KR102166490B1 - System and apparatus for cost estimation of architecture and construction - Google Patents

Abstract

Disclosed are a system for estimating construction and building costs to reflect a client′s opinion to a drawing and an apparatus thereof. According to the present invention, the system: acquires ground information of a target ground of a client who requests construction and building, wherein the ground information includes the area, slope, base ground quality, sunlight direction, and building purpose of the target ground; generates a required strength level of a requested building based on the ground information; generates required design and convenience levels based on a questionnaire evaluation of the client; acquires databased building objects, whose ground information satisfies the ground information of the target ground and a first required satisfaction level, among the databased building objects; selects building objects, whose strength, design, and convenience evaluation levels previously made for the acquired building objects satisfy the required strength, design, and convenience levels generated from the target ground and a second required satisfaction level, among the acquired building objects; outputs building information required for building a requested building on the target ground based on each selected building object, wherein the building information includes a building drawing, a building plan flowchart, construction equipment, and building materials corresponding to the databased building objects; and generates a building estimate of the requested building for the target ground based on the output building information.

Description

Construction and construction cost estimate system and device {SYSTEM AND APPARATUS FOR COST ESTIMATION OF ARCHITECTURE AND CONSTRUCTION}

The following examples relate to techniques for estimating construction and construction cost estimates.

In many cases, the process of estimating construction cost estimates for buildings and construction can be automated. On the other hand, the design process for estimating construction cost estimates for construction and construction is mostly done manually, making it difficult to say that it is actually completely automated. In addition, even if the design is automated, it is difficult to reflect the opinions of the client requesting architecture and construction, and it is difficult to include a more precise design reasoning process when inferring similar designs. Therefore, in the process of estimating construction cost estimates for construction and construction, it is required to properly reflect the opinions of the client, the process of design inference is precise, and an automated design and estimate inference method.

Korean Patent Application Publication KR10-2019-0122089 Korean Patent Application Publication No. KR10-1512293 Korean Patent Publication No. KR10-2039738 Korean Patent Application Publication No. KR10-1592762

In the embodiments, the opinion of the client is obtained through a questionnaire and is intended to be reflected in the design drawing and the estimate.

The embodiments are intended to be used as a reference for building design based on building objects databased through big data.

Examples are intended to design a building of the most suitable level from basic information about the soil.

Embodiments are intended to receive necessary data from online big data and refer to building design.

In the building and construction cost estimation system according to an embodiment, soil information of the target soil of the client who requested the building and construction work-the soil information is the area, slope, lower layer soil quality, sunlight direction and building of the target soil Obtaining-including use; Generating a robustness requirement of the requested building based on the soil information; The client's questionnaire evaluation-The questionnaire evaluation includes 40 to 100 questionnaire questions reflecting the evaluation of the requirements for the requested building, and each question is composed of an objective question and a subjective question. Generating a demand level and a convenience level; Acquiring the databaseized building objects from among the databased building objects, wherein the soil information of the databased building object satisfies the soil information of the target soil and a first requirement satisfaction level; Among the acquired building objects, a solidity evaluation degree, a design evaluation degree, and a convenience evaluation degree created in advance for the acquired building object, the robustness request degree, the design request degree, and the convenience request degree generated for the target soil. Selecting the obtained building objects satisfying the required satisfaction; Based on each selected building object, construction information required to construct the requested building on the target soil-the construction information is a construction plan corresponding to the databaseized building objects, a construction plan flow chart, construction equipment and construction materials Including-outputting; And generating a construction cost estimate of the requested building for the target soil, based on the output construction information.

According to an embodiment, the robustness evaluation degree of the databaseized building object is

Defect information reported in advance for the databaseized building object-The defect information is information provided by an evaluation of a building design expert and a construction safety expert for a building corresponding to the databased building object, for the building Based on the number, type, degree of defect management that occurred, and including the construction age of the building, it is made by a mechanical evaluation of the rigidity of the building, and the rigidity evaluation is expressed as a percentage value, and the rigidity evaluation is When determined, the design evaluation degree and the convenience evaluation degree are assigned to the remaining values of the percentage, and the design evaluation degree and the convenience evaluation degree are a random sample created online for the building corresponding to the database-formed building object The design evaluation degree and the convenience evaluation degree are obtained from the evaluation of the design and convenience of users, and a percentage value excluding the robustness evaluation degree, based on the proportional calculation of the design and convenience evaluation, according to the rational calculation. Discrete allocation, and the higher the design evaluation degree, the lower the convenience evaluation degree, and the higher the convenience evaluation degree, the lower the design evaluation degree. Calculated based on a review written on the building, and generating the solidity requirement of the target soil comprises: generating the solidity demand as a percentage value through mechanical evaluation based on the soil information of the target soil Including, the step of generating the design requirements and the convenience requirements of the target soil, separating the multiple-choice question and the subjective question of the questionnaire evaluation; Performing preprocessing of the multiple-choice question and the subjective question; Obtaining a result through mechanical evaluation of the pre-processed objective question and the subjective question through corresponding artificial neural networks, respectively; Generating a ratio of the design requirement and the convenience requirement based on the result of the mechanical evaluation; And generating the design requirement and the convenience requirement based on a ratio of the design requirement and the convenience requirement to a percentage value excluding the robustness requirement.

According to an embodiment, the pre-processing of the multiple-choice question and the subjective question may include: generating a first questionnaire vector so as to correspond to each row of the multiple-choice question; The step of separating the subjective questions into a predetermined basic unit-the basic unit follows the grammatical word unit division, but in the case of adverbs and contemplatives, it is indicated with the following adverbs, idioms, interjections, adjectives and verbs. ; Removing irradiation from the basic unit; Selecting a first meaning word from the basic unit from which the irradiation has been removed; Selecting a second meaning word based on a mechanical evaluation of the basic unit excluding the first meaning word; And generating a second questionnaire vector so that the first meaning word and the second meaning word correspond to each row for each question.

According to an embodiment, the robustness evaluation degree of the databaseized building object is

Generating a first evaluation vector based on the defect information previously reported for the databased building object; Applying the first evaluation vector to a first evaluation artificial neural network to obtain a first evaluation output; And generating the robustness evaluation degree based on the first evaluation output, wherein the first evaluation artificial neural network is generated based on defect information for 10000 buildings acquired as big data through online. First training evaluation vectors; And first training evaluation outputs obtained by applying the first training evaluation vectors to the first evaluation artificial neural network based on first training evaluation labels obtained in advance to correspond to the first training evaluation vectors, and the Learning in a backpropagation method through the difference between the first training evaluation labels, and obtaining an evaluation of the design and convenience of the random samples created for the building objects online for the design evaluation and the convenience evaluation ; Generating a second evaluation vector based on the evaluation of the design and convenience; Applying the second evaluation vector to a second evaluation artificial neural network to obtain a second evaluation output; And generating a proportional calculation of the evaluation of the design and convenience based on the second evaluation output, wherein the second evaluation artificial neural network includes 1000 randomly generated building objects and the corresponding Second training evaluation vectors generated based on 100 reviews randomly written for each random building object; And a second training evaluation vector obtained by applying the second training evaluation vectors to the second evaluation artificial neural network based on second training evaluation labels previously created by the second artificial neural network designer to correspond to the second training evaluation vectors. 2 learning in a backpropagation method based on differences between training evaluation outputs and the second training evaluation labels, and generating a first request vector based on the soil information of the target soil, wherein the robustness requirement is based on the soil information; Applying the first request vector to a first request artificial neural network to obtain a first request output; And generating the robustness requirement of the requested building corresponding to the target soil based on the first request output, wherein the first request artificial neural network is generated based on randomly generated 1000 soil information First training request vectors; And obtained by applying the first training request vectors to the first request artificial neural network based on first training request labels previously created by the first request artificial neural network designer to correspond to the first training request vectors. It is possible to learn in a backpropagation method through differences between the first training request outputs and the first training request labels.

