device 10. The inclination sensor 13 measures the inclination of the entire caisson. The inclination can be measured, for example, by measuring the elevation difference of the caisson using a height difference meter installed at a predetermined position on the caisson. The inclination sensor 13 acquires inclination information indicating the measured inclination of the entire caisson as sensor information and transmits it to the excavation shape prediction device 10. A plurality of cutting edge reaction force sensors 14 are installed along the outer periphery of the caisson on the underside of the cutting edge. The cutting edge is the part of the caisson where ground reaction forces are concentrated. The cutting edge reaction force sensors 14 measure the ground reaction forces acting on the cutting edge. The cutting edge reaction force sensors 14 acquire cutting edge reaction force information indicating the measured ground reaction forces acting on the cutting edge as sensor information and transmit it to the excavation shape prediction device 10. The range sensor 15 measures the shape of the object to be measured. The range sensor 15 is realized by, for example, a 3D (three-dimensional) scanner. In this embodiment, the range sensor 15 acquires excavation shape information (for example, a three-dimensional point cloud) indicating the shape of the object to be measured after excavation as sensor information, and transmits it to the excavation shape prediction device 10. Hereinafter, in the first embodiment, the object to be measured is, for example, the ground, more specifically, residual soil, for example.

[0017] <1-2. Equipment placement> Here, the arrangement of equipment will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the arrangement of equipment according to the first embodiment. The caisson body 20 according to the first embodiment shown in Fig. 2 is a reinforced concrete structure used when constructing a structure such as a bridge pier constructed underwater using the pneumatic caisson method. As shown in Fig. 2, the caisson body 20 is submerged by placing its lower end in the ground G and sinking due to its own weight or, in some cases, by pouring water into the body and causing it to sink due to the weight of the body and the water.

[0018] A friction sensor 11 is provided on a side wall 21 of the caisson body 20. This allows the friction sensor 11 to measure the friction force of the ground acting on the circumferential surface of the caisson when the caisson sinks. A work chamber 30 is provided below the bottom slab 22 of the caisson body 20 as a space for carrying out excavation work. The bottom of the bottom slab 22 is the ceiling of the work chamber 30. A barometric pressure sensor 12 is provided below the bottom slab 22. This allows the barometric pressure sensor 12 to measure the barometric pressure inside the work chamber 30. An inclination sensor 13 is provided on the top of the bottom slab 22 of the caisson body 20. This allows the inclination sensor 13 to measure the inclination of the entire caisson body 20. A cutting edge reaction force sensor 14 is provided on the bottom surface of the cutting edge 23. This allows the cutting edge reaction force sensor 14 to measure the ground reaction force acting on the cutting edge 23. Rails 40 are provided below the bottom slab 22 of the caisson body 20. An excavator 41 and a range sensor 15 are provided on the rails 40 so that they can travel on the rails 40. As a result, the excavator 41 can excavate earth and sand within the work chamber 30 by traveling along the rails 40. When the excavator 41 excavates earth and sand, residual soil 50 is formed. Note that the number of excavators 41 is not limited to one, and there may be more than one. Moreover, the range sensor 15 can measure the shape of the remaining soil 50 over a wide area of ​​the work chamber 30 by traveling on the rails 40.

[0019] <1-3. Functional configuration of excavation shape prediction device> 1, a description will be given of the functional configuration of the excavation shape prediction device 10. As shown in FIG. 1, the excavation shape prediction device 10 includes an input unit 110, a communication unit 120, a storage unit 130, a control unit 140, and an output unit 150.

[0020] The input unit 110 has a function of accepting input from a user. The input unit 110 is realized by an input device such as a keyboard, a mouse, a touch panel, etc. The input device may be a device that is provided in advance as hardware in the excavation shape prediction device 10, or may be a device that is externally connected to the excavation shape prediction device 10.

[0021] For example, the input unit 110 accepts input of basic information, ground information, feedback information, etc. of the caisson from the user. The basic information of the caisson is, for example, the caisson's body weight (kN), the caisson's outfit weight (kN), the load water (kN), and the outer periphery area. The ground information is, for example, information indicating the characteristics and properties of the ground to be excavated in construction using the caisson. Specifically, the ground information is information indicating the soil type (geology) such as sand, gravel, clay, and mudstone, and information indicating the ground strength such as the N value and uniaxial compressive strength. The feedback information is, for example, information indicating an evaluation of the construction performed based on the excavation shape predicted (output) by the excavation shape prediction device 10. The input unit 110 stores the basic information, ground information, feedback information, etc. of the caisson that has been accepted as input in the storage unit 130.

