JP2021008729A - Ground subsidence prediction system - Google Patents

Abstract

To provide a ground subsidence prediction system that can accurately predict ground subsidence at a specific point as an estimated ground subsidence point.SOLUTION: The ground subsidence prediction system comprises: a first learning unit 11 that learns the calculation of a value of a subsidence parameter at an arbitrary point from ground information at the arbitrary point by machine learning; an area information storage unit 12 that stores map data of a predetermined area and known point information; an estimated point setting unit 21 that sets the estimated point on the map; a peripheral information extraction unit 22 that extracts known point information for each known point existing around the estimated point; a peripheral subsidence parameter calculation unit 23 that causes the first learning unit 11 to calculate the value of the subsidence parameter from the extracted known point information for each known point; and a subsidence parameter estimation unit 24 that calculates the value of the subsidence parameter at the estimated point based on the subsidence parameter for each known point calculated by the peripheral subsidence parameter calculation unit 23.SELECTED DRAWING: Figure 2

Description

The present invention relates to a land subsidence prediction system for predicting land subsidence.

Conventionally, for example, it has been made to predict the ground condition of the planned construction site of an unconstructed building to be constructed. In the technique shown in Patent Document 1, first, the geological characteristic value of each layer is estimated from the boring data already obtained for a plurality of data around the ground estimation location, and the contour map of the geological characteristic value is generated from this geological characteristic value. To do. Next, the geological characteristic value of the ground estimation location is estimated from the generated contour map. From the geological characteristic value of the estimated ground location and the specification value of the structure constructed there, a predetermined calculation formula is applied to calculate the value of the influence that occurs on the ground during the construction of the structure.

Japanese Patent No. 5516211

However, in order to create a contour map around the ground estimation point (estimated point for estimating ground subsidence) in the ground subsidence prediction system shown in Patent Document 1, data of a large number of boring tests around the estimated point are acquired. There is a need to. In addition to this, even if the ground condition at the estimated point is known, it is not easy to accurately predict the ground subsidence at the estimated point.

The present invention has been made in view of these points, and in the present invention, a ground subsidence prediction system capable of accurately predicting land subsidence at an estimated point is provided by using a specific point as an estimated land subsidence point. provide.

In view of the above problems, the present invention is a ground subsidence prediction system that predicts land subsidence at a specific point as an estimated land subsidence point, and is obtained by a boring test at construction points of a plurality of existing buildings. The learning ground information including the depth from the ground surface of the construction site, the value of the soil parameter set by quantifying the soil quality according to the depth, and the N value in the standard penetration test according to the depth. , The value of the land subsidence parameter for learning set according to the ease of land subsidence based on the land subsidence amount of the existing building corresponding to the ground information for learning, as teacher data, the ground at an arbitrary point. In the first learning unit that learned the calculation of the settlement parameter value of the arbitrary point from the information by machine learning, the map data of a predetermined area, and a plurality of known points on the map of the map data, for each known point. Position information, depth from the ground surface obtained by the boring test for each known point, the value of the soil parameter set by quantifying the soil quality according to the depth, and N in the standard penetration test according to the depth. Known point information including a value is stored in the area around the stored area information storage unit, the estimated point setting unit that sets the estimated point on the map, and the estimated point set by the estimated point setting unit. The peripheral information extraction unit that extracts the known point information for each existing known point and the known point information for each known point extracted by the peripheral information extraction unit are input to the first learning unit and described. Based on the peripheral subsidence parameter calculation unit that causes the first learning unit to calculate the value of the subsidence parameter for each point and the subsidence parameter for each known point calculated by the peripheral subsidence parameter calculation unit, the estimated point It is characterized by including a settlement parameter estimation unit for calculating the value of the settlement parameter.

According to the present invention, in the first learning unit, for an existing building that has subsided so far, the learning ground information of the construction point of the existing building and the value of the subsidence parameter for learning are used as teacher data at an arbitrary point. The calculation of the value of the subsidence parameter at an arbitrary point is learned by machine learning from the ground information of. This machine learning may be performed by, for example, a deep neural network (DNN), which is one of the deep learning described later, or may be performed by using another algorithm.

Here, the construction points targeted by this teacher data are the points where the ground subsidence actually occurred, and at each construction point, the specifications of the existing building (for example, the purpose of the building, the number of times, and the building) with respect to the amount of the ground subsidence. Calculate the value of the settlement parameter for learning normalized by the numerical value of the area and the numerical value of the specifications of the foundation of the building. Therefore, this subsidence parameter is a numerical value of the ease of land subsidence for a building with the same specifications, in other words, a variable of the high risk of land subsidence. As the magnitude of the subsidence parameter increases, the land subsidence becomes easier, so the risk of land subsidence is high, and for example, a foundation structure that is difficult to subside is selected.

In the present invention, a specific point is used as an estimated point of land subsidence, and map data of a predetermined area including the estimated point and an area in which known point information for each known point is stored in a plurality of known points on the map of the map data. An information storage unit is provided.

Here, when the estimated point is set by the estimated point setting unit, the peripheral information extraction unit extracts the known point information for each known point existing around the estimated point. At this time, the peripheral information extraction unit may extract known points existing within a predetermined distance from the estimated points, or may extract a specific number of known points close to the estimated points.

Next, the peripheral subsidence parameter calculation unit inputs the known point information of these extracted known points to the first learning unit, and calculates the value of the subsidence parameter of the known points existing around the estimated point. The calculated settlement parameter value is a value that affects the settlement parameter value at the estimated point. Therefore, next, the subsidence parameter estimation unit calculates the value of the subsidence parameter of the estimated point based on the subsidence parameter of each known point existing around the estimated point.

In this way, in the present invention, the learning ground information at the construction site where the subsidence actually occurred and the value of the learning subsidence parameter set based on the ground subsidence amount of the existing building that actually subsided are taught. Since the learning is performed as data, the first learning unit can accurately learn the calculation of the value of the settlement parameter at any point including the estimated point. Then, at the estimated point, the value of the subsidence parameter of the known point around the estimated point is used to calculate the value of the subsidence parameter of the estimated point. Therefore, for example, even if there is no ground information of the estimated point, the position information of the estimated point is input. By doing so, even an inexperienced designer can reasonably and objectively predict the degree of subsidence of the estimated point with high accuracy. As a result, it is possible to prevent building defects due to land subsidence and reduce construction costs.

