KR102131138B1 - system and method predicting energy expenditure based on machine learning using 2D drawing - Google Patents

Abstract

Disclosed are a system for predicting energy consumption based on machine learning using a two-dimensional drawing and a method thereof. The system for predicting energy consumption includes: a building basic information input module that receives a learning model input factor including basic building information of a building from a user; a 2D drawing creation module that provides a 2D modeling creation tool to allow a user to create a 2D modeling drawing, sets color information for usage of each room of the created 2D modeling drawing, numerically receives floor height information of each room, and numerically receives height information of a window installed in the room to generate 2D modeling drawing information; an extraction module that extracts a learning model extraction factor based on the 2D modeling drawing information; a database of energy consumption for each variable according to a representative gross area in which information on the energy consumption for each variable according to the representative gross area is stored; an energy consumption prediction module that constructs a machine learning model factor including the learning model input factor received from the building basic information input module and the learning model extraction factor received from the extraction module, wherein the energy consumption prediction module matches the learning model factor and the energy consumption information for each variable according to the representative gross area and calculates and predicts the energy consumption for a corresponding building by performing machine learning based on the energy consumption information for each variable according to the representative gross area matched with the learning model factor.

Description

System and method predicting energy expenditure based on machine learning using 2D drawing}

The present invention relates to a system and method for predicting energy requirements based on machine learning, and more particularly, to a system and method for predicting energy requirements based on machine learning using 2D drawings that are two-dimensional.

Recently, due to climate change due to global warming, energy consumption and carbon dioxide generation in buildings are increasing, and various efforts have been made to reduce such energy consumption and carbon dioxide generation.

In particular, in order to improve the efficiency of building energy use, the government evaluates the energy consumption of heating, cooling, hot water, etc., and the amount of carbon dioxide generated through the design book of the building, and implements a system for certifying according to energy performance.

Accordingly, it is required to predict building energy performance before building or remodeling the building.

To this end, programs for simulating the amount of energy consumed by buildings, distribution, trends, etc. have been developed according to certain conditions. However, since such programs are complicated and difficult to use, there is a problem that an expert who can use the programs is needed. .

In addition, in the program for simulating the energy performance of existing low-rise residential buildings, it is possible to simulate the energy requirements of the building by creating drawings using other 2D drawing creation programs, creating complex envelope development drawings, and inputting heat permeation rates, etc. There was a problem in that the work efficiency was very complicated and cumbersome, and the efficiency of the work was very low.

In addition, recently, an effective and accurate method for simulating building energy based on 3D-based BIM (Building Information Modeling) information has been proposed, but in this case, excessive work is required to model and build 3D-based BIM information. In addition to having to be charged, time and money can be excessively charged, resulting in a decrease in work efficiency.

Patent No. 10-1418171 Patent No. 10-1576605 Patent Registration No. 10-2028861

The present invention was devised to solve the conventional problems as described above, and does not use complex and difficult to construct BIM information. It is a machine learning model factor required for machine learning based on basic building information and 2D modeling drawing input information. By constructing and performing machine learning based on this machine learning model factor and energy requirements database for each representative gross variable, it is possible to predict the energy demand of the building, simplifying energy demand forecasting for low-rise residential buildings and improving work efficiency. The objective is to provide a system and method for predicting energy requirements based on machine learning using two-dimensional drawings.

A machine learning-based energy demand prediction system using a two-dimensional drawing of the present invention for achieving the above object includes: a building basic information input module that receives a learning model input factor consisting of basic building information from a user; Provides 2D modeling creation tools to enable users to create 2D modeling drawings, sets color information for each room in each 2D modeling drawing, and numerically inputs floor height information for each room. A 2D drawing creation module configured to generate 2D modeling drawing information by numerically receiving window height information installed in the window; An extraction module for extracting a learning model extraction factor based on the 2D modeling drawing information; An energy requirement database for each variable according to the representative total area, in which energy requirement information for each variable according to the representative total area is stored; Constructing a machine learning model factor including the learning model input factor received from the building basic information input module and the learning model extraction factor received from the extraction module, and energy for each variable according to the learning model factor and the representative gross area. It includes an energy demand prediction module that estimates and estimates the energy demand for a building by matching machine requirements and performing machine learning based on energy demand information for each variable according to the representative gross area matched with the learning model factor. Is done.

