Member cross-section estimation device, member cross-section estimation method, and program
The member cross-section estimation device uses feature quantities and machine learning to efficiently estimate cross-sectional shapes of planned buildings, addressing large processing loads in structural design optimization systems.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- SHIMIZU CORP
- Filing Date
- 2022-01-25
- Publication Date
- 2026-07-22
AI Technical Summary
Existing structural design optimization systems face large processing amounts due to inappropriate consideration ranges of member cross-sections, necessitating a method to estimate reference cross-sections for planned buildings.
A member cross-section estimation device and method that calculates feature quantities for reference and target buildings, utilizing machine learning to estimate cross-sectional shapes without structural analysis data, using principal component analysis to associate feature quantities with cross-sectional shapes.
Enables efficient estimation of cross-sections for planned buildings by leveraging feature quantities and machine learning, reducing computational load and improving accuracy.
Smart Images
Figure 0007893614000001 
Figure 0007893614000002 
Figure 0007893614000003
Abstract
Description
Technical Field
[0001] The present invention relates to a member cross-section estimation device, a member cross-section estimation method, and a program.
Background Art
[0002] Patent Document 1 discloses a structural design optimization system that identifies an optimal combination of cross-sections of members constituting a structure to be built. This structural design optimization system includes a consideration range setting unit that sets a consideration range for the cross-sections of members constituting the structure, a structural analysis unit that sets a combination of cross-sections of members constituting the structure within the consideration range and performs a structural analysis on the set combination of cross-sections, a verification value determination unit that determines whether a verification value obtained by the structural analysis satisfies a predetermined verification condition, and a comparison value calculation unit that calculates a comparison value for performing a comparison based on a predetermined optimization determination condition for combinations of cross-sections of members within the consideration range that satisfy the predetermined verification condition. Based on the comparison value, it has a specific processing unit that identifies an optimal combination of cross-sections of members in accordance with a predetermined optimization determination condition.
Prior Art Documents
Patent Documents
[0003]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0004] However, in the above-described structural design optimization system, since the consideration range of the cross-sections of members is set and a structural analysis is performed on the combination of cross-sections within that consideration range, if the consideration range is not appropriate, the processing amount of the structural analysis may become extremely large. And in a planned building, there is a problem that it is necessary to estimate a reference cross-section in order to make this consideration range appropriate.
[0005] This invention has been made in view of these circumstances and provides a member cross-section estimation device, a member cross-section estimation method, and a program that can easily estimate the cross-section of a member of a building under planning. [Means for solving the problem]
[0006] This invention was made to solve the above-mentioned problems, and one aspect of the present invention comprises: a first member feature calculation unit that calculates a plurality of feature quantities for each of a plurality of members of a reference building based on information about the frame of the reference building from among the information about the reference building; a second member feature calculation unit that calculates a plurality of feature quantities for members of a target building based on information about the frame of the target building from among the information about the target building; and a member cross-section estimation unit that estimates the cross-sectional shape of a member of a target building from the cross-sectional shapes of a plurality of members of the reference building without using data related to structural analysis results, wherein each of the plurality of feature quantities for each of the plurality of members of the reference building and the plurality of feature quantities for members of the target building is a value indicating the floor on which the member is installed, a value indicating whether the member is a beam or a column, a value indicating the combination of the joining directions of the beams and columns joined at both ends of the member, a value indicating the sum of the member lengths of the beams joined at one end of the member, and a value indicating the sum of the member lengths of the beams joined at the other end of the member. Furthermore This indicates the length of the component. Value include.
[0007] Another aspect of the present invention is the member cross-section estimation device described above, wherein the member cross-section estimation unit includes a machine learning unit that performs machine learning to estimate the cross-section of the target building member from a plurality of feature quantities calculated by the second member feature quantity calculation unit, using at least a plurality of feature quantities calculated by the first member feature quantity calculation unit, and estimates the cross-sectional shape of the target building member using the results of machine learning by the machine learning unit.