[0022] The communication unit 120 has a function of transmitting and receiving various types of information. For example, the communication unit 120 receives sensor information from various sensor devices.

[0023] The storage unit 130 has a function of storing various types of information. The storage unit 130 is configured by a storage medium, such as a hard disk drive (HDD), a solid state drive (SSD), a flash memory, an electrically erasable programmable read-only memory (EEPROM), a random access read / write memory (RAM), a read-only memory (ROM), or any combination of these storage media.

[0024] For example, the storage unit 130 stores basic information and feedback information about the caisson input by the user. The storage unit 130 also stores a prediction model generated by machine learning. The prediction model is a model that learns the relationship between ground excavation and caisson subsidence in construction using a caisson. For example, the prediction model is a prediction model generated by machine learning using excavation shape information that indicates the shape of the ground and submerged state information that indicates information about the submerged state of the caisson as training data. Specifically, the prediction model uses information on the excavation shape before and after excavation and information on the sinking state before and after excavation as training data, and from the differences between each piece of information, it can learn how to excavate which ground (remaining excavated soil) and how to sink the caisson. For example, if the difference in the excavation shape information indicates that the remaining soil on the right side has been excavated, and the difference in the settling state information indicates that the caisson, which was leaning to the left, has now straightened out, the prediction model will learn that in order to straighten the caisson, which is leaning to the left, it is necessary to excavate the remaining soil on the right side into a specified shape. For example, suppose the difference in the excavation shape information indicates that the remaining soil on the left side has been excavated, and the difference in the sinking state information indicates that the caisson, which had shifted to the right from its designed position, has returned to its designed position. In this case, the predictive model learns that in order to return the caisson, which had shifted to the right from its designed position, to its designed position, it is sufficient to excavate the remaining soil on the left side to a specified shape. In this case, the caisson may be intentionally tilted and lowered to return it to its designed position, and then excavated again to eliminate the tilt. As a result, the prediction model receives input of excavation shape information indicating the shape of the remaining soil after excavation and sinking state information indicating the sinking state of the caisson after it has sunk due to excavation, and if there is an abnormality in the sinking state after excavation (for example, tilt or misalignment of the caisson), it predicts and outputs an excavation shape that can correct the abnormality.On the other hand, if there is no abnormality in the sinking state after excavation, the prediction model predicts and outputs an excavation shape that can maintain the sinking state.In other words, the prediction model can predict how the remaining soil should be excavated next, depending on the shape of the remaining soil after excavation and the sinking state of the caisson. The prediction model according to the first embodiment is assumed to be machine-learned based on information acquired at one construction site.

[0025] Note that the training data may include other information as long as it includes at least the installation state information and the excavation shape information. This allows the prediction model to predict the excavation shape of the ground with higher accuracy than when machine learning is performed using training data including only the installation state information and the excavation shape information. For example, the training data may include ground information. In other words, the prediction model may be a model that learns the relationship between the excavation of the ground and the subsidence of the caisson in construction using a caisson, as well as the relationship with the ground. Even in the same soil type, physical properties such as strength are not uniform and usually vary within a certain range. Furthermore, in most cases, soil quality changes as excavation progresses. Such differences in soil quality affect caisson settlement. For example, the rate at which a caisson settles differs between sandy and clayey soils. Specifically, in sandy soils, the caisson often settles immediately after the remaining soil is excavated, with little or no further settlement thereafter. In contrast, in clayey soils, the caisson often settles immediately after the remaining soil is excavated, with gradual settlement thereafter. This can cause the caisson to tilt or shift from its designed position depending on the soil quality. Therefore, having the prediction model learn soil information is effective in improving the accuracy of excavation shape predictions. When the prediction model also learns ground information, it predicts and outputs the excavation shape taking into account the ground information when it receives input of excavation shape information showing the shape of the remaining soil after excavation, sinking state information showing the sinking state of the caisson after it sinks due to excavation, and ground information showing information about the excavated ground. In the first embodiment, an example will be described below in which the prediction model is a prediction model that is machine-learned using training data including excavation shape information, submerged installation state information, and ground information.