In a more preferred embodiment, the peripheral information extraction unit extracts the known point information for each known point existing within a predetermined distance from the estimated point, and the ground subsidence prediction system uses the peripheral information. From the number of the known points extracted by the extraction unit and the value of the settlement parameter calculated by the settlement parameter estimation unit, a test execution judgment unit that determines the necessity of performing the boring test and the standard penetration test at the estimation point is used. Further prepare.

Here, when the known point information for each known point existing within a predetermined distance from the estimated point is extracted, for example, if the number of the extracted known points is less than the predetermined number, the calculated settling of the estimated points If the value of the parameter is larger than the predetermined value, or if both of these are true, there is a high possibility of land subsidence at the estimated point. Therefore, according to this aspect, the test execution determination unit can determine the necessity of performing the boring test and the standard penetration test at the estimation point in consideration of such a case, so that the ground at the estimation point can be determined. The possibility of subsidence can be predicted more accurately.

Further, as a preferred embodiment, the ground subsidence prediction system has a value of a subsidence parameter for learning of a selection point selected from construction points of a plurality of existing buildings, and a plurality of surrounding construction points within a predetermined range from the selection point. The value of the subsidence parameter for learning and the learning position information that specifies the relative position information for each construction point in the vicinity to the selected point are used as one data group, and a plurality of data obtained for each different selected point. The data group of the above is used as the teacher data, and the value of the subsidence parameter of a plurality of peripheral points existing around the arbitrary point and the relative position information of each of the peripheral points with respect to the arbitrary point are used to obtain the data of the arbitrary point. A second learning unit in which the calculation of the settlement parameter value is learned by machine learning is further provided, and the settlement parameter estimation unit includes the settlement parameter value for each known point calculated by the peripheral settlement parameter calculation unit. The relative position information for each known point with respect to the estimated point is input to the second learning unit, and the second learning unit is made to calculate the value of the settlement parameter of the estimated point.

According to this aspect, the second learning unit has learned the calculation of the value of the subsidence parameter of a predetermined construction point from the relative position information of the construction points in the vicinity thereof and the value of the subsidence parameter. Therefore, the subsidence parameter estimation unit can cause the second learning unit to calculate the value of the subsidence parameter of the estimated point in consideration of the orientation of the surrounding known points with respect to the estimated point. This makes it possible to calculate more accurate settlement parameter values.

In a more preferable embodiment, the second learning unit includes the position information for each known point in the predetermined area stored in the area information storage unit and the predetermined area calculated by the first learning unit. The value of the subsidence parameter for each known point of is machine-learned as teacher data. According to this aspect, since the second learning unit learns the calculation of the settlement parameter value in a predetermined area including the estimation point, the accuracy of the settlement parameter value at the estimation point can be improved.

In a more preferred embodiment, the subsidence parameter estimation unit is based on the value of the subsidence parameter for each known point calculated by the peripheral subsidence parameter calculation unit and the distance from the estimated point to each known point. The value of the settlement parameter at the estimated point is calculated.

According to this aspect, the value of the subsidence parameter of the estimated point is greatly affected by the value of the subsidence parameter for each known point and the distance from the estimated point to each known point. Therefore, by using these values, , The accuracy of the value of the settlement parameter at the estimated point can be improved.

According to the present invention, it is possible to accurately predict the land subsidence at the estimated point by using a specific point as the estimated land subsidence point.

It is a schematic conceptual diagram for demonstrating the ground subsidence prediction system which concerns on 1st Embodiment of this invention. It is a block diagram of the ground subsidence prediction system shown in FIG. It is a table which shows the teacher data for learning by the settlement parameter learning part (first learning part) shown in FIG. It is a schematic conceptual diagram of the neural network of the settlement parameter learning part (first learning part) shown in FIG. It is a figure which showed an example of the map data stored in the storage part shown in FIG. It is a table which showed an example of the subsidence parameter for each known point calculated by the peripheral subsidence parameter calculation part from the ground information of each known point. It is a figure for demonstrating the calculation method of the settlement parameter estimation part shown in FIG. It is a flow chart of a series of steps by the ground subsidence prediction system shown in FIG. It is a block diagram of the ground subsidence prediction system which concerns on 2nd Embodiment. It is a figure which showed an example of the map data for explaining the learning of the settlement estimation learning part (second learning part) shown in FIG. It is a schematic conceptual diagram of the neural network of the settlement estimation learning part shown in FIG. It is a figure for demonstrating the calculation method of the settlement parameter estimation part shown in FIG. It is a flow chart of a series of steps by the ground subsidence prediction system shown in FIG.

Hereinafter, the ground subsidence prediction system according to the first and second embodiments of the present invention will be described with reference to the drawings.

[First Embodiment]
1. 1. About the device configuration of the ground subsidence prediction system 1 The ground subsidence prediction system 1 according to the present embodiment uses, for example, a main server (host computer) 10 and a mobile terminal (communication terminal) 20 to perform ground subsidence at a specific point. As an estimated point, the ground subsidence of the estimated point is predicted. The ground subsidence prediction system 1 will be described below.

The main server 10 of the ground subsidence prediction system 1 inputs data to the arithmetic device 10A having artificial intelligence that calculates (learns) the state of ground subsidence (specifically, the value of the subsidence parameter) and the arithmetic device 10A. It is equipped with an input device 10B such as a keyboard. The arithmetic unit 10A includes a storage unit 10a composed of data such as regional information of a predetermined area to be described later, a ROM or RAM in which a program machine-learned by artificial intelligence is stored, and a value of a sinking parameter by the machine-learned program. It is provided with a calculation unit 10b including a CPU or the like for calculating. Specifically, the storage unit 10a stores the program of the sinking parameter learning unit 11 and the area information stored in the area information storage unit 12 shown in FIG. 2, and the calculation unit 10b stores the program in FIG. Learning (calculation) by the sinking parameter learning unit 11 shown is executed.