According to an embodiment, the step of selecting the second meaning word from the basic unit excluding the first meaning word obtained from the subjective question is based on the basic unit excluding the first meaning word, and the subjective Generating a first subjective vector to include the original sentence of the type item; Inputting the first subjective vector into a first subjective artificial neural network to obtain a first subjective output; And selecting the second meaning word from the basic unit excluding the first meaning word based on the first subjective output, wherein the first subjective artificial neural network randomly generates 100000 subjective answers First training subject vectors generated by excluding words corresponding to the first meaning word in; And based on first training subjective labels previously created by the first subjective artificial neural network designer to correspond to the first training subjective vectors, obtained by applying the first training subjective vectors to the first subjective artificial neural network. The step of learning in a backpropagation method through the difference between the first training subject outputs and the first training subject labels, and generating a ratio of the design request and the convenience request may include the first questionnaire vector and the second questionnaire vector Inputting into a first questionnaire artificial neural network and a second questionnaire artificial neural network corresponding to each; Generating a second request vector based on the outputs of the first questionnaire artificial neural network and the second questionnaire artificial neural network; Inputting the second request vector to a second request artificial neural network to obtain a second request output; And generating a ratio of the design request and the convenience request based on the second request output, wherein the first questionnaire artificial neural network and the second questionnaire artificial neural network are randomly generated 100000 objective answers. And first training questionnaire vectors and second training questionnaire vectors respectively generated based on first semantic words and second semantic words of the 100000 subjective responses used in the first subjective artificial neural network. And first training questionnaire labels and second training questionnaire labels previously created by designers of the first questionnaire artificial neural network and the second questionnaire artificial neural network to correspond to the first training questionnaire vectors and the second training questionnaire vectors. Based on the first training questionnaire outputs and second training questionnaire outputs obtained by applying the first training questionnaire vectors and the second training questionnaire vectors to the first questionnaire artificial neural network and the second questionnaire artificial neural network, respectively And the difference between the first training questionnaire labels and the second training questionnaire labels, and the second required artificial neural network is generated from the first training questionnaire outputs and the second training questionnaire outputs. Second training request vectors; And obtained by applying the second training request vectors to the second request artificial neural network based on second training request labels previously created by the second request artificial neural network designer to correspond to the second training request vectors. The learning may be performed in a backpropagation method through differences between the second training request outputs and the second training request labels.

In the embodiments, the opinion of the client may be acquired through a questionnaire and reflected in the design drawing and the estimate.

Embodiments may be used as a reference for building design based on building objects databased through big data.

The embodiments can design a building of the most suitable level from basic information about the soil.

Embodiments may receive necessary data from online big data and refer to building design.

1 is a flowchart illustrating a system for estimating construction and construction costs according to an embodiment.
FIG. 2 is a diagram for explaining a method of obtaining a robustness evaluation degree, a design evaluation degree, and a convenience evaluation degree according to an exemplary embodiment.
3 is a diagram for explaining a method of generating a robustness request, a design request, and a convenience request according to an embodiment.
4 is a diagram illustrating a method of processing a questionnaire question according to an exemplary embodiment.
5 is a diagram for describing an artificial neural network used to obtain a robustness evaluation degree, a design evaluation degree, and a convenience evaluation degree according to an embodiment.
6 is a diagram for explaining an artificial neural network used to generate a robustness requirement, a design requirement, and a convenience requirement according to an embodiment.
7 is an exemplary diagram of a configuration of an apparatus according to an embodiment

Hereinafter, exemplary embodiments will be described in detail with reference to the accompanying drawings. However, since various changes may be made to the embodiments, the scope of the rights of the patent application is not limited or limited by these embodiments. It should be understood that all changes, equivalents, or substitutes to the embodiments are included in the scope of the rights.

Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only, and may be changed in various forms and implemented. Accordingly, the embodiments are not limited to a specific disclosure form, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical idea.

Although terms such as first or second may be used to describe various components, these terms should be interpreted only for the purpose of distinguishing one component from other components. For example, a first component may be referred to as a second component, and similarly, a second component may be referred to as a first component.

When a component is referred to as being "connected" to another component, it is to be understood that it may be directly connected or connected to the other component, but other components may exist in the middle.

The terms used in the examples are used for illustrative purposes only and should not be interpreted as limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In the present specification, terms such as "comprise" or "have" are intended to designate the presence of features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, but one or more other features. It is to be understood that the presence or addition of elements or numbers, steps, actions, components, parts, or combinations thereof, does not preclude in advance.

Spatially relative terms "below", "beneath", "lower", "above", "upper", etc., as shown in the figure It can be used to easily describe the correlation between a component and other components. Spatially relative terms should be understood as terms including different directions of components during use or operation in addition to the directions shown in the drawings. For example, if a component shown in a drawing is turned over, a component described as "below" or "beneath" of another component will be placed "above" the other component. I can. Accordingly, the exemplary term “below” may include both directions below and above. Components may be oriented in other directions, and thus spatially relative terms may be interpreted according to orientation.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiment belongs. Terms as defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and should not be interpreted as an ideal or excessively formal meaning unless explicitly defined in this application. Does not.

In addition, in the description with reference to the accompanying drawings, the same reference numerals are assigned to the same components regardless of the reference numerals, and redundant descriptions thereof will be omitted. In describing the embodiments, when it is determined that a detailed description of related known technologies may unnecessarily obscure the subject matter of the embodiments, the detailed description thereof will be omitted.

1 is a flowchart illustrating a system for estimating construction and construction costs according to an embodiment.

According to an embodiment, a control device (hereinafter, referred to as a control device) for a system for estimating construction and construction costs may acquire soil information of a target soil of a client who requested a building and construction work (101).

The control device according to an embodiment may include a server and a user terminal. The server and the user terminal are connected to a network and can exchange data through wired or wireless communication. The control device may be used by architectural designers and construction designers to search for a building construction design of soil requested by a customer, and may also be used to search for a construction design of a building on soil owned by architectural designers and construction designers.

The server may be its own server owned by a person or organization that provides services using the server; May be a cloud server; It may be a peer-to-peer (p2p) set of distributed nodes. The server has a computing function of a typical computer; Save/reference function; Input/output function; And it may be configured to perform all or part of the control function. The server may include at least one artificial neural network that performs an inference function. The server may be configured to communicate with the user terminal via wired or wireless.

The server communicates with the user terminal and can perform operations to find a building design optimized for the target soil. Accordingly, the service provided through the server may be suitable for a person or organization that specializes in building and construction, such as architectural designers and construction designers.

The user terminal may be a desktop computer, a laptop computer, a tablet, a smartphone, or the like. The user terminal includes an operation function of a general computer; Save/reference function; Input/output function; And it may be configured to perform all or part of the control function. The user terminal may be configured to communicate with the server by wire or wirelessly.

The user terminal may install an application linked with the server or may be connected to a web page linked with the server. The user terminal may transmit soil information of a target soil for designing a building to the server through a web page or an application, and obtain a building design method and estimate of the target soil calculated and generated by the server. Through this, the user of the user terminal can present the building design and estimate of the target soil to the customer. Users of the user terminal may be mainly architectural designers and construction designers, but there are no special restrictions on the user's job. As long as the processing capacity of the server allows, there is no particular limitation on the number of user terminals.

Hereinafter, embodiments will be described centering on the operation of the server, but the embodiments are not limited by the subject or aspect of communication, and various application examples may be employed.