[0026] The control unit 140 has a function of controlling the overall operation of the excavation shape prediction device 10. The control unit 140 is realized, for example, by causing a CPU (Central Processing Unit) provided as hardware in the excavation shape prediction device 10 to execute a program. As shown in FIG. 1, the control unit 140 includes a sensor information acquisition unit 141, a sinking state information acquisition unit 142, a ground information acquisition unit 143, an excavation shape information acquisition unit 144, a feedback information acquisition unit 145, a learning unit 146, and a prediction unit 147.

[0027] The sensor information acquisition unit 141 has a function of acquiring sensor information. For example, the sensor information acquisition unit 141 acquires each piece of sensor information transmitted from the friction sensor 11, the air pressure sensor 12, the tilt sensor 13, the cutting edge reaction force sensor 14, and the range sensor 15 and received by the communication unit 120.

[0028] The sinking state information acquisition unit 142 has a function of acquiring sinking state information. For example, the sinking state information acquisition unit 142 acquires sinking state information regarding the sinking state of the caisson after it has sunk due to excavation, based on sensor information acquired by each sensor device provided on the caisson body 20. Specifically, the installation state information acquisition unit 142 acquires the subsidence balance, caisson inclination, and cutting edge reaction force as installation state information. The subsidence balance is information that can be calculated based on the sensor information acquired by the sensor information acquisition unit 141 and the basic information of the caisson stored in the memory unit 130. The caisson inclination is information indicated by the inclination information acquired by the sensor information acquisition unit 141. The cutting edge reaction force is information indicated by the cutting edge reaction force information acquired by the sensor information acquisition unit 141.

[0029] To obtain the subsidence balance, the installation state information obtaining unit 142 first calculates the subsidence force using the following formula. Subsidence force = caisson body weight + caisson equipment weight + loaded water The caisson body weight, caisson equipment weight, and load water shown in the above formula are information included in the basic information of the caisson stored in the memory unit 130. Next, the installation state information acquisition unit 142 calculates the subsidence resistance force using the following formula. Subsidence resistance = (average friction force x outer area) + (air pressure inside the box x excavation area) The average friction force shown in the above formula is the average value of the friction forces measured by the multiple friction sensors 11, and the outer circumferential area is information included in the basic information of the caisson stored in the memory unit 130. The internal caisson air pressure shown in the above formula is the air pressure inside the working chamber 30 measured by the air pressure sensor 12. The excavation area is an area calculated from the excavation shape information measured by the range sensor 15, for example. Then, the installation state information acquisition unit 142 calculates the subsidence balance using the following formula. Subsidence balance = Subsidence force - Subsidence resistance When no settlement occurs, a ground reaction force equivalent to the above-mentioned settlement balance acts on the underside of the caisson. When the work chamber 30 is excavated and the settlement balance exceeds the available ground reaction force, settlement occurs. Even if the work chamber is excavated evenly, depending on the condition of the ground, such as different soil properties in some areas, the distribution of ground reaction force may not be uniform, and tilting may occur during settlement. Therefore, it is important to understand the relationship between ground information, excavation shape information, and settlement state information.

[0030] The ground information acquisition unit 143 has a function of acquiring ground information. For example, the ground information acquisition unit 143 acquires the ground information from the storage unit 130.

[0031] The excavation shape information acquisition unit 144 has a function of acquiring excavation shape information. For example, the excavation shape information acquisition unit 144 acquires excavation shape information indicating the shape of the remaining soil after excavation, based on the sensor information acquired by the sensor information acquisition unit 141. As an example, the excavation shape information acquisition unit 144 acquires the excavation shape information from the sensor information.

[0032] The feedback information acquisition unit 145 has a function of acquiring feedback information. For example, the feedback information acquisition unit 145 acquires the feedback information from the storage unit 130.