The mobile terminal 20 of the ground subsidence prediction system 1 can communicate with the main server 10 via a network, and displays a calculation device 20A that calculates the value of the subsidence parameter of the estimated point and a touch panel that inputs and displays data. -Equipped with an input unit 20B. The arithmetic unit 10A is an arithmetic unit consisting of a storage unit 20a including a ROM or RAM for storing data from the main server 10 and data input by the display / input unit 20B, and a CPU or the like for calculating sinking parameters of estimated points. 20b and. Specifically, the storage unit 20a stores data from the main server 10, data input by the display / input unit 20B, an application (program) for calculation by the calculation unit 20b, and the like, and the calculation unit 20b stores the data and the like. , The peripheral information extraction unit 22, the peripheral subsidence parameter calculation unit 23, the subsidence parameter estimation unit 24, and the like shown in FIG. 2 execute calculations.

In the present embodiment, the ground subsidence prediction system 1 individually includes a main server 10 and a mobile terminal 20, and the number of the mobile terminal 20 is particularly limited as long as the above-mentioned application is installed. It is not something that is done.

Further, the ground subsidence prediction system 1 may not be composed of the main server 10 and the mobile terminal 20, but may be configured as one system. Specifically, the storage unit 10a of the main server 10 stores the data stored in the storage unit 20a of the mobile terminal 20, and the calculation unit 10b of the main server 10 performs the calculation by the calculation unit 20b of the mobile terminal 20. Therefore, the mobile terminal 20 may be omitted. Further, the mobile terminal 20 does not perform the above-mentioned storage and calculation, but the main server 10 performs all storage and calculation, and the mobile terminal 20 inputs the position information of the estimated point and the main server 10 calculates. It may serve only to perform with the output of the value of the settling parameter at the estimated point. Hereinafter, in the present embodiment, the functions of the software of the ground subsidence prediction system 1 will be described with reference to the block diagram of FIG.

In the present embodiment, the arithmetic unit 10A of the main server 10 includes a subsidence parameter learning unit (first learning unit) 11 and a regional information storage unit 12. The arithmetic unit 20A of the mobile terminal 20 includes an estimation point setting unit 21, a peripheral information extraction unit 22, a peripheral settlement parameter calculation unit 23, and a settlement parameter estimation unit 24. The data of the arithmetic unit 10A of the main server 10 and the data of the arithmetic unit 20A of the mobile terminal 20 can be transmitted and received to each other via the network.

2. 2. About the model of the land subsidence prediction system 1 2-1. Subsidence parameter learning unit 11
The subsidence parameter learning unit 11 (hereinafter referred to as "first learning unit 11") provides learning ground information at construction points P1, P2, P3, ... Of a plurality of existing buildings A, B, C, ..., And each construction point P1. , P2, P3, ... Using the values of the learning sinking parameters associated with, as teacher data, the calculation of the sinking parameter values at any point from the ground information at any point is a machine. It was learned by learning. Here, the "point" that is the target of the teacher data learned by the first learning unit 11 is a construction point existing in the construction site where the existing building was constructed, and the ground subsidence actually occurs and the amount of land subsidence is measured. It is a point in the construction site that was constructed, or a point in the construction site where the amount of land subsidence was estimated from the specifications of the existing building. The target "point" of the data input to the first learning unit 11 after learning is a known point where the boring test and the standard penetration test were performed in a predetermined area, and the data input from these test results. The data output by is the value of the settlement parameter at a known point.

Specifically, the "learning ground information" is the construction points P1, P2, P3, ... Obtained by the boring test of the construction points P1, P2, P3, ... Of the existing buildings A, B, C, ... It includes the depth from the ground surface, the value of the soil parameter set by quantifying the soil quality according to the depth, and the N value in the standard penetration test according to the depth. Here, the "depth" to be input is a "depth" that can specify the depth that constitutes the soil, such as the depth of the bottom of the layer that constitutes each soil, and the layer thickness of that soil. As long as it can be input, the value of the soil parameter and the value of "depth" associated with the N value are not particularly limited.

Generally, in a boring test, the type of soil is investigated according to the depth from the ground surface. Here, the "soil parameter" is a parameter that is quantified and set for each soil quality. For example, the soil quality includes clay, silt, fine sand, coarse sand, fine gravel, medium gravel, coarse gravel, cobble, or volta, and is quantified for each soil quality. For example, a numerical value corresponding to the particle size distribution according to the type of soil may be set. For example, as the particle size distribution according to the type of soil increases, the bearing capacity that supports the load from the building increases (it is difficult for the ground to subside), so the values are assigned in this order so that the values become smaller. .. For example, as shown in FIG. 3, the numerical values of the soil parameters are assigned to the soil A to E.

"N value in standard penetration test" means N value measured in standard penetration test (test based on JIS A 1219), converted N value based on Swedish sounding test (test based on JIS A 1212), etc. The larger the N value, the greater the bearing capacity that supports the load from the building.

As shown in FIG. 3, at each construction point P1, P2, P3, ..., The depth, the value of the soil parameter for each depth, and the N value for each depth are assigned as teacher data. Of these learning ground information, the values of the soil parameters of each construction site P1, P2, P3, ..., For example, the soil quality is recognized from the image of the boring log, and artificial intelligence using machine learning is used. May be set.

The "learning subsidence parameter" associated with each construction site is a variable that is set to a value according to the ease of land subsidence based on the amount of land subsidence of the existing building, and the larger the value of this variable, the more. It is set to a value that facilitates land subsidence.

In the present embodiment, the value of the "learning subsidence parameter" takes into account the specifications of the existing building (for example, the purpose of the building, the number of times, the building area, and the specifications of the foundation of the building) with respect to the amount of land subsidence at the construction site. It is a value normalized by a numerical value of (specification). For example, this subsidence parameter is a numerical value of the ease of land subsidence (risk of land subsidence) for buildings with the same specifications. In other words, as the magnitude of the subsidence parameter increases, the ground subsidence becomes easier, so a foundation structure that is less likely to subside is selected. Since the value of such a learning subsidence parameter is teacher data that is the basis of learning, even if an expert sets a numerical value based on the experience from the amount of subsidence of the existing building and the specifications of the existing building. Often, it may be set as follows.