[0008] Another aspect of the present invention is the member cross-section estimation device described above, wherein the machine learning unit performs principal component analysis on a plurality of feature quantities calculated by the first member feature quantity calculation unit, and the member cross-section estimation unit extracts from a plurality of reference building members that are closest in distance to the principal component of the target building member in the space of the principal component of the analysis result by the machine learning unit, and the cross-sectional shape of the extracted member is used as the estimated cross-sectional shape.
[0009] Another aspect of the present invention is a method for estimating a member cross-section using a member cross-section estimation device, comprising: a first step of calculating a plurality of feature quantities for each of a plurality of members of a reference building based on information about the frame of the reference building from among the information about the reference building; a second step of calculating a plurality of feature quantities for members of a target building based on information about the frame of the target building from among the information about the target building; and a third step of estimating the cross-sectional shape of a member of a target building from the cross-sectional shapes of a plurality of members of the reference building without using data related to structural analysis results, wherein each of the plurality of feature quantities for each of the plurality of members of the reference building and the plurality of feature quantities for members of the target building is a value indicating the floor on which the member is installed, a value indicating whether the member is a beam or a column, a value indicating the combination of the joining directions of the beams and columns joined at both ends of the member, a value indicating the sum of the member lengths of the beams joined at one end of the member, and a value indicating the sum of the member lengths of the beams joined at the other end of the member. Furthermore This indicates the length of the component. Value include.
[0010] Another aspect of the present invention is a program for a computer to function as a first member feature calculation unit that calculates a plurality of feature quantities for each of a plurality of members of a reference building based on structural information of the reference building, from among the information of the reference building; a second member feature calculation unit that calculates a plurality of feature quantities for members of a target building based on structural information of the target building, from among the information of the target building; and a member cross-section estimation unit that estimates the cross-sectional shape of a member of a target building from the cross-sectional shapes of a plurality of members of the reference building, without using data relating to structural analysis results, wherein each of the plurality of feature quantities for each of the plurality of members of the reference building and the plurality of feature quantities for members of the target building is a value indicating the floor on which the member is installed, a value indicating whether the member is a beam or a column, a value indicating the combination of the joining directions of beams and columns joined at both ends of the member, a value indicating the sum of the member lengths of beams joined at one end of the member, and a value indicating the sum of the member lengths of beams joined at the other end of the member. Furthermore This indicates the length of the component. Value include. [Effects of the Invention]
[0011] According to this invention, it is possible to estimate the cross-section of the structural members of a planned building. [Brief explanation of the drawing]
[0012] [Figure 1] This is a schematic block diagram showing the functional configuration of a member cross-section estimation device 10 according to one embodiment of the present invention. [Figure 2] This is a flowchart illustrating the operation of the first member feature quantity calculation unit 12 in the same embodiment. [Figure 3] This is a schematic diagram illustrating the types of joints a0 to a11 in the same embodiment. [Figure 4] This is a schematic diagram illustrating the total beam member lengths at ends i and j in the same embodiment. [Figure 5]This graph illustrates the operation of the member cross-section estimation unit 16 in the same embodiment. [Modes for carrying out the invention]
[0013] The embodiments of the present invention will be described below with reference to the drawings. Figure 1 is a schematic block diagram showing the functional configuration of a member cross-section estimation device 10 according to one embodiment of the present invention. The member cross-section estimation device 10 estimates the cross-sections of members of a planned building using information on the framework of a reference building and information on the cross-sections of each member of that reference building. In the case of a planned building, the configuration of columns and beams, i.e., the framework, is determined, but the cross-sectional shapes of each column and beam are not determined. Furthermore, it is desirable that the reference building has a framework similar to that of the planned building.