According to an embodiment, the server of the control device may obtain soil information of the target soil from the user terminal. The soil information of the target soil may include the area of the target soil, the slope, the soil quality of the lower layer, the direction of sunlight, and the use of the building. The area of the target soil may generally be expressed in units of m ² , but is not limited thereto. Depending on the type of the target soil, the method of calculating the area may vary, and in general, the area measured by the method specified by the real estate-related legislation of the country or region to which the target soil belongs may be used. The slope of the target soil can be measured based on the horizontal plane, and the steeper the slope is, the more unsuitable land for building design is, so higher robustness may be required when designing a building. The soil quality of the lower layer of the target soil can be expressed as a ratio of the density, erosion and fertility of the soil obtained from constituents constituting the lower layer of the target soil. The direction of sunlight on the target soil can be used as an indicator to set the direction in building and construction design, and can be used to change the design according to the surrounding environment. In particular, when the transmission of sunlight is obstructed according to the surrounding topographical features, the architectural and construction design taking into account all calculations can be made. The use of the building is the purpose reported to the target soil, and the value obtained in advance can be used. When a client wants to use the construction and construction cost estimation system, when he/she wants to change the use of the soil, he/she can specify the changed value for the purpose of the target soil by registering the application for the use change in the construction and construction cost estimation system.

According to an embodiment, the user of the user terminal may be a construction designer and a construction designer of the target soil, and may plan a building design and estimate of the target soil for a requester of a request building for the target soil. In this process, the user terminal may acquire soil information of the target soil through the customer's terminal (PC, notebook, smartphone, etc.) or through the user's input.

According to an embodiment, the control device may generate a robustness requirement of the requested building based on soil information (102).

According to an embodiment, the control device may input soil information into an artificial neural network to generate a robustness requirement of the requested building through mechanical evaluation. The soil information can be used to generate the first request vector corresponding to the input layer of the first request neural network, and the control device can generate the robustness requirement of the request building from the first request output corresponding to the output layer of the first request neural network. have.

According to an embodiment, the control device may generate the robustness requirement of the requested building from the area, slope, lower layer soil quality, sunlight direction, and building use of the target soil included in the soil information, each element individually, Or it can be evaluated comprehensively. For example, even if the slope is steep, if the soil quality is sufficiently strong and the area is large, a higher building can be designed.Conversely, even if the slope is flat, a relatively low building is designed if the soil quality is weak and the area is narrow. can do.

According to an embodiment, the control device may generate a design request and a convenience request based on the customer's questionnaire evaluation (103).

The design demand obtained from the client's questionnaire evaluation may be a weight allocated according to the degree to which aesthetics are satisfied to users using the requested building of the target soil. The convenience demand obtained from the survey evaluation may be a weight allocated so as to increase as the degree of convenience that users feel while using it increases.

According to an embodiment, the design demand and the convenience demand of the requested building may be first expressed as a ratio value of 100. Accordingly, as the design demand increases, the convenience demand may be relatively lower. For example, if the design requirement is 30%, the convenience requirement can be 70%; Conversely, if the design requirement is 70%, the convenience requirement may be 30%. The final design demand and convenience demand of the requested building can be determined according to the% value of the pre-generated robustness demand.If the robustness demand is determined in advance as 30%, the sum of the design demand and the convenience demand will be adjusted to be 70%. I can. In this adjustment process, the ratio of the existing design requirement and convenience requirement can be used, and can be calculated through the process of converting the ratio calculated as 100% to 70%. The specific operation of generating the robustness requirement, design requirement, and convenience requirement of the requested building will be described later with reference to FIG. 3.

According to an embodiment, the control device may obtain, among the databased building objects, the databased building objects in which the soil information of the databased building object satisfies the soil information of the target soil and the first requirement satisfaction level ( 104).

According to an embodiment, the building object may be an object including information on buildings previously built on a corresponding soil. The information of buildings includes the name of the building, the purpose of the building, the location of the building, the direction of the building, the size of the building (width, height, height, etc.), and the structure within the building (elevator, stairs, number of separated spaces, and columns). Number, etc.), building construction information (construction plans, construction plan flow charts, construction equipment and construction materials, etc.), estimates corresponding to construction information, soil information of the soil including buildings, solidity evaluation drawings prepared in advance for buildings, design evaluation Information such as degree and convenience evaluation degree may be included. The control device may pre-database and store each building object corresponding to each soil on which the construction of the building is completed.

According to an embodiment, the first request satisfaction level may be a criterion for determining a first comparison value of each databaseized building object. The first comparison value may be a value obtained by comparing a difference between soil information of the soil including the databased building object and soil information of the target soil.

According to an embodiment, the control device may generate a first comparison value of each databaseized building object based on soil information including each databaseized building object and soil information of a target soil. Subsequently, the control device may determine whether the first comparison value of each database-formed building object satisfies the first requirement satisfaction level. Subsequently, the control device may obtain databaseized building objects that satisfy the first requirement satisfaction level.

According to an embodiment, if the first comparison value is less than 100, for example, the control device may determine that the first request satisfaction level is satisfied. In order to generate the first comparison value, the control device may compare the horizontal length of the target soil and calculate a percentage (%) difference in the horizontal length of the soil information of the soil including each databased building object. ; Compared with the vertical length of the target soil, it is possible to calculate what percentage (%) the vertical length of the soil information of each databased building object is different; Compared with the slope of the target soil, it is possible to calculate what percentage (%) the slope of the soil soil information including each databased building object is different; Compared with the soil quality of the lower layer of the target soil, it is possible to calculate what percentage (%) the difference of the soil quality of the lower layer of the soil information of each databased building object is; Compared with the sunlight direction of the target soil, it is possible to calculate what percentage (%) the angle difference of the sunlight direction of the soil of the building object shows; Compared with the building use of the target soil, it is possible to calculate what percentage (%) difference in the building use of the building object. The difference in percent (%) for building uses can be based on pre-specified values. In general, buildings are used as detached houses, apartment houses, type 1 neighborhood living facilities, type 2 neighborhood living facilities, cultural and assembly facilities, religious facilities, sales facilities, transportation facilities, medical facilities, educational research facilities, elderly people's facilities, and training. Facilities, sports facilities, business facilities, lodging facilities, amusement facilities, factories, warehouse facilities, dangerous goods storage and treatment facilities, automobile related facilities, animal and plant related facilities, resource circulation related facilities, correctional and military facilities, broadcasting and communication facilities, power generation It can be classified into 29 types of facilities, cemetery related facilities, tourist rest facilities, funeral halls, and campsite facilities, and you can determine the percentage (%) of the difference in use of the building based on the difference by obtaining a number that is appropriate for each order. The control device is the difference in percent (%) of the transverse length; Difference in percent (%) of vertical length; Difference in percent (%) of the slope; Difference in percent (%) of lower soil quality; Difference in percent (%) of the angle in the direction of sunlight; A first comparison value may be generated by summing the difference in percent (%) of the building use.

According to an embodiment, for example, the control device calculates that the width of the soil information of the first databased building object is 11 m, and a difference of 10 percent (%) is compared with 10 m of the width of the target soil. Can; Compared with the length of the target soil of 15 m, it is possible to calculate that the length of the soil information of the first databased building object is 12 m, which is a 20 percent (%) difference; Compared with the slope of the target soil of 15°, it is possible to calculate that the slope of the soil information of the first databased building object is 15°, and that there is a difference of 0 percent (%); The soil quality of the lower layer of the target soil is dense: erosion: the soil quality of the soil information of the building object in the first database is 60:10:30 compared to that of 70:10:20, which is 10% (%) difference Can compute what is going on; Compared with the sunlight direction of the target soil 15°, it is possible to calculate that the sunlight direction of the soil information of the first databased building object is 20°, which is 33 percent (%) different; It can be calculated that the building use of the target soil is a difference of 3.4 percent (%) as the building use of the building object in the first database compared to the single-family house (1 point), with the apartment house (2 points). The control device is 10, which is the difference in percent (%) of the horizontal length; 20, the difference in percent of the vertical length; 0, the difference in percent (%) of the slope; 10, which is the difference in percent (%) of lower soil quality; 33, which is the difference in percent (%) in the direction of sunlight; The first comparison value of 76.4 can be generated by summing the difference of 3.4, which is the percentage (%) of the building use.