Specifically, the value of the "learning subsidence parameter" may be a value obtained by dividing the amount of land subsidence of the existing building by a value corresponding to the building specifications of the existing building shown below. In the present embodiment, the "value corresponding to the building specifications of the existing building" is, for example, (1) the value of the usage parameter set by quantifying the usage of the existing building, (2) the number of floors of the existing building, and (3). It is a value obtained by multiplying the building area of the existing building and (4) the value of the foundation parameter set by quantifying the foundation type selected for the existing building.

The "use parameter" is a parameter according to the use of the existing building, and since the weight of the existing building fluctuates according to the use, for example, a hospital, a school, a factory, an office building, etc., these uses It is quantified and set according to. For example, the values of the application parameters are assigned so as to increase in order of the application in which the ground subsidence is likely to occur.

The “building floor” is the number of floors of an existing building or a non-existing building, and the “building area” is an area corresponding to the foundation of an existing building or a non-existing building.

The "foundation parameter" is a parameter set by quantifying the foundation type selected for the existing building or the non-existing building. Specifically, (1) a direct foundation structure with piles, (2) a pile draft foundation structure with piles, (3) a pile foundation structure with friction piles, and (4) support piles supported by a support layer. The pile foundation structure, etc. can be mentioned, and these foundation forms (forms of foundation structure) are quantified and set. For example, numerical values are assigned so that the numerical values become larger in the order of the foundation structure where the ground subsidence is likely to occur.

Further, in the case of a pile foundation structure, a numerical value of a size corresponding to the length of the friction pile, the type of the friction pile, etc. may be set in the order in which the ground subsidence is likely to occur. Further, in addition to the foundation structure, for example, a foundation structure whose ground has been improved by cement milk may be included, and the above-mentioned numerical values are set for this structure as well.

In this way, as the teacher data, as shown in FIG. 3, the value of the "learning subsidence parameter" for each construction point is set in association with the learning ground information for each construction point. For example, at the construction point P1, the value of the learning settlement parameter is 50, and at the construction point P2, the value of the learning settlement parameter is 32. Therefore, when a building having the same specifications is constructed at the construction point P1 and the construction point P2, the ground subsidence is more likely to occur at the construction point P1. At the construction points P1 to P4 shown in FIG. 3, the ground is easily subsided in the order of the construction points P4, P1, P3, and P2.

Using the learning ground information and the learning subsidence parameter values at each construction point P1, P2, ... Shown in FIG. 3 as teacher data, the calculation of the subsidence parameter value at any point is performed from the ground information at any point. Learn by machine learning.

In the present embodiment, in the first learning unit 11, the output values calculated for the input values of the existing buildings A, B, C, ... Constructed so far are the respective existing buildings A, B, C, ... Then, the calculation of the value of the subsidence parameter is learned by machine learning so as to converge to the value of the subsidence parameter (actual subsidence parameter) set according to the ease of land subsidence based on the actual amount of land subsidence. ..

This learning may be performed, for example, by repeatedly correcting the correction coefficient to be multiplied by each variable for a predetermined mathematical formula consisting of variables according to the number of input values of the ground information for learning. In the present embodiment, as shown in FIG. 4, the first learning unit 11 includes a deep neural network (DNN: hereinafter referred to as “neural network”) 11', and the neural network 11'is ground information at the construction site. Is used as the input value, and the value of the settlement parameter is used as the output value.

In this embodiment, the neural network 11'has an input layer 11A. The input layer 11A uses the depths N1, N2, ... From the ground surface and the soil parameter values and N values corresponding to these as input values for each construction point P1, P2, P3, P4, .... The input layer 11A is composed of a plurality of input layer neuron elements 11a according to the number of input values. The input layer neuron element 11a of the input layer 11A can be increased or decreased according to the number of input condition data, and a model of the neural network 11'is constructed according to the number of input data due to this increase or decrease.

The neural network 11'has an output layer 11E. The output layer 11E is composed of one output layer neuron element 11e that outputs the value of the settlement parameter. The neural network 11'has three intermediate layers 11B, 11C and 11D. The three intermediate layers 11B, 11C, and 11D are provided between the input layer 11A and the output layer 11E. Each intermediate layer 11B, 11C, 11D includes a plurality of intermediate layer neuron elements 11b, 11c, 11d that are directly or indirectly coupled to these elements.

In the present embodiment, the intermediate layer is composed of three layers. For example, the intermediate layer may be composed of one, two layers, four or more layers, and the intermediate layer is three layers. It is not limited to layers. Further, the intermediate layer neuron elements 11b, 11c, 11d constituting the intermediate layers 11B, 11C, 11D are the number corresponding to the number of the input layer neuron elements 11a, but the number of the input layers is increased according to the number of input data. The number of neuron elements 11a may be changed, and each intermediate layer may have a number of intermediate layer neuron elements different from the number of input data.

The intermediate layers 11B, 11C, and 11D perform a predetermined calculation using the value of the neuron parameter of the neuron element in the same layer on the input layer 11A side, and output the calculation result to the neuron element on the output layer 11E side. It is a thing. Specifically, the intermediate layer neuron elements 11b, 11c, 11d and the output layer neuron element 11e are values of neuron parameters input from the input layer neuron element 11a or the intermediate layer neuron elements 11b, 11c, 11d on the input layer 11A side. Is used to perform a predetermined operation.

Specifically, the intermediate layer neuron elements 11b, 11c, 11d that execute the calculation and the output layer neuron element 11e each have a predetermined activation function, and the input data (parameter value) is used as the activation function. The value of the neuron parameter is calculated by substituting it into the activation function. For example, in each intermediate layer neuron element 11b of the intermediate layer 11B, the input value of each input layer neuron element 11a is input as the value of the neuron parameter, and the value of the neuron parameter for each input layer neuron element 11a is calculated by the activation function. Will be done.