[0033] The learning unit 146 has a function of generating a prediction model by performing supervised learning on a learning model using training data, which is a data set for learning. For example, the learning unit 146 is a prediction model that has been machine-learned using prepared excavation shape information and submerged state information as training data. Specifically, the learning unit 146 uses a dataset in which the difference in submerged state information before and after excavation is an explanatory variable and the difference in excavation shape information before and after excavation is an objective variable, and causes the learning model to machine-learn the correspondence between the excavation of the ground and the subsidence of the caisson. In this way, the learning unit 146 can generate a prediction model that can predict and output the excavation shape of the ground in the next excavation using the submerged state information and the excavation shape information as input data.

[0034] Furthermore, the learning unit 146 uses a dataset in which the difference in the submerged state information before and after excavation and the difference in the ground information before and after excavation are used as explanatory variables, and the difference in the excavation shape information before and after excavation is used as a target variable, and machine-learns the correspondence between the excavation of the ground and the subsidence of the caisson, as well as the correspondence with the ground information, into the learning model. As a result, the learning unit 146 can use the submerged state information, excavation shape information, and ground information as input data to generate a prediction model that can predict and output the excavation shape of the ground in the next excavation, taking the ground information into consideration. The learning model may be, for example, a deep learning model using a convolutional neural network (CNN), a deep CNN (DCNN), a decision tree, a hierarchical Bayes model, or a support vector machine (SVM).

[0035] The learning unit 146 also has a function of re-learning the generated prediction model. For example, the learning unit 146 re-learns the prediction model using the submerged installation state information, excavation shape information, and ground information acquired after the generation of the prediction model. The learning unit 146 also re-learns the prediction model based on the feedback information acquired by the feedback information acquisition unit 145. This allows the learning unit 146 to improve the prediction accuracy of the excavation shape of the remaining soil in the next excavation.

[0036] The prediction unit 147 has a function of predicting the excavation shape of the ground. For example, the prediction unit 147 uses a prediction model stored in the storage unit 130 to predict the excavation shape of the remaining soil in the next excavation.

[0037] When the prediction model is a model that has been machine-learned using the submerged state information and excavation shape information as training data, the prediction unit 147 inputs the submerged state information acquired by the submerged state information acquisition unit 142 and the excavation shape information acquired by the excavation shape information acquisition unit 144 into the prediction model, and simulates and outputs the excavation shape of the ground for the next excavation predicted from the submerged state information after excavation and the excavation shape information.

[0038] When the prediction model is a model that has been machine-learned using the submerged state information, excavation shape information, and ground information as training data, the prediction unit 147 inputs the submerged state information acquired by the submerged state information acquisition unit 142, the excavation shape information acquired by the excavation shape information acquisition unit 144, and the ground information acquired by the ground information acquisition unit 143 into the prediction model, and simulates and outputs the excavation shape of the ground for the next excavation predicted from the submerged state information, excavation shape information, and ground information after excavation.

[0039] The output unit 150 has a function of outputting various information. The output unit 150 is realized by, for example, a display device such as a display. The display device may be a device that is provided in advance as hardware in the excavation shape prediction device 10, or may be a device that is externally connected to the excavation shape prediction device 10. For example, the output unit 150 displays on a display device the excavation shape predicted by the prediction unit 147. This allows the user to check the predicted excavation shape.

[0040] The excavation shape prediction device 10 acquires sensor information from all sensor devices connected to the device itself and predicts the excavation shape using all the acquired sensor information. Therefore, the excavation shape prediction device 10 can output a predicted excavation shape by comprehensively judging the sinking state of the caisson. By carrying out construction based on the excavation shape, the user can further improve the accuracy of construction.

[0041] <1-4. Processing flow> The flow of processing in the excavation shape prediction device 10 according to the first embodiment will be described with reference to Fig. 3. Fig. 3 is a flowchart showing an example of the flow of processing in the excavation shape prediction device according to the first embodiment. In the example shown in Fig. 3, it is assumed that a prediction model generated in advance by the learning unit 146 is stored in the storage unit 130.