[0014] As shown in Figure 1, the member cross-section estimation device 10 comprises a reference building information input unit 11, a first member feature quantity calculation unit 12, a target building information input unit 14, a second member feature quantity calculation unit 15, a member cross-section estimation unit 16, and a member information output unit 17. The reference building information input unit 11 accepts input of information about a reference building. The information about the reference building includes information about the building's frame and information about the cross-sections of each of its members. The information about the building's frame includes at least information representing the length of each of the columns and beams constituting the frame, and information representing the connection of each of the columns and beams. The information about the cross-sections of each member is information representing the cross-sectional shape of each of the columns and beams. The input method may be to have the operator specify a location in a storage device such as a flash memory or hard disk connected to the member cross-section estimation device 10, or a location on a network, and read a file from that location, or to receive it from another device.
[0015] The first member feature quantity calculation unit 12 calculates a plurality of feature quantities for each of the plurality of members of the reference building. The first member feature quantity calculation unit 12 calculates these feature quantities using the structural information among the information of the reference building received by the reference building information input unit 11. In the present embodiment, the plurality of calculated feature quantities include a value indicating the floor where the member is installed, a value indicating whether the member is a beam or a column, a value indicating the combination of the joining directions of the beams and columns joined at both ends of the member, a value indicating the total length of the members of the beams joined at one end of the member, a value indicating the total length of the members of the beams joined at the other end of the member, and a value indicating the length of the member. Note that the plurality of calculated feature quantities only need to include at least one of these, and may include other values. Note that the value indicating the combination of the joining directions may be held as two independent values.
[0016] The target building information input unit 14 receives the input of information on the target building, such as the planned building. The information on the target building includes the structural information of the building. The structural information of the building is the same as in the case of the reference building, and at least includes information representing the length of each of the columns and beams constituting the structure and information representing the joining of the columns and beams. Also, the input method in the target building information input unit 14 is the same as that in the reference building information input unit 11, but may be different from the reference building information input unit 11.
[0017] The second member feature quantity calculation unit 15 calculates a plurality of feature quantities for the members of the target building. The second member feature quantity calculation unit 15 calculates these feature quantities using the structural information among the information of the reference building received by the reference building information input unit 11. The feature quantities calculated by the second member feature quantity calculation unit 15 and the calculation method of these feature quantities are the same as those of the first member feature quantity calculation unit 12.
[0018] The member cross-section estimation unit 16 estimates the cross-sectional shape of the members of the target building from the cross-sectional shapes of the members of the reference building using the feature quantities calculated by the first member feature quantity calculation unit 12 and the second member feature quantity 15. The member cross-section estimation unit 16 includes a machine learning unit 13.
[0019] The machine learning unit 13 performs machine learning for estimating the cross-section of the member of the target building from a plurality of feature amounts calculated by the second member feature amount calculation unit 15, using at least the plurality of feature amounts calculated by the first member feature amount calculation unit 12. As a machine learning method, any method such as cluster analysis, principal component analysis, neural network, etc. may be used. Here, the case of using principal component analysis will be described. First, the machine learning unit 13 performs principal component analysis on the plurality of feature amounts calculated by the first member feature amount calculation unit 12. Next, the machine learning unit 13 associates the value of the principal component obtained by the principal component analysis with the cross-sectional shape. This association is performed by associating the cross-sectional shape of the member with the value of the principal component corresponding to the plurality of feature amounts of each member of the reference building. Note that the principal component calculated by the principal component analysis may be only the first principal component or may include the second principal component and subsequent components.
[0020] The member cross-section estimation unit 16 estimates the cross-sectional shape of the member of the target building from a plurality of feature amounts related to the member of the target building, using the result of the machine learning by the machine learning unit 13. In this embodiment, the case where the principal component analysis is used in the machine learning unit 13 will be described. The member cross-section estimation unit 16 extracts, from among the plurality of members of the reference building, the member having the shortest distance from the principal component of the member of the target building in the principal component space of the analysis result by the machine learning unit 13, and sets the cross-sectional shape of the extracted member as the cross-sectional shape of the estimation result. In the calculation formula of the principal component of the member of the target building, those obtained by performing principal component analysis on the plurality of feature amounts related to each member of the reference building in the machine learning unit 13 are used. Note that the member cross-section estimation unit 16 may not include the machine learning unit 13, and may use, as the estimation result of the cross-sectional shape of the member of the target building, the cross-sectional shape of the member having the minimum normalized Euclidean distance between the feature amounts of the member of the target building and the plurality of members of the reference building.