At this time, the value of the neuron parameter output by the intermediate layer neuron elements 11b, 11c, 11d of each intermediate layer 11B, 11C, 11D is relative to the value of the neuron parameter calculated by the activation function in the element. , It is calculated by multiplying by the weighting coefficient, and is output to the output layer neuron element 11e or the intermediate layer neuron elements 11c and 11d on the output layer 11E side.

In the present embodiment, the values of the subsidence parameters of the existing buildings A, B, C, ... Output by the output layer neuron element 11e are in advance with respect to the values of the actual subsidence parameters of the existing buildings A, B, C, ... The weighting coefficient multiplied by the value of the neuron parameter of each intermediate layer neuron element 11b, 11c, 11d is repeatedly corrected until it converges within the set range. In this way, in the present embodiment, the calculation (method) of the value of the settlement parameter can be learned in advance by machine learning. In this way, machine learning of the first learning unit 11 is executed using the learning ground information and the values of the learning subsidence parameters at the plurality of construction points P1, P2, P3, ... As teacher data, and at any point. From the ground information, a model for calculating the value of the subsidence parameter is generated.

In the present embodiment, in the intermediate layer neuron elements 11b, 11c, 11d, the weighting coefficient multiplied by the value of the neuron parameter calculated by the activation function is applied to the intermediate layer neuron elements 11b, 11c, 11d. Although provided, in addition to this, the output layer neuron element 11e may also be provided with a weighting coefficient to be multiplied by the value of the neuron parameter calculated by the activation function.

2-2. Area information storage unit 12
As shown in FIG. 5, the area information storage unit 12 stores map data of a predetermined area and known point information at a plurality of known points T1, T2, ... On the map M1 of the map data. This known point information is set by quantifying the depth from the ground surface and the soil quality according to the depth obtained by the position information for each known point T1, T2, ..., The boring test for each known point T1, T2, ... It includes the values of soil parameters and the N value in the standard penetration test according to the depth, and these data are linked for each known point. Since the ground information of the depth, the soil parameter value, and the N value shown here is the same type of information as that described in the learning ground information, detailed description thereof will be omitted. The known points on the map M1 of the map data may include the construction points used in machine learning.

Specifically, each known point is a point where a boring test and a standard penetration test are carried out, and the ground information obtained by these tests is the position information of each known point T1, T2, ... On the map M1. It is tied to. It should be noted that the building may not have been constructed at this known point in the present or in the past, and if the ground subsidence due to the building occurs at this known point, the area information storage unit 12 actually measures it. Subsidence parameters associated with the amount of land subsidence that has been or are estimated may be further stored. The predetermined area that is the target of the map data may be, for example, an area such as a ward or a city, or an area that continuously includes one or a plurality of common soil layers. Such map data is transmitted from the main server 10 to the mobile terminal 20, and is displayed on the display / input unit 20B of the mobile terminal 20.

2-3. Estimated point setting unit 21
As shown in FIG. 5, the estimation point setting unit 21 sets the estimation point G on the map M1. In the present embodiment, the operator inputs the position information of the estimated point G on the map M1 displayed on the display / input unit 20B of the mobile terminal 20. As a result, the estimated point G on the map M1 is set in the arithmetic unit 20A of the mobile terminal 20.

2-4. Peripheral information extraction unit 22
As shown in FIGS. 2 and 5, the peripheral information extraction unit 22 extracts known point information for each of the known points T1, T7, T8, ... Existing around the estimated point G set by the estimated point setting unit 21. (See FIG. 6). In the present embodiment, as shown in FIGS. 5 and 7, known point information for each known point T1, T7, T8, ... Existing within a predetermined range RG including the estimated point G is extracted.

Here, in the present embodiment, the peripheral information extraction unit 22 extracts known point information for each known point existing within a predetermined distance r from the estimated point G. Specifically, as shown in FIGS. 5 and 7, the region extracted by the peripheral information extraction unit 22 is a circular region (within a circle having a radius of 3 km) within a certain distance r (for example, 3 km) from the estimation point G. Is. However, the peripheral information extraction unit 22 may extract known point information of a specific (constant) number of known points in ascending order of distance from the estimated point G, for example.

2-5. Peripheral subsidence parameter calculation unit 23
The peripheral subsidence parameter calculation unit 23 inputs the known point information for each of the known points T1, T7, T8, ... Extracted by the peripheral information extraction unit 22 shown in FIG. 6 into the first learning unit 11, and the known points shown in FIG. The first learning unit 11 is made to calculate the value of the settlement parameter for each of the points T1 and T7. As described above, the first learning unit 11 has learned the calculation of the value of the subsidence parameter of the arbitrary point from the ground information of the arbitrary point, so that the subsidence parameter for each known point T1, T7, T8, ... The value of can be calculated accurately.

Here, if the value of the actual subsidence parameter is already stored in the extracted known point in the area information storage unit 12, the peripheral subsidence parameter calculation unit 23 will perform the subsidence parameter of the known point. It is not necessary to calculate the value. Further, as another aspect, even when the actual settlement parameter value not based on the calculation is already stored, the peripheral settlement parameter calculation unit 23 can perform the settlement parameter value at the known point. The value of the settlement parameter calculated at another known point may be corrected from the difference or ratio between the calculated value of the settlement parameter and the actual value of the settlement parameter.

2-6. Subsidence parameter estimation unit 24
As shown in FIG. 7, the subsidence parameter estimation unit 24 calculates the value of the subsidence parameter of the estimation point G based on the subsidence parameters for each known point T1, T7, T8, ... Calculated by the peripheral subsidence parameter calculation unit 23. To do. Here, the value of the settling parameter of the estimated point G may be calculated by, for example, the average value of the settled parameters for each of the calculated known points T1, T7, T8, ....

However, in the present embodiment, the values of the settlement parameters for each known point T1, T7, T8, ... Calculated by the peripheral settlement parameter calculation unit 23 and the distances L1, L7, L8 for each known point extracted from the estimated point G. Based on ..., The value of the subsidence parameter of the estimated point G is calculated. As a result, the value of the subsidence parameter of the estimated point G is the value of the subsidence parameter for each known point T1, T7, T8, ... And the distance L1 for each known point T1, T7, T8, ... Extracted from the estimated point G. , L7, L8 ... Have a great influence, and by using these values, the accuracy of the value of the settlement parameter at the estimation point G can be improved.