[0042] 3, first, the sensor information acquisition unit 141 acquires sensor information (step S101). Specifically, the sensor information acquisition unit 141 acquires the sensor information received by the communication unit 120 from the friction sensor 11, the air pressure sensor 12, the tilt sensor 13, the cutting edge reaction force sensor 14, and the range measurement sensor 15. Next, the installation state information acquisition unit 142 acquires the installation state information (step S102). Specifically, the installation state information acquisition unit 142 acquires the subsidence balance, the caisson inclination, and the cutting edge reaction force as installation state information based on the sensor information acquired by the sensor information acquisition unit 141 and the basic information of the caisson stored in the storage unit 130. Next, the excavation shape information acquisition unit 144 acquires the excavation shape information (step S103). Specifically, the excavation shape information acquisition unit 144 acquires the excavation shape information from the sensor information acquired by the sensor information acquisition unit 141. Next, the ground information acquisition unit 143 acquires the ground information (step S104). Specifically, the ground information acquisition unit 143 acquires the ground information stored in the storage unit . Next, the prediction unit 147 predicts the excavation shape (step S105). Specifically, the prediction unit 147 inputs the submerged state information acquired by the submerged state information acquisition unit 142, the excavation shape information acquired by the excavation shape information acquisition unit 144, and the ground information acquired by the ground information acquisition unit 143 into the prediction model stored in the storage unit 130. After the input, the prediction unit 147 predicts the excavation shape output from the prediction model as the excavation shape for the next excavation. Next, the output unit 150 outputs the excavation shape predicted by the prediction unit 147 (step S106). Next, the feedback information acquisition unit 145 attempts to acquire feedback information (step S107). If the feedback information is acquired (step S107 / YES), the process proceeds to step S108. On the other hand, if the feedback information is not acquired (step S107 / NO), the process ends. When the process proceeds to step S108, the learning unit 146 causes the prediction model to re-learn (step S108). Specifically, the learning unit 146 causes the prediction model to re-learn using the feedback information acquired by the feedback information acquisition unit 145, and then ends the process.

[0043] If feedback information cannot be acquired in step S107, step S107 may be repeated until feedback information can be acquired without terminating the process, or the process may be repeated from step S101. Furthermore, after re-learning is performed in step S108, the process may be repeated from step S101 without terminating the process.

[0044] 2. Second embodiment The first embodiment has been described above. Next, the second embodiment will be described. In the above-described first embodiment, an example in which a prediction model performs machine learning based on information acquired at one construction site has been described, but the present invention is not limited to such an example. In the second embodiment, an example in which a prediction model performs machine learning based on information acquired at multiple construction sites will be described. Note that, hereinafter, explanations that overlap with the explanation in the first embodiment will be omitted as appropriate.

[0045] <2-1. Configuration of excavation shape prediction system> The configuration of the excavation shape prediction system according to the second embodiment will be described with reference to Fig. 4. Fig. 4 is a block diagram showing an example of the configuration of the excavation shape prediction system 1a according to the second embodiment. As shown in Fig. 4, the configuration of the excavation shape prediction system 1a according to the second embodiment is similar to the configuration of the excavation shape prediction system 1 according to the first embodiment described with reference to Fig. 1, and therefore, redundant description will be omitted.

[0046] <2-2. Equipment placement> The arrangement of the devices according to the second embodiment is similar to the arrangement of the devices according to the first embodiment described with reference to FIG. 2, and therefore a duplicated description will be omitted.

[0047] <2-3. Functional configuration of excavation shape prediction device> As shown in Figure 4, the configuration of the excavation shape prediction system 1a of the second embodiment differs from the configuration of the excavation shape prediction system 1 of the first embodiment in that the control unit 140a of the excavation shape prediction device 10a is equipped with an other site information acquisition unit 148.

[0048] The other site information acquisition unit 148 has a function of acquiring other site information including installed state information and excavation shape information at other sites. When the other site information acquisition unit 148 acquires other site information, the learning unit 146 causes the prediction model to perform machine learning using the other site information as well. That is, the prediction model according to the second embodiment is a model that has undergone machine learning using training data that includes not only installed state information and excavation shape information acquired at the site where construction is currently being carried out, but also installed state information and excavation shape information acquired at other sites. In this way, by having the prediction model perform machine learning using other site information as well, the prediction accuracy of the excavation shape is improved, and the prediction model can be used universally at a variety of sites.