According to an embodiment, for example, since 76.4, which is a first comparison value of the building object converted to the first database, is a value of 100 or less, the control device determines that the building object converted to the first database satisfies the first demand satisfaction level. , It is possible to obtain a building object converted to a first database.

According to an embodiment, the control device may determine whether the n-th database-formed building object satisfies the first request satisfaction level, using the method described above, for the remaining database-formed building objects, and the first request satisfaction level Each databaseized building object that satisfies can be obtained.

According to an embodiment, the control device includes a solidity evaluation degree, a design evaluation degree, and a convenience evaluation degree created in advance for the obtained building object among the acquired building objects, and Acquired building objects that satisfy the convenience requirement and the second requirement may be selected (105).

According to an embodiment, a degree of robustness evaluation, a design evaluation degree, and a convenience evaluation degree of building objects in a database may be generated based on inference of a learned artificial intelligence (artificial neural network). For example, the degree of robustness evaluation of each databased building object may decrease as the number of defects occurs more frequently, more frequently, based on defect information including whether or not defects occurred in the corresponding building. The design evaluation degree may increase as the number of reviews determined to include the mention that the design of the building is aesthetic in the reviews collected by the control device for the building corresponding to each databased building object. The convenience evaluation map of each databased building object is, in the reviews collected by the control device for the building corresponding to each databased building object, the more reviews determined to include a mention that the spatial arrangement design is convenient, the more Can increase. A detailed operation of the control device inferring and generating a solidity evaluation degree, a design evaluation degree, and a convenience evaluation degree of a database-formed building object will be described later with reference to FIG. 2.

The second request satisfaction according to an embodiment may be a criterion for determining a second comparison value of each referenced building object. The control device includes a solidity evaluation diagram, a design evaluation diagram, and a convenience evaluation diagram of each referenced building object; A second comparison value of each acquired building object may be generated based on the robustness requirement, design requirement, and convenience requirement of the target soil. Subsequently, the control device may determine whether the second comparison value of each acquired building object satisfies the second requirement satisfaction level. Subsequently, the control device may select the acquired building objects that satisfy the second requirement satisfaction level.

According to an embodiment, if the second comparison value is less than 15, for example, the control device may determine that the second request satisfaction level is satisfied. In order to generate the second comparison value, the control device may calculate a ratio of the robustness requirement, the design requirement, and the convenience requirement as a percentage (%), based on soil information and questionnaire evaluation on the target soil. Subsequently, the control device determines the difference between the ratio of the solidity evaluation, design evaluation, and convenience evaluation of each referenced building object and the solidity demand, design and convenience demand of the target soil. It can be set as the second comparison value. If the second comparison value is, for example, less than 5, the control device may determine that the range of values assigned to the second similarity is satisfied.

For example, the control device may have generated 40% of the robustness requirement of the target soil, 30% of the design requirement, and 30% of the convenience requirement. On the other hand, the first acquired building object may have a solidity evaluation of 50%, a design evaluation of 20%, and a convenience evaluation of 30%. The control device may set 10, which is a difference between the robustness demand ratio of the first acquired building object, and the robustness evaluation ratio, as the second comparison value. In the calculation, firstly, the difference between the robustness demand ratio and the robustness evaluation ratio can be used, and then the design demand ratio and the design evaluation ratio, or the convenience demand. If the difference between the ratio and the convenience evaluation ratio is different from the difference between the robustness demand ratio and the robustness evaluation ratio, the difference can be used. Since the second comparison value of the referenced first building object is less than 15, the control device determines that the referenced first building object satisfies the second requirement, and selects the referenced first building object.

According to an embodiment, the control device may determine whether the n-th acquired building object satisfies the second requirement satisfaction level, using the method described above, for the remaining acquired building objects, and satisfies the second requirement satisfaction level. Each acquired building object can be selected.

According to an embodiment, the control device may output construction information necessary for constructing a requested building in the target soil based on each selected building object (106).

According to an embodiment, the control device may extract construction information of each selected building object. Also, the control device may generate a blueprint model of a requested building corresponding to the target soil based on soil information of the target soil, or obtain a blueprint model of the target soil through a user terminal or the like. Subsequently, the control device may output each building design plan for the target soil by combining the extracted construction information of each building object with the design drawing model of the requested building of the target soil.

For example, the control device may select three building objects from among the acquired building objects for building design of the target soil. The control device may extract construction information from each selected building object. In addition, the control device may generate a request building blueprint model of the target soil based on soil information of the target soil, or obtain a blueprint model of the target soil through a user terminal. Subsequently, the control device may infer a result of combining each of the extracted building construction information to the requested building design model of the target soil using an artificial neural network, and correct and output three construction information for the target soil.

According to an embodiment, the control device may generate a construction cost estimate of the requested building for the target soil based on the output construction information (107).

The control device can calculate the cost required for the construction of the requested building for each output. Specifically, the control device, based on the existing building construction history information, in order to complete the construction according to each output, the cost of purchasing basic building materials; The cost of purchasing, renting and installing elevators, emergency power supplies, emergency water supply and frequency generators, etc.; Other design costs; The cost of installing various facilities such as water, electricity and gas; Other overall construction equipment rental cost and manpower cost required for completion of the construction are calculated, and these costs are summed to calculate the cost for each building and construction work.

The control device may transmit each output building design and each calculated cost to the user terminal. Through this, the user terminal can display each building design and an estimated cost required for each building design.

Through the above, the control device may perform an operation of generating a building design method in consideration of the robustness, aesthetics, and convenience of the target soil based on the building objects in the database. Through this, a user of a user terminal using a service performed by a control device can obtain a building design method in consideration of robustness, aesthetics, and convenience. Through this, it is possible to help architectural designers and construction designers to more easily proceed with building design.

FIG. 2 is a diagram for explaining a method of obtaining a robustness evaluation degree, a design evaluation degree, and a convenience evaluation degree according to an exemplary embodiment.

According to an embodiment, each of the database-formed building objects may include numerical data such as a robustness evaluation 203, a design evaluation 206, and a convenience evaluation 207. The degree of robustness evaluation 203 of the databased building object may increase as the number of defect management decreases, based on the defect information 202 collected for the building corresponding to the databased building object. Defect information 202 is information provided by an evaluation of a building design expert and a construction safety expert for a building corresponding to a building object in a database, and the number, type, degree, and construction age of the building It may include. The number of defect management can be used as a measure for dividing the severity of defects according to the number, and the type and degree of defects can be mechanically classified according to a predetermined value. The building age of a building can be based on the value determined at the time of initial design, and can also be used as a value to evaluate the building by comparing the occurrence of the number, type, and degree of defect management against the building age.

The design evaluation degree 206 and the convenience evaluation degree 207 of the databased building object may be calculated based on the design and convenience evaluation 204 in a review made by random samples online. The design evaluation map 206 may increase as the number of reviews determined to include a mention that the building design is aesthetic in a review written by random samples online. Convenience evaluation map 207 of databased building objects is determined to include a comment that the building design is convenient from reviews collected on buildings corresponding to databased building objects in reviews written by random samples online. The more reviews there are, the more it can be. The robustness evaluation degree 203, the design evaluation degree 206, and the convenience evaluation degree 207 may have a sum of 100%. When the robustness evaluation degree 203 is first determined, the remaining percentage (%) value A method may be selected according to the ratio of the design evaluation degree 206 and the convenience evaluation degree 207. Accordingly, the higher the design evaluation 206, the lower the convenience evaluation 207, and the higher the convenience evaluation 207, the lower the design evaluation 206 may be.