The extraction module includes a coordinate extraction unit for extracting coordinate values for each room and each window from 2D modeling drawing information created through the 2D drawing creation module, coordinate information extracted from the coordinate extraction unit, and the floor height and height information. Based on the window height information, it is configured to calculate the total floor area, the main area, the window area ratio, and the long and short constipation, and calculate the number of toilets, the number of rooms, and the number of balconies through the color information. It may be configured to include; a factor calculation unit for obtaining the learning model extraction factors consisting of the number, number of rooms, the number of balcoty.

The learning model input factors include building use, completion year, area, building location, occupancy schedule, and number of occupants. The learning model extraction factors are gross area, circumference, window area ratio, long and short constipation, number of toilets, number of rooms, number of balconies Can be configured.

On the other hand, the method for predicting energy requirements based on machine learning using the two-dimensional drawing of the present invention includes: a) The basic building information input module is the basic building information for the building, such as building use, completion year, area, building location, occupancy schedule, and occupant Receiving a learning model input factor composed of numbers; b) the 2D drawing creation module creates a 2D modeling drawing by the user, sets color information for each use on each 2D drawing, and receives the floor height information of each room and window height information of each window; c) the extraction module obtaining a learning model extraction factor through the 2D modeling drawing information created in step b); d) The learning model input factor and the learning model extraction factor are configured as machine learning model factors necessary for machine learning, matching the machine learning model factor and energy requirement information for representative gross area variables, and matching the machine learning model factor. It can be configured to include; performing the machine learning with the energy requirements information for each representative gross area variable to estimate and estimate the energy requirements for the building.

In step c), the extraction module extracts coordinates for each room through shape information of the 2D modeling drawing information, and based on the extracted coordinate information, floor height information, and window height information, the total area, circumference, and window area ratio, Calculate the long and short constipation, calculate the number of toilets, the number of rooms, and the number of balconies through the color information, and extract the learning model consisting of the total area, circumference, window area ratio, long and short constipation, number of toilets, number of rooms, number of balconies Can be configured to obtain.

According to the above, the present invention acquires the learning model factors necessary for the machine learning model based on the input basic building information and the 2D modeling drawing information, without the need to use complex and difficult to build 3D BIM information. This is a method to estimate the energy demand of a building by performing machine learning based on learning model factors and energy requirement information for representative gross area variables. It can improve the convenience of energy demand forecasting, and the time required for energy demand forecasting. It has the effect of greatly reducing the work efficiency.

In addition, the present invention can significantly reduce the cumbersome and time-consuming work process of creating a complex envelope development or inputting a heat permeation rate without using a separate 2D drawing creation program, thereby improving the work efficiency of energy demand forecasting. There is an effect that a more concise and convenient energy demand prediction can be made.

1 is a block diagram of a system for predicting energy consumption based on machine learning using a two-dimensional drawing according to an embodiment of the present invention,
2 is a conceptual diagram of a whole process of a system for predicting energy requirements based on machine learning using a 2D drawing according to an embodiment of the present invention,
3 is a flow chart of a method for predicting energy demand based on machine learning using a two-dimensional drawing according to an embodiment of the present invention.

The present invention can be applied to various changes and can have various embodiments, and specific embodiments are illustrated in the drawings and described in detail in detail for carrying out the invention. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. In describing each drawing, similar reference numerals are used for similar components.

Terms such as first, second, A, and B can be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from other components. For example, the first component may be referred to as a second component without departing from the scope of the present invention, and similarly, the second component may be referred to as a first component. The term and/or includes a combination of a plurality of related described items or any one of a plurality of related described items.

When an element is said to be "connected" or "connected" to another component, it is understood that other components may be directly connected to or connected to the other component, but there may be other components in between. It should be. On the other hand, when a component is said to be "directly connected" or "directly connected" to another component, it should be understood that no other component exists in the middle.

Terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, terms such as “include” or “have” are intended to indicate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, and that one or more other features are present. It should be understood that the existence or addition possibilities of fields or numbers, steps, operations, components, parts or combinations thereof are not excluded in advance.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person skilled in the art to which the present invention pertains. Terms, such as those defined in a commonly used dictionary, should be interpreted as having meanings consistent with meanings in the context of related technologies, and should not be interpreted as ideal or excessively formal meanings unless explicitly defined in the present application. Does not.

Hereinafter, a system and method for predicting energy requirements based on machine learning using a two-dimensional drawing according to a preferred embodiment of the present invention will be described in detail with reference to FIGS. 1 to 3 with reference to the accompanying drawings.

The system and method for estimating energy requirements based on machine learning using the two-dimensional drawings described below are implemented through hardware components such as a computer including software and/or software capable of executing each component and its operations described below. It will be understood by those skilled in the art.

Referring to Figure 1, the machine learning-based energy requirements prediction system 100 using the secondary drawing of the present invention is a basic building information input module 110, 2D drawing creation module 120, extraction module 130, representative It is configured to include the energy requirements database 140 for each area variable, the energy requirements prediction module 150.

The energy demand prediction module 150 of the present invention is configured to calculate and predict energy demand for a corresponding building using a machine learning model such as deep learning, and at this time, it is necessary for the machine learning model in the energy demand prediction module 150 Machine learning model factors are composed of 13 machine learning model factors as shown in Table 1 below.

The 13 machine learning model factors required for the machine learning model are factors that affect the building energy requirements, and include basic information about the building, including information about the actual shape of the building.

number Machine learning model factors Argument Description One Building use (A) Deviation of energy requirements according to building characteristics 2 Year Built (B) Structure heat transmission rate regulation 3 Region (C) Reflecting climate conditions 4 Building location (D) Definition of insulation boundary conditions 5 Appointment Schedule (E) Pattern of energy demand for occupants 6 Number of occupants (F) Pattern of energy demand for occupants 7 Total area (G) (Surface information)-Total building area value (m2) 8 Incense (H) Positioning the main window in the living room 9 Window area ratio (I) Window area/Total outer wall area value 10 Constipation (J) Width: Vertical length value matching 11 Number of toilets (K) (Color classification of rooms) Use of heating and non-heating space information 12 Number of rooms (L) (Room color classification) Heat storage element calculation 13 Number of balconies (M) (Color classification of rooms) Use of heating and non-heating space information

In the present invention, 13 machine learning model factors are 6 learning model input factors (numbers 1 to 6, building use (A), completion year (B), area (C), building location (D), and occupancy schedule (E)). , Number of occupants (E), and 7 learning model extraction factors (numbers 7 to 13, total area (G), circumference (H), window area ratio (I), long and short constipation (J), number of toilets (K), rooms Number (L), number of balcony (M)), 6 learning model input factors are configured to be input from the user through the building basic information input module 110, and 7 learning model extraction factors are 2D. It is configured to be calculated based on the extraction and extraction information from the 2D modeling drawing information of the corresponding building created by the user through the drawing creation module 120.

Building basic information input module 110 is the basic building information consisting of building use (A), completion year (B), area (C), building location (D), occupancy schedule (E), and number of occupants (E) from the user Is configured to receive the input, and the received basic information of the building is configured to be transmitted to the energy demand prediction module 150 as an input factor of the learning model.

The 2D drawing creation module 120 provides a 2D drawing creation tool so that a user can create 2D modeling drawings for the building. The 2D modeling drawing information created by the 2D drawing creation module may be configured to include shape information about the target room and window. In addition, the 2D drawing creation module 120 has a color setting unit for setting color for each room use, such as a bathroom, a room, and a balcony, for each room of the created 2D modeling drawing, and a floor height for numerically inputting the floor height of each room It may be configured to further include an input unit and a window height input unit for numerically inputting a window height installed in the room.

Accordingly, the 2D modeling drawing information created through the 2D drawing creation module 120 is configured to include the above-mentioned shape information, color information set in each room, floor height information of each room, and window height information installed in the room. Can be.

The extraction module 130 may be configured to extract seven learning model extraction factors from 2D modeling drawing information created through the 2D drawing creation module 120.