[0049] When machine learning is performed using information from other sites, as in the prediction model according to the second embodiment, the feedback information acquisition unit 145 may also acquire, as feedback information, an evaluation of the construction work performed based on the excavation shape output from the other sites. When the feedback information acquisition unit 145 acquires feedback information from the other sites, the learning unit 146 re-trains the prediction model based on the feedback information including the evaluation from the other sites. In this way, the learning unit 146 not only improves the prediction accuracy of the excavation shape at one site using the prediction model, but also improves the prediction accuracy of the excavation shape when the prediction model is used for general purposes at a variety of sites.

[0050] <2-4. Processing flow> The processing flow in the excavation shape prediction device 10a according to the second embodiment is the same as the processing flow in the excavation shape prediction device 10 according to the first embodiment described with reference to Fig. 3. The differences will be described below.

[0051] In the case of the excavation shape prediction device 10a according to the second embodiment, in step S102 shown in FIG. 3, the other site information acquisition unit 148 also acquires sinking state information of other sites. In step S103 shown in FIG. 3, the other site information acquisition unit 148 also acquires excavation shape information of other sites. In step S104 shown in FIG. 3, the other site information acquisition unit 148 also acquires ground information of other sites. In step S107 shown in FIG. 3, the feedback information acquisition unit 145 also attempts to acquire feedback information of other sites. In step S108 shown in FIG. 3, if feedback information of other sites has also been acquired, the learning unit 146 re-learns the prediction model using the feedback information of other sites.

[0052] As described above, the excavation shape prediction device 10 according to the first embodiment of the present invention and the excavation shape prediction device 10a according to the second embodiment include the excavation shape information acquisition unit 144, the sinking state information acquisition unit 142, and the prediction unit 147. The excavation shape information acquisition unit 144 acquires excavation shape information indicating the shape of the ground after excavation based on sensor information acquired by a sensor device provided on the body of a caisson during construction using a caisson. The sinking state information acquisition unit 142 acquires sinking state information regarding the sinking state of the caisson after it has subsided due to excavation of the ground based on the sensor information acquired by the sensor device. The prediction unit 147 predicts the excavation shape of the ground in the next excavation from the acquired excavation shape information and sinking state information using a prediction model that has learned the relationship between excavation of the ground and subsidence of the caisson.

[0053] With this configuration, even people without much experience or knowledge can easily obtain information that accurately indicates what kind of subsidence will occur depending on how they excavate, thereby reducing the occurrence of tilting of the structure due to human error and problems due to lack of experience. Therefore, the excavation shape prediction device 10 according to the first embodiment and the excavation shape prediction device 10a according to the second embodiment of the present invention make it possible to improve the accuracy of construction in construction using a caisson.

[0054] In the above-described embodiment, an example has been described in which the sensor device that acquires the excavation shape information is a range sensor, but the present invention is not limited to such an example. For example, the sensor device that acquires the excavation shape information may be a distance measurement sensor such as a radar or a lidar. The distance measurement sensor measures the shape of the object to be measured by measuring the distance to the object to be measured in the work chamber based on its own position.

[0055] The above describes the embodiments of the present invention. Note that the excavation shape prediction device 10 according to the first embodiment and the excavation shape prediction device 10a according to the second embodiment may be partially or entirely implemented by a computer. In this case, a program for implementing the functions may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed. Note that the term "computer system" as used herein includes hardware such as an OS and peripheral devices. Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into a computer system. Furthermore, the term "computer-readable recording medium" may also include devices that dynamically store programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or telephone lines, or devices that store programs for a fixed period of time, such as volatile memory within a computer system serving as a server or client. The program may be for implementing some of the functions described above, or may be capable of implementing the functions in combination with a program already stored in the computer system, or may be implemented using a programmable logic device such as an FPGA (Field Programmable Gate Array).