[0021] The component information output unit 17 outputs component information for the component of the target building, including information showing the cross-sectional shape estimated by the component cross-section estimation unit 16. The output method may be display on a screen, output to a file, or transmitted to another device.
[0022] Figure 2 is a flowchart illustrating the operation of the first member feature calculation unit 12 in this embodiment. The second member feature calculation unit 15 operates similarly to the first member feature calculation unit 12, but instead of reference building information, it uses information about the target building. The first member feature calculation unit 12 processes each member of the reference building from step S2 to step S7 (step S1).
[0023] In step S2, the first member feature calculation unit 12 obtains a value indicating the floor on which the member is installed. The floor on which it is installed may be included in the structural information, or it may be estimated from the information representing the connections of each column and beam. The value indicating the floor may be, for example, "1" for the first floor and "2" for the second floor, but the closer the floors, the closer the values should be, and other numerical values may also be assigned.
[0024] In step S3, the first member feature calculation unit 12 obtains a value indicating the type of member (a value indicating whether it is a beam or a column). The type of member may be included in the frame information, or it may be estimated from the information representing the connections of the columns and beams. For example, the value indicating the type of member may be "1" for a column and "2" for a beam. Alternatively, instead of separating columns and beams into "1" and "2" respectively, the data may be treated as data in the column and beam categories, and the cross-section may be extracted for each column and beam separately.
[0025] In step S4, the first member feature calculation unit 12 obtains values indicating the connection types at the i and j ends of the member. The connection type is a combination of the connection directions of the beams and columns that are connected. In this embodiment, connection types a0 to a11 are defined as connection types for each end, and the values indicating the connection types for each end are set to "0" to "11" in order. Of the two ends of the member, the end with the smaller value indicating the connection type is designated as the i end, and the other end as the j end. The values indicating the connection types at the i and j ends are the sum of the value indicating the connection type at the i end multiplied by 100 and the value indicating the connection type at the j end. For example, if the connection type at the j end is connection type a3 and the connection type at the j end is connection type a8, then the values indicating the connection types at the i and j ends will be 3 × 100 + 8 = 308. The first member feature calculation unit 12 uses the frame information to determine the connection types at the i and j ends of the member.
[0026] Figure 3 is a schematic diagram illustrating connection types a0 to a11 in this embodiment. In Figure 3, the x, y, and z axes are, for example, the x-axis representing the direction of the long side in the horizontal plane of the building, the y-axis representing the direction of the long side in the horizontal plane of the building, and the z-axis representing the vertical direction (direction of gravity). The orientation of these axes is such that the positive direction of the z-axis is opposite to gravity, and the positive directions of the x and y axes are determined so that the x, y, and z axes are right-handed. The x, y, and z axes are not limited to this example and may be changed depending on the location or shape of the building, or they may be left-handed. Furthermore, the x, y, and z axes may be the same at all nodes, or they may be different at each node. If the x, y, and z axes are different at each node, they may be determined, for example, based on a specific beam or column. A node is the endpoint of a column or beam that constitutes the frame, or a point where another column or beam is connected to such a column or beam.
[0027] In Figure 3, black circles represent nodes, solid lines indicate that a member is attached in that direction, and dashed lines indicate that a member is not attached in that direction. For directions where neither solid nor dashed lines are present, whether or not a member is attached in that direction is irrelevant to whether or not it corresponds to that type of joint.
[0028] As shown in Figure 3, type a0 is a type of joint where there are no member attachments in the positive direction of each of the x, y, and z axes, and in the negative direction of each of the x and y axes. Type a1 is a type of joint where there is one member attachment in only one of the four directions of the positive and negative x and y axes, and no member attachments in the remaining three directions and in the positive direction of the z axis. Type a2 is a type of joint where there are two members attachments in two directions that form a 180-degree angle with each other, and no member attachments in the remaining two directions and in the positive direction of the z axis. Type a3 is a type of joint where there are two members attachments in two directions that form a 90-degree angle with each other, and no member attachments in the remaining two directions and in the positive direction of the z axis.