According to an embodiment, the control device may collect a review on a building corresponding to the databased building object in order to obtain a design evaluation 206 and a convenience evaluation 207 of the databased building object. Reviews may be collected through web pages or applications that provide social media, blogs, map services, and geospatial information services. The collection of reviews may be performed by a keyword extraction method, a hashtag extraction method, a text mining method, or the like.

According to an embodiment, the control device may determine whether a statement that a building design is aesthetic or a statement that a building design is convenient is included in the collected reviews. The determination may be made by a keyword extraction method, a hashtag extraction method, a text mining method, or the like. The control device can receive necessary data from online big data and refer to it for building design.

In the reviews collected on the building corresponding to the building object in the database, the more reviews determined that the control device includes a mention that the building design is aesthetic, the more the design evaluation of the building design of the building object in the database (206) Can be set to a higher value. For example, the more review articles determined that the control device includes mentions of a beautiful exterior, a modern feel, a foreign home-like feel, etc., in a review collected on a building corresponding to a database-based building object, It is possible to increase the design evaluation degree 206 of the building design of the databased building object.

In addition, in the reviews collected on the space corresponding to the building object in the database, the more reviews determined to include the mention that the building design is convenient, the more the control device evaluates the convenience of the building design of the building object in the database ( 207) can be set to a higher value. For example, in the reviews collected on the buildings corresponding to the building objects in the database, the review determined that the control device includes mentions that it is convenient because there are many elevators, it is good because there is no noise between floors, or it is convenient because there are many toilets. As the number increases, the degree of convenience evaluation 207 of the building design of the databased building object can be increased.

The specific values of the robustness evaluation diagram 203, design evaluation diagram 206, and convenience evaluation diagram 207 of each database-formed building object will be generated based on the inference of the learned artificial intelligence (artificial neural network). I can. The learning operation of artificial intelligence will be described later with reference to FIG. 5.

For example, in obtaining the robustness evaluation 203, the design evaluation 206 and the convenience evaluation 207 of the building design of the building object corresponding to the building-1, the control device -1 reviews can be collected. Subsequently, the control unit can classify the reviews with reference to the building design from the reviews on the building-1. Then, from the classified reviews, the control device can determine the reviews with positive comments on the building design.

The control unit determined that in the review article on Building-1, there were 4 review articles including the mention that the building design is aesthetic and 6 review articles including the mention that the building design is convenient. The ratio 205 of the design evaluation degree 206 and the convenience evaluation degree 207 of the building design may be set to 4:6. The specific values of the design evaluation degree 206 and the convenience evaluation degree 207 and ratio 205 of the building design of the building-1 object may be generated based on the inference of the learned artificial intelligence (artificial neural network) of the control device. The ratio of the generated design evaluation degree 206 and the convenience evaluation degree 207 may be finally used for calculation of the rest except for the percent (%) value occupied by the robustness evaluation degree 203. For example, if the robustness evaluation 203 determined based on the defect information 202 is 50%, the design evaluation 206 and the convenience evaluation 207, the ratio 205, of 4:6 are used. The design evaluation degree 206 may be calculated as 20%, and the convenience evaluation degree 207 may be calculated as 30%.

According to an embodiment, the control device may obtain a robustness evaluation degree 203, a design evaluation degree 206, and a convenience evaluation degree 207 of the building design of the databaseized building objects 201. The robustness evaluation degree 203 refers to the degree of robustness of a building, and is based on the fact that no defects occur for as long as possible. The design evaluation diagram 206 seeks for a client and an actual user to feel the beauty of the building for the requested building. Therefore, it may be good to reflect the client's design requirements as much as possible. On the other hand, the convenience evaluation map 207 of the building design is convenient in the use of the requested building, and can be interpreted as a numerical value of how much the installed structures increase the convenience. The robustness evaluation diagram 203, design evaluation diagram 206, and convenience evaluation diagram 207 of the building design of the databased building objects 201 may be used to create a building design plan suitable for the target soil.

The control device according to an embodiment may be used as a reference for building design based on building objects databased through big data.

3 is a diagram for explaining a method of generating a robustness request, a design request, and a convenience request according to an embodiment.

According to an embodiment, the operation of the control device generating the robustness demand 302, the design demand 306, and the convenience demand 307 of the target soil may include the following detailed operations.

According to an embodiment, the control device may generate the robustness requirement 302 based on the soil information 301 of the target soil. The robustness requirement 302 may include whether the target soil is suitable for erecting a building, and may include information on the rigidity required for the building to be erected. The robustness requirement 302 may be evaluated based on the area, slope, soil quality of the lower layer, the direction of sunlight, and the use of the building, which are included in the soil information 301 of the target soil, and each element is individually, Or it can be obtained by comprehensively evaluating. The generation of the robustness requirement 302 can be obtained by mechanical evaluation through artificial intelligence (artificial neural network). The control unit can design the most suitable level of building from basic information about the soil.

According to an embodiment, the control device may generate a design request 306 and a convenience request 307 of the target soil, each of which may be generated by values of the questionnaire evaluation 303 written by the client. The questionnaire evaluation 303 prepared by the client may consist of a total of 40 to 100 questions, and the questions of the questionnaire evaluation 303 may be divided into an multiple-choice question 304 and a subjective question 305. In the multiple-choice question 304, when a corresponding answer for each question is determined, it may be reflected in the design request level 306 and the convenience request level 307 through values set for the corresponding answer in advance. The subjective question 305 may undergo mechanical evaluation through a process of pre-processing the answer to each question. A detailed description of the pre-processing process of the subjective question 305 will be described later with reference to FIG. 4.

According to an embodiment, the control device may perform mechanical evaluation of the objective question 304 and the subjective question 305 through corresponding artificial neural networks, respectively. Based on the result of this evaluation, the design requirement ( 306) and the convenience requirement 307 may be generated.

According to an embodiment, after the percentage (%) of the robustness requirement 302 is first determined, the design requirement 306 and the convenience requirement 307 are excluded from the percent (%) of the robustness requirement 302. It can be calculated from the remaining percentage, in which case the ratio of the design requirement 306 and the convenience requirement 307 to the remaining percentage can be used. For example, if the robustness requirement 302 is 40%, and the ratio of the design requirement 306 and the convenience demand 307 is 5:5, the final design requirement 306 is 30%, and the convenience demand Figure 307 may be determined as 30%. A detailed description of the artificial neural network used to calculate the robustness requirement 302, the design requirement 306, and the convenience requirement 307 will be described later with reference to FIG. 6.

The control device according to an embodiment may acquire a client's opinion through a questionnaire and reflect it in a design drawing and an estimate.

4 is a diagram illustrating a method of processing a questionnaire question according to an exemplary embodiment.

According to an embodiment, the control device may self-process the sentences acquired through the subjective question of the questionnaire question. The control device may classify the acquired sentences into a basic unit 401. The basic unit 401 is basically classified based on a grammatical word unit, that is, a space, but if the word is an adverb or a tube sentence according to the search result of the database in the control device stored in advance, the following adverb, tube sentence, interjection, and adjective It can be distinguished by notation with and verbs. For example,'beautiful house' is divided into'beautiful' and'house' in the existing word unit, but'beautiful house' can be viewed as a unit because'beautiful' is a ductal sentence previously stored in the database. For example, in the case of the expression'very beautiful house' or'very very inconvenient staircase','very' is an adverb, so'beautiful' is attached with the guan detective, and'beautiful' is also a guan detective, so'very beautiful house 'Can be viewed as a unit. In the same way,'very very uncomfortable stairs' is also attached with'very,' because'too much' is an adverb. 'Too very inconvenient stairs' may be recognized as one word in the basic unit 401.