More specifically, in the present embodiment, the subsidence parameter estimation unit 24 sets the subsidence parameter so that the value of the subsidence parameter of the estimated point G decreases as the distance of each known point extracted from the estimated point increases. After correcting the values of, the average value of these values is estimated as the value of the settlement parameter at the estimation point G. Thereby, the value of the subsidence parameter of the known point closer to the estimated point G can be reflected in the value of the subsidence parameter of the estimated point G. Such a calculation may be performed based on the number 1 shown below.

Here, DG is the value of the subsidence parameter of the estimated point, Dk is the value of the subsidence parameter of each extracted known point, and Lk is the distance of each extracted known point from the estimated point. Is the distance to extract the known points from the estimated points, and N is the number of the extracted known points. Note that α is a correction coefficient, and this correction coefficient is, for example, a constant set so that the value of the sinking parameter at the estimated point calculated by Equation 1 matches the value of the actual sinking parameter at the estimated point. Is. However, this formula does not include the Dk of the known points not extracted by the peripheral information extraction unit 22, and the known points not extracted are not counted in the number of N.
Therefore, when this number 1 is applied in the case shown in FIG. 7, it becomes the following number 2.

Here, when the values of the actual subsidence parameters are already stored in some of the extracted known points in the area information storage unit 12, the subsidence parameter estimation unit 24 uses the subsidence parameters of the estimated points. The actual settlement parameters may be used to calculate the value of.

3. 3. About the ground subsidence prediction method using the ground subsidence prediction system 1 The ground subsidence prediction method using the ground subsidence prediction system 1 according to the present embodiment will be described below with reference to FIG. First, in step S1, the first learning unit 11 performs machine learning using the learning ground information and the value of the learning subsidence parameter. As a result, a model of the neural network 11'that outputs the value of the subsidence parameter from the ground information is constructed.

Next, in step S2, the area information storage unit 12 stores the map data of a predetermined area and the known point information at a plurality of known points T1, T2, ... On the map of the map data. The main server 10 executes the steps S1 and S2 in any order. Next, in step S3, the application of the mobile terminal 20 is started, and the map M of the map data stored in the main server 10 is displayed together with the known points T1, T2, ....

Next, in step S4, the operator inputs the position information of the estimated point G on the map M displayed on the mobile terminal 20. As a result, the estimated point setting unit 21 sets the estimated point G on the map M by associating the position information of the estimated point G with the map data. Next, in step S5, the peripheral information extraction unit 22 extracts known point information (peripheral information) for each of the known points T1, T7, T8, ... Existing around the estimated point G.

Next, in step S6, the peripheral subsidence parameter calculation unit 23 transmits ground information for each known point T1, T7, T8, ... From the mobile terminal 20 to the arithmetic unit 10A of the main server 10 via the network, and the main The first learning unit 11 in the server 10 calculates the subsidence parameter values D1, D7, D8, ... For each known point T1, T7, T8, .... The first learning unit 11 transmits the calculated result to the arithmetic unit 20A of the mobile terminal 20 via the network, and the result is input to the settlement parameter estimation unit 24.

Next, in step S7, the subsidence parameter estimation unit 24 calculates the distances L1, L7, L8 ... For each known point extracted from the estimation point G. In step S8, the subsidence parameter estimation unit 24 sets the subsidence parameter values D1, D7, D8, ... For each known point T1, T7, T8, ..., And the distances L1, L7, extracted from the estimated point G to the known point. The value of the subsidence parameter of the estimated point G is calculated based on L8. Finally, in step S9, the value of the sinking parameter of the estimated point G is input to the display / input unit 20B of the mobile terminal 20, and the result is displayed.

In this embodiment, the learning ground information at the construction site where the subsidence actually occurred and the value of the subsidence parameter for learning set based on the subsidence amount of the existing building that actually subsided are learned as teacher data. , The first learning unit 11 can accurately learn the calculation of the value of the settlement parameter at an arbitrary point. At the estimated point G, the value of the subsidence parameter of the estimated point G is calculated using the value of the subsidence parameter of the known points around it. Therefore, even if there is no ground information of the estimated point G, the estimated point G is subsided. The degree of ease can be predicted with high accuracy.

[Second Embodiment]
The ground subsidence prediction system 1 according to the second embodiment of the present invention will be described below with reference to FIGS. 9 to 13. The difference between the ground subsidence prediction system 1 according to the second embodiment and that of the first embodiment is that (1) as shown in FIG. 9, a subsidence estimation learning unit 13 is newly provided and subsidence is settled by using the subsidence estimation learning unit 13. Teacher data of the points where the parameter estimation unit 24 calculated the settlement parameters, (2) the points where the test execution judgment unit 25 was newly provided, and (3) the known point information of the predetermined area for the learning of the first learning unit. It is a point corrected (re-learned) by using as. Therefore, the same configuration as that of the first embodiment will be omitted in detail, and different configurations will be described below.

The subsidence estimation learning unit 13 (hereinafter referred to as "second learning unit 13") has the values of the subsidence parameters of a plurality of peripheral points existing around an arbitrary point and the relative position of each peripheral point with respect to the arbitrary point. From the information, the calculation of the value of the subsidence parameter at an arbitrary point was learned by machine learning.

As shown in FIG. 10, for example, the teacher data of the second learning unit 13 includes the values of the learning sinking parameters of the selected points selected from the construction points W1 to W37 of the plurality of existing buildings and the values within a predetermined range from the selected points. A set of teacher data (one data group) of the values of the learning sinking parameters of a plurality of surrounding construction points and the learning position information that specifies the relative position information of each surrounding construction point with respect to the selected point. ), And it is a plurality of data groups obtained for each different selection point.