[0029] Joint type a4 is a type of joint where three members are attached in three of the four directions of the positive and negative x and y axes, and there are no members attached in the remaining direction or in the positive direction of the z axis. Joint type a5 is a type of joint where four members are attached in all four directions of the positive and negative x and y axes, and there are no members attached in the positive direction of the z axis. Joint type a6 is a type of joint where, in addition to the case of joint type a0, there is also a member attached in the positive direction of the z axis.
[0030] Joint type a7 is a type of joint where, in addition to the case of joint type a1, there is an attachment of a member in the positive direction of the z axis. Joint type a8 is a type of joint where, in addition to the case of joint type a2, there is an attachment of a member in the positive direction of the z axis.
[0031] Joint type a9 is a type of joint where, in addition to the case of joint type a3, the member is attached in the positive direction of the z axis. Joint type a10 is a type of joint where, in addition to the case of joint type a4, the member is attached in the positive direction of the z axis. Joint type a11 is a type of joint where, in addition to the case of joint type a5, the member is attached in the positive direction of the z axis.
[0032] Returning to Figure 2, in step S5, the first member feature calculation unit 12 obtains the total beam member length at end i. The total beam member length at end i is a value that represents the sum of the member lengths of the beams joined at end i of that member. If the member is a beam, the length of that beam is also included. The first member feature calculation unit 12 obtains the lengths of all beams joined at end i from the frame information and calculates the total beam member length at end i by adding them together. The unit of length is, for example, meters, but other units may also be used.
[0033] In step S6, the first member feature calculation unit 12 obtains the total beam member length at end j. The total beam member length at end j is a value that represents the sum of the member lengths of the beams joined at end j of that member, and is calculated in the same manner as at end i. Figure 4 is a schematic diagram illustrating the total beam member lengths at end i and end j in this embodiment. Figure 4 shows only the beams in a part of the frame, and the columns are omitted. The member that is the subject of feature calculation is beam L3, shown by the thick line. At end i, beams L1 and L2 are joined to beam L3, so the total beam member length at end i is the length of beam L1 + the length of beam L2 + the length of beam L3. At end j, beams L4, L5, and L6 are joined to beam L3, so the total beam member length at end j is the length of beam L3 + the length of beam L4 + the length of beam L5 + the length of beam L6. The same considerations apply to column members.
[0034] Returning to Figure 2, in step S7, the first member feature calculation unit 12 obtains a value indicating the length of the member. The first member feature calculation unit 12 obtains the length of the member from the frame information. The unit of length is, for example, meters, but other units may be used.
[0035] Figure 5 is a graph illustrating the operation of the member cross-section estimation unit 16 in this embodiment. In the graph of Figure 5, the horizontal axis represents the first principal component calculated by the machine learning unit 13, and the vertical axis represents the second principal component calculated by the machine learning unit 13. Here, we will explain using the example where the machine learning unit 13 has calculated up to the second principal component for multiple features of each member. The rectangles in the graph of Figure 5 are plots of various building members used as a reference. The letters a through l attached to each rectangle indicate the cross-sectional shape of that member. Therefore, the rectangles in the graph of Figure 5 represent the results of machine learning performed by the machine learning unit 13.
[0036] The diamond shapes in the graph of Figure 5 represent the plots of each structural member of the target building at the positions of its principal components. The structural member cross-section estimation unit 16 calculates the principal components of the structural members of the target building by substituting the feature quantities calculated by the second structural member feature quantity calculation unit 15 into the calculation formula obtained by the principal component analysis performed by the machine learning unit 13. For each structural member of the target building, the structural member cross-section estimation unit 16 selects the structural member closest to the position of its principal component from among the structural members of the reference building, and uses the cross-sectional shape of the selected structural member as the estimated cross-sectional shape.