[0056] The embodiments of the present invention have been described in detail above with reference to the drawings, but the specific configuration is not limited to that described above, and various design changes can be made within the scope of the gist of the present invention. [Explanation of symbols]

[0057] 1, 1a... Excavation shape prediction system, 10, 10a... Excavation shape prediction device, 11... Friction sensor, 12... Air pressure sensor, 13... Tilt sensor, 14... Cutting edge reaction force sensor, 15... Range sensor, 20... Body, 21... Side wall, 22... Bottom slab, 23... Cutting edge, 30... Work chamber, 40... Rail, 41... Excavator, 50... Residual soil, 110... Input unit, 120... Communication unit, 130... Memory unit, 140, 140a...control unit, 141...sensor information acquisition unit, 142...sinking state information acquisition unit, 143...ground information acquisition unit, 144...excavation shape information acquisition unit, 145...feedback information acquisition unit, 146...learning unit, 147...prediction unit, 148...other site information acquisition unit, 150...output unit, G...ground

Claims

1. In construction using a caisson, an excavation shape information acquisition unit acquires excavation shape information indicating the shape of the ground after excavation based on sensor information acquired by a sensor device installed on the body of the caisson; A sinking state information acquisition unit that acquires sinking state information indicating information about the sinking state of the caisson after it has subsided due to excavation of the ground based on the sensor information acquired by the sensor device; A prediction unit that predicts the excavation shape of the ground in the next excavation from the acquired excavation shape information and sinking state information using a prediction model that has learned the relationship between the excavation of the ground and the subsidence of the caisson; Equipped with The prediction model predicts and outputs an excavation shape that can improve the abnormality when there is an abnormality in the submerged state after excavation based on the input excavation shape information and the submerged state information, and predicts and outputs an excavation shape that can maintain the submerged state when there is no abnormality in the submerged state after excavation. Drilling shape prediction device.

2. a ground information acquisition unit that acquires ground information indicating information on the ground to be excavated in construction using a caisson; Furthermore, The prediction model is a model that learns the relationship between the excavation of the ground and the subsidence of the caisson as well as the relationship with the ground, The prediction unit predicts the excavation shape of the ground in the next excavation from the acquired excavation shape information, submerged state information, and ground information using the prediction model. The excavation shape prediction device according to claim 1 .

3. an other site information acquisition unit that acquires other site information including excavation shape information and sinking state information at other sites; Furthermore, The prediction model is a model that learns the relationship between the excavation of the ground and the subsidence of the caisson at multiple sites, The excavation shape prediction device according to claim 1 or 2.

4. a feedback information acquisition unit that acquires, as feedback information, an evaluation of the construction work performed based on the predicted excavation shape; a learning unit that re-learns the prediction model based on the acquired feedback information; The excavation shape prediction device according to claim 1 , further comprising:

5. The feedback information acquisition unit also acquires, as feedback information, an evaluation of construction work performed at another site based on the predicted excavation shape, the learning unit re-learns the prediction model based on feedback information including evaluations at the other sites. The excavation shape prediction device according to claim 4 .

6. an excavation shape information acquisition process in which the excavation shape information acquisition unit acquires excavation shape information indicating the shape of the ground after excavation based on sensor information acquired by a sensor device installed on the body of the caisson during construction using the caisson; A sinking state information acquisition process in which a sinking state information acquisition unit acquires sinking state information indicating information about the sinking state of the caisson after it has sunk due to excavation of the ground based on the sensor information acquired by the sensor device; A prediction process in which a prediction unit predicts the excavation shape of the ground in the next excavation from the acquired excavation shape information and subsidence state information using a prediction model that has learned the relationship between the excavation of the ground and the subsidence of the caisson; Including, The prediction model predicts and outputs an excavation shape that can improve the abnormality when there is an abnormality in the submerged state after excavation based on the input excavation shape information and the submerged state information, and predicts and outputs an excavation shape that can maintain the submerged state when there is no abnormality in the submerged state after excavation. Excavation shape prediction method.

7. Computer, In construction using a caisson, an excavation shape information acquisition means acquires excavation shape information indicating the shape of the ground after excavation based on sensor information acquired by a sensor device installed on the body of the caisson; A sinking state information acquisition means for acquiring sinking state information indicating information about the sinking state of the caisson after it has subsided due to excavation of the ground based on the sensor information acquired by the sensor device; a prediction means for predicting the excavation shape of the ground in the next excavation from the acquired excavation shape information and sinking state information using a prediction model that has learned the relationship between the excavation of the ground and the subsidence of the caisson; It functions as The prediction model predicts and outputs an excavation shape that can improve the abnormality when there is an abnormality in the submerged state after excavation based on the input excavation shape information and the submerged state information, and predicts and outputs an excavation shape that can maintain the submerged state when there is no abnormality in the submerged state after excavation. program.

Images