For example, in FIG. 10, the construction point W1 on the map M2 of a predetermined area is selected as the selection point. For the selected construction point W1, the surrounding construction points W6, W7, W9, ... Within a predetermined range R1 (for example, distance r) are extracted from the construction point W1. Then, the value of the subsidence parameter of the construction point W1 and the value of the subsidence parameter for each of the surrounding construction points W6, W7, W9, ... Are used as the value of the subsidence parameter for learning. It is used as learning position information that specifies the relative positions of the surrounding construction points W6, W7, W9, ... With respect to the construction point (selection point) W1. The learning position information of the surrounding construction points W6, W7, W9, ... May be the coordinates of the Cartesian coordinate system with the selected point as the origin, or may be the coordinates of the polar coordinate system. The construction site used in the learning of the second learning unit may include the construction site used in the learning of the first learning unit.

Using these as a set of teacher data, as shown in FIG. 11, the values of the learning sinking parameters for each of the surrounding construction points W6, W7, W9, ..., And the learning of the surrounding construction points W6, W7, W9, ... The position information is input to the neural network 13', and the calculation of the value of the sinking parameter is learned by machine learning so that the output value converges to the value of the learning sinking parameter of the construction point W1. Teacher data is extracted by selecting such learning as a selection point for a plurality of other construction points (for example, W4, W36, etc. in FIG. 10), and the teacher data is used to determine the sinking parameter. The calculation of the value is learned by machine learning.

The neural network 13'of the second learning unit 13 also includes an input layer 13A, intermediate layers 13B to 13D, and an output layer 13E, similarly to the neural network 11'of the first learning unit 11. These layers are composed of neuron elements 13a to 13e, and perform machine learning by repeatedly correcting a weighting coefficient that is multiplied by a value of a neuron parameter.

Here, for example, the value of the subsidence parameter used for the teacher data at each construction point W1 to W34 may be the value of the subsidence parameter of each construction point calculated by the first learning unit 11, and the actual value of each point may be used. It may be the value of the settlement parameter. Here, among the teacher data, it is preferable to set the construction point where the actual sinking parameter value is stored as the selection point. As a result, the machine learning of the second learning unit 13 is performed so as to converge to the actual value of the settlement parameter, so that the value of the settlement parameter can be calculated with higher accuracy.

Further, in the present embodiment, the second learning unit 13 extracts seven construction points W6, W7, W9, ... Existing from the construction points W1 within a predetermined range R1 with respect to the selected construction points W1. However, by selecting another construction site or changing the size of a predetermined range, the number of construction sites around this area can be compared to multiple other numbers (numbers other than 7). , The same learning may be performed. As a result, the value of the subsidence parameter of the estimated point G can be calculated more accurately by the second learning unit 13, which will be described later, according to the number of known points extracted by the peripheral information extraction unit 22.

Here, the second learning unit 13 uses the position information for each of the known points T1 to T33 in the predetermined area stored in the area information storage unit 12 as teacher data, and the known points calculated in the first learning unit 11. It may be machine-learned using the values of the settlement parameters for each of T1 to T33. As a result, the second learning unit 13 learns the calculation of the value of the sinking parameter in the predetermined area including the estimated point G, so that the accuracy of the value of the sinking parameter of the estimated point G can be improved.

In the present embodiment, as shown in FIG. 12, the subsidence parameter estimation unit 24 uses the subsidence parameter values D1, D7, T8, ... For each known point T1, T7, T8, ... Calculated by the peripheral subsidence parameter calculation unit 23. And the relative position information (X1, Y1), (X7, Y7), (X8, Y8), ... For each known point with respect to the estimated point are input to the second learning unit 13, and the second learning unit 13 To calculate the value of the subsidence parameter of the estimated point G.

Here, as shown in FIG. 12, the position information for each known point T1, T7, T8, ... May be a coordinate in Cartesian coordinates with the estimated point G as the origin (0,0), and the estimated point. It may be the coordinates in the polar coordinate system with G as the origin.

In the second learning unit 13, for the selected construction point W1, a model of the neural network generated according to the number of construction points W6, W7, W9, ... Existing within a predetermined range R1 from the construction point W1 is generated. Select and use this model to calculate the value of the subsidence parameter at the estimated point.

From the number of known points extracted by the peripheral information extraction unit 22 and the value of the subsidence parameter calculated by the subsidence parameter estimation unit 24, the test execution determination unit 25 determines whether or not to perform the boring test and the standard penetration test at the estimated points. To judge. Specifically, when the number of known points extracted by the peripheral information extraction unit 22 is less than a predetermined number (for example, less than 6 points), the number of extracted points is small, so that the test execution determination unit 25 , It is judged that it is necessary to carry out the boring test and the standard penetration test at the estimated point G.

On the other hand, when the number of known points extracted by the peripheral information extraction unit 22 is a predetermined number or more (for example, 6 or more points), the extracted points are sufficient, and the test execution determination unit 25 estimates. It is determined whether or not the boring test and the standard penetration test at point G are performed. For example, in the present embodiment, since the number of known points is 7, it is determined whether or not the test is carried out.

Further, even if it is determined that the test is not carried out, if the value of the subsidence parameter calculated by the subsidence parameter estimation unit 24 exceeds a predetermined value (for example, exceeds 30), the ground Since there is a high possibility of subsidence, the test execution determination unit 25 determines that it is necessary to carry out the boring test and the standard penetration test at the estimation point G. On the other hand, when the value of the subsidence parameter calculated by the subsidence parameter estimation unit 24 is equal to or less than a predetermined value (for example, 30 or less), the possibility of ground subsidence is low, so the test execution determination unit 25 determines the estimation point G. It is judged whether or not the boring test and the standard penetration test are performed.

In this way, when the number of known points extracted by the peripheral information extraction unit 22 is less than a predetermined number, or when the value of the subsidence parameter calculated by the subsidence parameter estimation unit 24 exceeds a predetermined value. In addition, the test execution determination unit 25 determines that it is necessary to carry out the boring test and the standard penetration test at the estimation point G. As a result, the possibility of land subsidence at the estimated point G can be predicted more accurately.