According to an embodiment, the control device may remove the investigation of words classified by mood units. The type of investigation may include'silver, silver, two, a, a', etc., but is not limited thereto. The type of survey may be previously stored in the database. The control device may select the first meaning word 402 among words of the basic unit 401 from which the irradiation has been removed.

The first meaning word 402 according to an exemplary embodiment may mean a word having a clear meaning by itself. An obvious word may be a word that can infer the meaning of design and convenience by itself. For example,'North Europe','Southbound','Convenient','Modern','Aesthetic','Convenient','Pretty', etc. may be included, but are not limited thereto.

According to an embodiment, the control device may select the second meaning word 403 by performing a mechanical evaluation on the basic unit 401 excluding the first meaning word 402. In the mechanical evaluation of the basic unit 401 except for the first meaning word 402, the semantic unit can be made clear by including all the existing subjective question answer contents.

The second meaning word 403 according to an exemplary embodiment may mean a word whose meaning is not clear by itself and thus requires determination of the context. For example, expressions such as'like' and'not so much' may be included, but are not limited thereto. For example, in the case of sentences such as'the sun is rising' and'the design is not so good', it can be confirmed that'heard' and'not so much' are difficult to infer only with those words. In this case, it is necessary to check the classification criteria words included before and after the word in the sentence. In the case of'heard', the word'year' may be the classification criterion word, and in the case of'not so much', the word'design' may be the classification criterion word. Each classification criterion word preferentially selects a classification criterion word located before based on the corresponding word, but if not, a classification criterion word that is closest to the back of the corresponding word may be selected as the classification criterion word of the corresponding word. The classification criterion word may be selected based on a value previously stored in the database, and may be modified by an administrator who manages the database.

According to an embodiment, the control device may generate the second questionnaire vector so that the first meaning word 402 and the second meaning word 403 correspond to each row for each question. The second questionnaire vector according to an embodiment may generate a vector to correspond to each item number of the subjective question. For example, the answer to No. 1 of the subjective question is'I wish it would be a Nordic style', and subjective If the answer to #2 of the hyung question is'I wish it would be designed so that the sun goes well toward the south', then for each basic unit (401), the number 1 of the subjective question is'Nordic','If the wind', and'I would like it'. In the subjective question, No. 2 can be'To Yum Hyang','Go to the sun','To listen well','If it's designed', and'I'd like it.' If each survey is removed, the subjective questionnaire number 1 is'North Europe','wind' and'I like it', and the subjective questionnaire number 2 is'south-facing','sun','to listen to','if designed', ' I wish it would be. The first meaning word 402 obtained from subjective question No. 1 may be'Nordic', and the second meaning word 403 may be'Nordic style wish' deduced from'I like it'. Since'Northern Europe' has already been included as the first meaning word 402, the second meaning word 403 may be deleted. The first meaning word 402 acquired by subjective question No. 2 may have a'south-facing', and the second meaning word 403 is'sun,' through'I hope it's designed to be good for the sun', 'Listen well' can be selected. Accordingly, the second questionnaire vector may include “North Europe” in the first row, and “southward” in the second row; 'year'; May include'to listen to'. The second questionnaire vector may be generally generated by adding one row to the number of subjective questions, and may include all of the answer sentences of the subjective question in the last row.

5 is a diagram for describing an artificial neural network used to obtain a robustness evaluation degree, a design evaluation degree, and a convenience evaluation degree according to an embodiment.

The artificial neural network may be a component included in the control device, and may be learned through the control device or a separate learning device.

The first evaluation artificial neural network 503 may output a robustness evaluation diagram 505 of each databaseized building object based on the previously reported defect information 501 of the databased building objects. The defect information 501 of each database-formed building object may be processed and generated as a first evaluation vector 502, and the first evaluation vector 502 corresponds to the input layer node of the first evaluation artificial neural network 503 It can be a vector. The first evaluation artificial neural network 503 outputs a first evaluation output 504, and the first evaluation output 504 may be a vector corresponding to an output layer node of the first evaluation artificial neural network 503. Based on the first evaluation output 504, the control device may generate a robustness evaluation map 505 of databased building objects.

The second evaluation artificial neural network 508 receives a review including an evaluation 506 for the design and convenience of the building from the reviews created by random samples online for the building corresponding to the building object, The design evaluation degree 511 and the convenience evaluation degree 512 of may be output as numerical values. The evaluation 506 for the design and convenience of the review may be processed and generated as a second evaluation vector 507, and the second evaluation vector 507 is a vector corresponding to the input layer node of the second evaluation artificial neural network 508 Can be The process in which the evaluation 506 for the design and convenience of the review is processed and generated into the second evaluation vector 507 is to classify the reviews into sentences, divide the classified sentences into spaces, place them in each column of the vector, and arrange them. The values of each column can be coded and processed, but is not limited thereto. The process of classifying into sentences may include periods, exclamation marks, quotation marks, double quotation marks, question marks, and line breaks, but is not limited thereto. The second evaluation artificial neural network 508 outputs the second evaluation output 509, and the second evaluation output 509 may be a vector corresponding to an output layer node of the second evaluation artificial neural network 508. Based on the second evaluation output 509, the control device may generate a proportional calculation 510 of the evaluation 506 for design and convenience of building objects.

Hereinafter, a process of learning an artificial neural network through a learning device will be described.

First, the learning device may acquire training data and a label.

For learning the first evaluation artificial neural network 503, the learning device may acquire defect information 501 for 10000 buildings acquired as big data through online as each training data. Labels may include results deduced from defect information 501 for buildings by architectural designers and construction designers.

For learning the second evaluation artificial neural network 508, the learning device may acquire 1000 randomly generated random building objects and 100 reviews randomly created for each random building object corresponding thereto, as each training data. Labels may include results evaluated by designers, architectural designers and construction designers of artificial neural networks.

Now, the learning device can generate an input of an artificial neural network from the training data.

The learning device may use the training data as an input of the artificial neural network or may generate an input of the artificial neural network after undergoing a known process of removing unnecessary information from each training data. Inputs corresponding to each artificial neural network may be referred to as nth training evaluation vectors. The input of the first evaluation artificial neural network 503 for learning may be first training evaluation vectors, and the input of the second evaluation artificial neural network 508 may be second training evaluation vectors.

Next, the learning device can apply the input to the artificial neural network.

The artificial neural network included in the server may be an artificial neural network that is learned according to supervised learning. The artificial neural network may be a convolutional neural network (CNN) or a recurrent neural network (RNN) structure suitable for training through supervised learning.

Then, the learning device may obtain an output from the artificial neural network.

The output of the first evaluation artificial neural network 503 may include a robustness evaluation degree 505 of a building object inferred from the defect information 501 of the building object. Specifically, the artificial neural network may output a value of the robustness evaluation degree 505 as a number based on the number, type, and degree of defect management of a building object, and a building age. The output of the first evaluation artificial neural network 503 may be a first evaluation output 504.

The output of the second evaluation artificial neural network 508 may include a proportional calculation 510 of the evaluation 506 for the design and convenience of the building design of the building object. Specifically, when there are significantly many reviews determined that artificial neural networks contain mentions that the building design is aesthetic; Few reviews have been determined to contain a statement that the building design is aesthetic; By analyzing the pattern according to the case where there is no review determined that the building design includes a mention that it is aesthetic, the design allocation value of the building design of the building object can be output as a number. At this time, the determination of whether or not a mention that the building design is aesthetic is included in the review may be performed by a keyword extraction method, a hashtag extraction method, a text mining method, or the like. The reasoning about the convenience allocation value can take the same method as that of the design allocation value. When the design allocation value and the convenience allocation value are determined, the second evaluation artificial neural network 508 can calculate the ratio of the two and perform a proportional calculation 510 of the evaluation 506 for the design and convenience of the building design of the building object. have. The output of the second evaluation artificial neural network 508 may be a second evaluation output 509.