[0037] Therefore, in Figure 5, the rhombuses plotted near the rectangles labeled with the letter d, which represents the cross-sectional shape, are also labeled with the letter d, which represents the same cross-sectional shape. Similarly, the rhombuses plotted near the rectangles labeled with the letters e, f, g, h, j, and l, which represent the cross-sectional shapes, are also labeled with the letters e, f, g, h, j, and l, respectively, which represent the same cross-sectional shape.
[0038] Furthermore, the machine learning unit 13 may perform cluster analysis, and the member cross-section estimation unit 16 may estimate the cross-sectional shape based on which cross-sectional shape cluster the member belongs to. Alternatively, the machine learning unit 13 may generate a neural network that obtains the cross-sectional shape from the features, and the member cross-section estimation unit 16 may use that neural network to estimate the cross-sectional shape. In addition, a method using the normalized Euclidean distance between features may be used. In these cases as well, the same features as in this embodiment can be used.
[0039] As described above, the member cross-section estimation device 10 in this embodiment includes a first member feature calculation unit 12 that calculates multiple feature quantities for each of multiple members of a reference building, a second member feature calculation unit 15 that calculates multiple feature quantities for the members of the target building, a machine learning unit 13 that performs machine learning to estimate the cross-section of the target building member from the multiple feature quantities calculated by the second member feature calculation unit 15, using at least the multiple feature quantities calculated by the first member feature calculation unit 12, and a member cross-section estimation unit 16 that estimates the cross-sectional shape of the target building member from the multiple feature quantities for the target building member using the results of machine learning by the machine learning unit 13. Furthermore, each of the multiple feature quantities relating to each of the multiple reference building members and the multiple feature quantities relating to the target building member includes at least one of the following: a value indicating the floor on which the member is installed; a value indicating whether the member is a beam or a column; a value indicating the combination of the connection directions of the beams and columns joined at both ends of the member; a value indicating the sum of the member lengths of the beams joined at one end of the member; a value indicating the sum of the member lengths of the beams joined at the other end of the member; and a value indicating the length of the member. This makes it possible to easily estimate the cross-section of the building member under planning.
[0040] Alternatively, the member cross-section estimation device 10 may be realized by recording a program for realizing the functions of the member cross-section estimation device 10 in Figure 1 onto a computer-readable recording medium, loading the program recorded on this recording medium into a computer system, and executing it. The term "computer system" here includes hardware such as the operating system and peripheral devices.
[0041] Furthermore, "computer system" shall also include the homepage provisioning environment (or display environment) if a WWW system is being used. Furthermore, "computer-readable recording media" refers to portable media such as flexible disks, magneto-optical disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into computer systems. Moreover, "computer-readable recording media" also includes those that dynamically hold programs for a short period of time, such as communication lines used when transmitting programs over networks such as the Internet or communication lines such as telephone lines, and those that hold programs for a certain period of time, such as volatile memory inside computer systems that act as servers or clients in such cases. In addition, the above-mentioned programs may be for the purpose of realizing some of the functions described above, and may also be able to realize the above-mentioned functions in combination with programs already recorded in the computer system.
[0042] Furthermore, each functional block of the member cross-section estimation device 10 in Figure 1 described above may be individually chipped, or some or all of them may be integrated into a single chip. Also, the integrated circuit method is not limited to LSIs; it may be implemented using dedicated circuits or general-purpose processors. Both hybrid and monolithic designs are acceptable. Some functions may be implemented by hardware, while others may be implemented by software. Furthermore, with advancements in semiconductor technology, if technologies such as integrated circuit development that can replace LSIs emerge, it will also be possible to use integrated circuits based on those technologies.