The ground subsidence prediction method using the ground subsidence prediction system 1 according to the present embodiment will be described below with reference to FIG. In steps S1 and S2, the same procedure as in steps S1 and S2 shown in FIG. 8 is performed. In the present embodiment, in step S21, for the first learning unit 11 that has undergone machine learning in S1, among the data of the known points stored in step S2, for the known point where the ground subsidence actually occurred. The ground information and the subsidence parameter are used as teacher data, and learning by the first learning unit 11 is performed again. As a result, the model generated in step S1 is corrected, and the tendency of land subsidence in the predetermined area can be relearned by the first learning unit 11.

In steps S3 to S6, the same procedure as in steps S3 to S6 shown in FIG. 8 is performed. Next, in step S61, the subsidence parameter estimation unit 24 calculates the coordinates of the known points as shown in FIG.

In step S81, the sinking parameter estimation unit 24 moves the position information (X1, Y1), (X7, Y7), (X8, Y8) for each known point T1, T7, T8, ... From the mobile terminal 20 via the network. ), ... And the values of the sinking parameters D1, D7, D8, ... Associated with this are transmitted to the arithmetic unit 10A of the main server 10. The second learning unit 13 in the main server 10 calculates the value of the sinking parameter of the estimated point G. The second learning unit 13 transmits the calculated result to the arithmetic unit 20A of the mobile terminal 20 via the network, and the result is input to the test execution determination unit 25.

Next, in step S82, whether or not the test execution determination unit 25 needs to carry out the boring test and the standard penetration test at the estimated point G from the number of the extracted known points and the value of the settling parameter of the estimated point G. To judge. Finally, in step S9, the value of the sinking parameter at the estimated point G and the result of the necessity of conducting the test are input to the display / input unit 20B of the mobile terminal 20, and this result is displayed.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and various designs are designed without departing from the spirit of the present invention described in the claims. You can make changes.

In the second embodiment, the test execution determination unit determines the necessity of performing the boring test and the standard penetration test at the estimated points from the number of the extracted known points and the calculated values of the sinking parameters of the estimated points. For example, the test execution determination unit uses one or more selected from the average value of the distances from the estimated points to the extracted known points, the number of extracted known points, or the calculated value of the settling parameter of the estimated points. It may be determined whether or not a boring test and a standard penetration test are performed at the estimated point. Further, the first learning unit may be used to create a map showing the value of the subsidence parameter associated with the map data in a predetermined area.

1: Ground subsidence prediction system, 10: Main server, 20: Mobile terminal, 11: Subsidence parameter learning unit (1st learning unit), 12: Regional information storage unit, 13: Subsidence estimation learning unit (2nd learning unit), 21: Estimated point setting unit, 22: Peripheral information extraction unit, 23: Peripheral subsidence parameter calculation unit, 24: Subsidence parameter estimation unit, 25: Test execution judgment unit, G: Estimated point, T1 to T33: Known point

Claims (5)

It is a ground subsidence prediction system that predicts the land subsidence at a specific point as the estimated land subsidence point.
The depth from the ground surface of the construction site obtained by the boring test at the construction site of a plurality of existing buildings, the value of the soil parameter set by quantifying the soil quality according to the depth, and the standard penetration according to the depth. The value of the learning subsidence parameter set according to the ease of ground subsidence based on the learning ground information including the N value in the test and the ground subsidence amount of the existing building corresponding to the learning ground information. The first learning unit, which learned the calculation of the value of the subsidence parameter at an arbitrary point from the ground information at an arbitrary point by machine learning, using
Depending on the map data of a predetermined area and a plurality of known points on the map of the map data, the position information for each known point, the depth from the ground surface obtained by the boring test for each known point, and the depth. The area information storage unit where the known point information including the value of the soil parameter set by quantifying the soil quality and the N value in the standard penetration test according to the depth is stored is
An estimated point setting unit that sets the estimated point on the map,
Peripheral information extraction unit that extracts the known point information for each known point existing around the estimated point set by the estimated point setting unit, and
Peripheral subsidence parameter that inputs the known point information for each known point extracted by the peripheral information extraction unit to the first learning unit and causes the first learning unit to calculate the value of the subsidence parameter for each known point. Calculation part and
A ground subsidence prediction system including a subsidence parameter estimation unit that calculates the value of the subsidence parameter at the estimated point based on the subsidence parameter for each known point calculated by the peripheral subsidence parameter calculation unit. .. The peripheral information extraction unit extracts the known point information for each known point existing within a predetermined distance from the estimated point.
The ground subsidence prediction system carries out a boring test and a standard penetration test at the estimated points based on the number of the known points extracted by the peripheral information extraction unit and the values of the subsidence parameters calculated by the subsidence parameter estimation unit. The ground subsidence prediction system according to claim 1, further comprising a test execution determination unit for determining necessity. The ground subsidence prediction system has a value of a learning subsidence parameter of a selected point selected from construction points of a plurality of existing buildings and a value of a learning subsidence parameter of a plurality of surrounding construction points within a predetermined range from the selected point. And the learning position information that specifies the relative position information for each construction point in the vicinity to the selected point are used as one data group, and a plurality of data groups obtained for each different selected point are used as teacher data. , Calculation of the value of the subsidence parameter of the arbitrary point from the value of the subsidence parameter of a plurality of peripheral points existing around the arbitrary point and the relative position information of each of the peripheral points with respect to the arbitrary point. It also has a second learning department learned by machine learning.
The subsidence parameter estimation unit obtains the value of the subsidence parameter for each known point calculated by the peripheral subsidence parameter calculation unit and the relative position information for each known point with respect to the estimated point. The ground subsidence prediction system according to claim 1 or 2, wherein the value of the subsidence parameter of the estimated point is calculated by the second learning unit. The second learning unit includes position information for each known point in the predetermined area stored in the area information storage unit and subsidence for each known point in the predetermined area calculated by the first learning unit. The ground subsidence prediction system according to claim 3, wherein the value of the parameter is machine-learned as teacher data. The subsidence parameter estimation unit is based on the value of the subsidence parameter for each known point calculated by the peripheral subsidence parameter calculation unit and the distance from the estimated point to each known point, and the subsidence parameter of the estimated point. The land subsidence prediction system according to claim 1 or 2, wherein the value of is calculated.

Images