Thereafter, the learning device can compare the output and the label. The process of comparing the output of the artificial neural network corresponding to the inference with the label corresponding to the correct answer may be performed by calculating a loss function. As the loss function, a known mean squared error (MSE), cross entropy error (CEE), etc. may be used. However, the present invention is not limited thereto, and loss functions used in various artificial neural network models may be used as long as a deviation, error, or difference between the output of the artificial neural network and the label can be measured. The first evaluation artificial neural network 503 is based on the difference between the first training evaluation output and the first training evaluation labels, and the second evaluation artificial neural network 508 is based on the difference between the second training evaluation output and the second training evaluation labels. can do.

Next, the learning device may optimize the artificial neural network based on the comparison value. By updating the weight of the nodes of the artificial neural network so that the comparison value of the learning device becomes smaller, the output of the artificial neural network corresponding to the inference and the label corresponding to the correct answer can be gradually matched. Through this, the artificial neural network It can be optimized to output an inference close to the correct answer. Specifically, the learning apparatus may optimize the artificial neural network by repeating the process of resetting the weight of the artificial neural network so that the loss function corresponding to the comparison value approaches the estimated value of the minimum value. For the optimization of artificial neural networks, known backpropagation algorithms, stochastic gradient descent, and the like can be used. However, the present invention is not limited thereto, and weight optimization algorithms used in various neural network models may be used.

The learning device can train an artificial neural network by repeating this process.

Through this, the learning device is based on the defect information 501 of the building objects; A first evaluation artificial neural network 503 that outputs a robustness evaluation degree 505 for a corresponding building object may be trained.

In addition, the learning device is based on a review including an evaluation 506 for design and convenience of a building design among reviews collected for a building corresponding to a building object; A second evaluation artificial neural network 508 that outputs a proportional calculation 510 of the evaluation 506 for the design and convenience of the building design of the corresponding building object may be trained.

6 is a diagram for explaining an artificial neural network used to generate a robustness requirement, a design requirement, and a convenience requirement according to an embodiment.

The artificial neural network may be a component included in the control device, and may be learned through the control device or a separate learning device.

The first requested artificial neural network may output a robustness requirement of the requested building corresponding to the target soil based on soil information of the target soil. The soil information of the target soil may be processed and generated as a first request vector, and the first request vector may be a vector corresponding to an input layer node of the first request artificial neural network. The first request artificial neural network outputs a first request output, and the first request output may be a vector corresponding to an output layer node of the first request artificial neural network. Based on the first requested output, the control device may generate a robustness requirement of the requested building corresponding to the target soil.

According to an embodiment, a subjective question among questionnaires may undergo a pre-processing process, through which a first meaning word and a second meaning word may be selected. Among them, an artificial neural network may be used to select the second semantic word, and the artificial neural network used in this case may be a first subject artificial neural network.

The first subjective artificial neural network may output the second meaning word based on the basic unit excluding the first meaning word and the original sentence of the subjective question. The basic unit excluding the first semantic word and the original sentence of the subjective question may be processed into a first subjective vector, and the first subjective vector may be a vector corresponding to an input layer node of the first subjective artificial neural network. The first subjective artificial neural network may output a first subjective output, and the first subjective output may be a vector corresponding to an output layer node of the first subjective artificial neural network. The control device may select a second meaning word based on the first subjective output.

The first questionnaire vector created to correspond to each row for each question from the objective questionnaire of the questionnaire question and the second questionnaire vector generated to correspond to each row of the first and second semantic words of the subjective questionnaire May correspond to input layer nodes of the first questionnaire artificial neural network and the second questionnaire artificial neural network, respectively. The outputs of the first questionnaire artificial neural network and the second questionnaire artificial neural network may be used to generate a second request vector.

The second requested artificial neural network may receive outputs of the first questionnaire artificial neural network and the second questionnaire artificial neural network as inputs, and may output a ratio of a design request and a convenience request for a target soil. The outputs of the first questionnaire artificial neural network and the second questionnaire artificial neural network may be processed and generated as a second request vector, and the second request vector may be a vector corresponding to an input layer node of the second request artificial neural network. The second request artificial neural network outputs the second request output, and the second request output may be a vector corresponding to an output layer node of the second request neural network. Based on the second request output, the control device may generate a ratio of a design request and a convenience request corresponding to the requested building of the target soil.

Hereinafter, a process of learning an artificial neural network through a learning device will be described.

First, the learning device may acquire training data and a label.

For learning the first required artificial neural network, the learning device may acquire information about 1000 pieces of soil that is randomly generated as each training data. Labels may include results evaluated by designers, architectural designers and construction designers of artificial neural networks.

For learning the first subjective artificial neural network, the learning apparatus may acquire basic units excluding words corresponding to the first semantic word from 100000 randomly generated subjective answers as each training data. Labels may include results evaluated by designers, architectural designers and construction designers of artificial neural networks.

For learning the first questionnaire artificial neural network and the second questionnaire artificial neural network, the learning device is the first semantic words and the second meaning of 100000 objective answers randomly generated and 100000 subjective answers used in the first subjective artificial neural network. Words can be acquired as individual training data. Labels may include results evaluated by designers, architectural designers and construction designers of artificial neural networks.

For learning the second required artificial neural network, the learning device may acquire first training questionnaire outputs and the second training questionnaire outputs as respective training data. Labels may include results evaluated by designers, architectural designers and construction designers of artificial neural networks.

Now, the learning device can generate an input of an artificial neural network from the training data.

The learning device may use the training data as an input of the artificial neural network or may generate an input of the artificial neural network after undergoing a known process of removing unnecessary information from each training data. Inputs corresponding to each artificial neural network may be referred to as nth training request vectors, nth training subject vectors, and nth training questionnaire vectors. Accordingly, the first training request vectors, the second training request vectors, the first training subject vectors, the first training questionnaire vectors, and the second training questionnaire vectors may be used as inputs of the corresponding artificial neural network.

Next, the learning device can apply the input to the artificial neural network.

The artificial neural network included in the server may be an artificial neural network that is learned according to supervised learning. The artificial neural network may be a convolutional neural network (CNN) or a recurrent neural network (RNN) structure suitable for training through supervised learning.

Then, the learning device may obtain an output from the artificial neural network.

The output of the first requested artificial neural network may be an inference about the robustness requirement of the requested building corresponding to the target soil from soil information of the target soil. Specifically, the demand for solidity of a building is a numerical percentage value based on an evaluation of the solidity of the building, and can be increased if the building requires high solidity. When the value of the robustness requirement is determined, the design requirement and the convenience requirement may be determined by distributing the remaining percentage values.

The output of the first subjective artificial neural network may be an inference for the second meaning word from the subjective question among questionnaires prepared by the client. Specifically, the artificial neural network may identify a classification criterion word from the subjective question and output a second meaning word of the subjective question through a combination of words positioned before and after the subjective question. In this case, a method of obtaining a classification criterion word and a combination of a word before and after in the subjective question may be performed by a keyword extraction method, a hashtag extraction method, a text mining method, or the like.

The output of the first questionnaire artificial neural network and the second questionnaire artificial neural network is an evaluation of the design and convenience corresponding to each from the first and second meaning words of the objective and subjective questionnaires among the questionnaire questions created by the client. It could be reasoning.

The output of the second requested artificial neural network may be a ratio of a design request and a convenience request obtained from the first questionnaire vector and the second questionnaire vector. Specifically, the design demand and convenience demand of the building are values obtained by distributing a percentage excluding the robustness demand, and may be a numerical percentage value based on an evaluation of the client's requirements for design and convenience. As the design demand increases, the convenience demand may decrease. Conversely, when the design demand decreases, the convenience demand may increase.