[0043] While embodiments of this invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments and may include design modifications and the like that do not depart from the spirit of this invention. [Explanation of symbols]
[0044] 10 Member cross-section estimation device, 11 Reference building information input unit, 12 First member feature calculation unit, 13 Machine learning unit, 14 Target building information input unit, 15 Second member feature calculation unit, 16 Member cross-section estimation unit, 17 Member information output unit
Claims
1. A first member feature quantity calculation unit calculates multiple feature quantities for each of the multiple members of the reference building based on the structural information of the reference building, among the reference building information, A second member feature calculation unit calculates multiple feature quantities relating to the members of the target building based on the structural information of the target building, among the information of the target building. Without using data related to structural analysis results, a member cross-section estimation unit estimates the cross-sectional shape of a member of the target building from the cross-sectional shapes of multiple members of the reference building, using the frame-based feature quantities calculated by the first member feature quantity calculation unit and the second member feature quantity calculation unit. Equipped with, Each of the multiple feature quantities relating to each of the multiple members of the aforementioned reference building and each of the multiple feature quantities relating to the aforementioned building member includes a value indicating the floor on which the member is installed, a value indicating whether the member is a beam or a column, a value indicating the combination of the joining directions of the beams and columns joined at both ends of the member, a value indicating the sum of the member lengths of the beams joined at one end of the member, a value indicating the sum of the member lengths of the beams joined at the other end of the member, and a value indicating the length of the member. A device for estimating the cross-sectional area of a structural member.
2. The member cross-section estimation unit is, The machine learning unit performs machine learning to estimate the cross-section of the target building member from a plurality of feature quantities calculated by the second member feature quantity calculation unit, using at least the plurality of feature quantities calculated by the first member feature quantity calculation unit. The member cross-section estimation device according to claim 1, which estimates the cross-sectional shape of the target building member using the results of machine learning performed by the machine learning unit.
3. The machine learning unit performs principal component analysis on the multiple feature quantities calculated by the first member feature quantity calculation unit. The member cross-section estimation unit extracts, from among multiple reference building members, the one closest in distance to the principal component of the target building member in the space of the principal component of the analysis result by the machine learning unit, and sets the cross-sectional shape of the extracted member as the estimated cross-sectional shape. The member cross-section estimation device according to claim 2.
4. A method for estimating the cross-section of a member using a member cross-section estimation device, A first step is to calculate multiple feature quantities for each of the multiple members of the reference building based on the structural information of the reference building, among the information of the reference building to be referenced. A second step involves calculating multiple feature quantities relating to the members of the target building based on the structural information of the target building, among the information of the target building. Without using data related to structural analysis results, a third step is to estimate the cross-sectional shape of the member of the target building from the cross-sectional shapes of multiple members of the reference building, using the feature quantities based on the frame calculated in the first and second steps, and It has, Each of the multiple feature quantities relating to each of the multiple members of the aforementioned reference building and each of the multiple feature quantities relating to the aforementioned building member includes a value indicating the floor on which the member is installed, a value indicating whether the member is a beam or a column, a value indicating the combination of the joining directions of the beams and columns joined at both ends of the member, a value indicating the sum of the member lengths of the beams joined at one end of the member, a value indicating the sum of the member lengths of the beams joined at the other end of the member, and a value indicating the length of the member. Method for estimating the cross-section of a structural member.
5. Computers, A first member feature quantity calculation unit calculates multiple feature quantities for each of the multiple members of the reference building based on the structural information of the reference building, among the reference building information. A second member feature calculation unit calculates multiple feature quantities related to the members of the target building based on the structural information of the target building, among the information of the target building. Without using data related to structural analysis results, a member cross-section estimation unit estimates the cross-sectional shape of a member of the target building from the cross-sectional shapes of multiple members of the reference building, using feature quantities based on the frame calculated by the first member feature quantity calculation unit and the second member feature quantity calculation unit. It is a program designed to function as such. Each of the multiple feature quantities relating to each of the multiple members of the aforementioned reference building and each of the multiple feature quantities relating to the aforementioned building member includes a value indicating the floor on which the member is installed, a value indicating whether the member is a beam or a column, a value indicating the combination of the joining directions of the beams and columns joined at both ends of the member, a value indicating the sum of the member lengths of the beams joined at one end of the member, a value indicating the sum of the member lengths of the beams joined at the other end of the member, and a value indicating the length of the member. program.