Thereafter, the learning device can compare the output and the label. The process of comparing the output of the artificial neural network corresponding to the inference with the label corresponding to the correct answer may be performed by calculating a loss function. As the loss function, a known mean squared error (MSE), cross entropy error (CEE), etc. may be used. However, the present invention is not limited thereto, and loss functions used in various artificial neural network models may be used as long as a deviation, error, or difference between the output of the artificial neural network and the label can be measured.

Next, the learning device may optimize the artificial neural network based on the comparison value. By updating the weight of the nodes of the artificial neural network so that the comparison value of the learning device becomes smaller, the output of the artificial neural network corresponding to the inference and the label corresponding to the correct answer can be gradually matched. Through this, the artificial neural network It can be optimized to output an inference close to the correct answer. Specifically, the learning apparatus may optimize the artificial neural network by repeating the process of resetting the weight of the artificial neural network so that the loss function corresponding to the comparison value approaches the estimated value of the minimum value. For the optimization of artificial neural networks, known backpropagation algorithms, stochastic gradient descent, and the like can be used. However, the present invention is not limited thereto, and weight optimization algorithms used in various neural network models may be used. The learning device can train an artificial neural network by repeating this process.

7 is an exemplary diagram of a configuration of an apparatus according to an embodiment.

The device 701 according to an embodiment includes a processor 702 and a memory 703. The processor 702 may include at least one of the devices described above with reference to FIGS. 1 to 6, or may perform at least one method described above with reference to FIGS. 1 to 6. Specifically, the device 701 may be a server 100, a user terminal 110, or an artificial neural network learning device. A person or organization using the device 701 may provide services related to some or all of the methods described above with reference to FIGS. 1 to 6.

The memory 703 may store information related to the above-described methods or a program in which methods described below are implemented. The memory 703 may be a volatile memory or a nonvolatile memory.

The processor 702 can execute a program and control the device 701. The code of a program executed by the processor 702 may be stored in the memory 703. The device 701 is connected to an external device (eg, a personal computer or a network) through an input/output device (not shown), and may exchange data through wired or wireless communication.

The device 701 may be used to train an artificial neural network or use the learned artificial neural network. The memory 703 may include a learning artificial neural network or a learned artificial neural network. The processor 702 may train or execute an artificial neural network algorithm stored in the memory 703. The apparatus 701 for training the artificial neural network and the apparatus 701 using the learned artificial neural network may be the same or may be separate.

The embodiments described above may be implemented as a hardware component, a software component, and/or a combination of a hardware component and a software component. For example, the devices, methods, and components described in the embodiments include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA). array), programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions, such as one or more general purpose computers or special purpose computers. The processing device may execute an operating system (OS) and one or more software applications executed on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of software. For the convenience of understanding, although it is sometimes described that one processing device is used, one of ordinary skill in the art, the processing device is a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that it may include. For example, the processing device may include a plurality of processors or one processor and one controller. In addition, other processing configurations are possible, such as a parallel processor.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -A hardware device specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of the program instructions include not only machine language codes such as those produced by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

The software may include a computer program, code, instructions, or a combination of one or more of these, configuring the processing unit to behave as desired or processed independently or collectively. You can command the device. Software and/or data may be interpreted by a processing device or to provide instructions or data to a processing device, of any type of machine, component, physical device, virtual equipment, computer storage medium or device. , Or may be permanently or temporarily embodyed in a transmitted signal wave. The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

As described above, although the embodiments have been described by the limited drawings, a person of ordinary skill in the art can apply various technical modifications and variations based on the above. For example, the described techniques are performed in a different order from the described method, and/or components such as a system, structure, device, circuit, etc. described are combined or combined in a form different from the described method, or other components Alternatively, even if substituted or substituted by an equivalent, an appropriate result can be achieved.

Therefore, other implementations, other embodiments and claims and equivalents fall within the scope of the following claims.

Claims (3)

In the method of estimating construction and construction cost,
Obtaining, by a control device, soil information of the target soil of the client who requested the building and construction work-the soil information includes the area, slope, lower layer soil quality, sunlight direction, and building usage of the target soil;
Generating, by the control device, a solidity requirement of the requested building based on the soil information;
The control device evaluates the client's questionnaire-The questionnaire evaluation includes 40 to 100 questionnaire questions that reflect the evaluation of the requirements for the requested building, and each question consists of an objective question and a subjective question. Creating a design requirement and a convenience requirement based on the basis;
Acquiring, by the control device, the databaseized building objects in which the soil information of the databased building object satisfies the soil information of the target soil and a first demand satisfaction level from among the building objects databased by the control device;
Among the building objects obtained by the control device, a solidity evaluation degree, a design evaluation degree, and a convenience evaluation degree created in advance for the acquired building object, the robustness request degree, the design request degree, and the convenience evaluation degree generated for the target soil Selecting the obtained building objects satisfying a requirement degree and a second requirement satisfaction level;
Construction information required for the control device to construct the requested building in the target soil based on each selected building object-the construction information is a construction plan corresponding to the databaseized building objects, a construction plan flow chart, and construction Outputting-including equipment and construction materials; And
Generating a construction cost estimate of the requested building for the target soil based on the construction information output by the control device
Including,
The robustness evaluation degree of the databased building object is
Defect information reported in advance for the databaseized building object-The defect information is information provided by an evaluation of a building design expert and a construction safety expert for a building corresponding to the databased building object, for the building Based on the number, type, degree of defect management that occurred, and including the building age of the building, it is made with a mechanical evaluation of the robustness of the building,
The firmness evaluation is expressed as a percentage value,
When the robustness evaluation degree is determined, the design evaluation degree and the convenience evaluation degree are allocated to the remaining values of the percentage,
The design evaluation degree and the convenience evaluation degree,
It is obtained from the evaluation of the design and convenience of random samples created online for the building corresponding to the databased building object,
Based on the proportional calculation of the evaluation of the design and convenience, a percentage value excluding the robustness evaluation degree is discretely allocated to the design evaluation degree and the convenience evaluation degree according to the rational calculation,
The higher the design evaluation degree, the lower the convenience evaluation degree,
The higher the convenience evaluation, the lower the design evaluation,
Evaluation of the above design and convenience
The random samples are calculated based on reviews written on the building,
The step of generating the robustness requirement of the target soil,
Generating, by the control device, the robustness requirement as a percentage value through mechanical evaluation based on the soil information of the target soil
Including,
Generating the design requirement and the convenience requirement of the target soil,
Separating, by the control device, the objective question and the subjective question of the questionnaire evaluation;
Performing, by the control device, pre-processing of the objective and subjective questions;
Obtaining, by the control device, a result of mechanical evaluation through an artificial neural network corresponding to each of the objective and subjective questions preprocessed;
Generating, by the control device, a ratio of the design requirement and the convenience requirement based on a result of the mechanical evaluation; And
Generating, by the control device, the design request and the convenience request based on a ratio of the design request and the convenience request to a percentage value excluding the robustness request
Including
How to estimate construction and construction costs.
delete The method of claim 1,
The step of performing pre-processing of the multiple-choice question and the subjective question
Generating, by the control device, a first questionnaire vector so that the objective questionnaire corresponds to each row for each question;
The basic unit in which the control device pre-determines the subjective question-The basic unit follows the grammatical word unit division, but in the case of adverbs and idioms, it is marked with the following adverbs, idioms, interjections, adjectives and verbs. Separating into;
Removing, by the control device, irradiation from the basic unit;
Selecting, by the control device, a first meaning word from the basic unit from which the irradiation has been removed;
Selecting, by the control device, a second meaning word based on a mechanical evaluation of the basic unit excluding the first meaning word; And
Generating, by the control device, a second questionnaire vector so that the first meaning word and the second meaning word correspond to each row for each question
Including
How to estimate construction and construction costs.

Images