Specifically, the extraction module 130 includes a coordinate extraction unit 132 for extracting coordinate values for each room and each window from the 2D modeling drawing information created through the 2D drawing creation module 120, and the coordinate extraction unit 132 ), calculate the total area (G), circumference (H), window area ratio (I), and long-side constipation (J) based on the coordinate information, floor height and window height information extracted from the ), the number of openings (L), the number of balconies (M), and the total area (G), circumference (H), window area ratio (I), long and short constipation (J), number of toilets (K), number of rooms (L) , Factor calculation unit 134 for acquiring 7 learning model extraction factors composed of the number of balconies (M).

Since some studies that extracted 2D information from the existing building energy simulation model extracted only basic coordinates or length information about the room, it was impossible to extract the learning model factors necessary for the machine learning model, but the extraction module 130 of the present invention The coordinate extraction unit 132 extracts coordinate information for each room and each window from the 2D modeling drawing information, and based on the extracted coordinate information and the floor height information, window height information, and color information included in the 2D modeling information. , 7 learning model extraction factors required for machine learning model (No. 7~13, total area (G), circumference (H), window area ratio (I), long and short constipation (J), number of toilets (K), number of rooms (L) ), the number of balconies (M)) can be calculated and extracted.

Table 2 shows seven learning model extraction factors and extraction methods extracted by the extraction module 130 from 2D modeling drawing information.

number Learning model extraction factors Extraction method Extraction place One Total area (G) Calculate by multiplying after extracting X,Y coordinates of drawing 2D drawing 2 Incense (H) Calculate the coordinates of the living room window and multiply by the vector for the direction 2D drawing 3 Window area ratio (I) Calculate the area of the wall with a window and divide it with the area of the window 2D drawing 4 Constipation (J) After extracting the X and Y coordinates of the drawing, calculate the ratio of X and Y 2D drawing 5 Number of toilets (K) Count the number after grasping the color input 2D drawing 6 Number of rooms (L) Count the number after grasping the color input 2D drawing 7 Number of balconies (M) Count the number after grasping the color input 2D drawing

The energy requirements database 140 for each representative area variable is configured to store energy requirement information according to the representative total area, and is configured to provide energy consumption information for each variable according to the representative total area to the energy consumption prediction module 150.

The energy demand prediction module 150 includes a machine learning lab anatomy 152 in which a machine learning model is rescued and performs machine learning, and can be learned to predict the energy demand for the building through the built machine learning model. .

Specifically, the energy requirement prediction module 150 includes six learning model input factors (building use (A), completion year (B), area (C), building location (D), from the building basic information input module 110), Receiving schedule (E), number of occupants (E), and extracting seven learning models from the extraction module 130 (Gross Area (G), Perimeter (H), Window Area Ratio (I), Long-term Constipation (J)) , The number of toilets (K), the number of rooms (L), the number of balconies (M)), and is configured to receive energy requirement information for each variable according to the representative floor area from the energy requirement database 140 for each representative floor area variable.

Next, the energy demand prediction module 150 matches energy demand information for each representative floor area variable based on 13 machine learning model factors including 6 learning model input factors and 7 learning model extraction factors, and 13 machines The energy requirements for each building are inserted into the machine learning execution unit 152 to match the learning model factors, and machine learning is performed to predict and calculate the energy requirements for the building. It may be configured to be displayed on the display unit.

As described above, the present invention acquires the learning model factors necessary for the machine learning model based on the input basic building information and the 2D modeling drawing information, without the need to use complex and difficult to construct 3D BIM information, and the acquired learning model factors Based on the energy requirement information for each of the representative gross area variables, it is possible to estimate the energy requirements of buildings through machine learning, thereby improving the convenience of estimating energy requirements, and the time required to estimate energy requirements is also significantly reduced. Speed can be secured.

In particular, the present invention is provided with a 2D drawing creation module in the system without using a separate 2D drawing creation program, thereby enabling convenient 2D modeling drawing, thereby improving work efficiency of energy demand prediction.

Meanwhile, with reference to FIG. 3, a method for predicting energy consumption based on machine learning using a 2D drawing will be described.

Referring to FIG. 3, the building basic information input module 110 allows the user to use the building (A), completion year (B), area (C), building location (D), occupancy schedule (E), and number of occupants ( E) receives basic building information (S101).

Here, the basic information of buildings consisting of building use (A), year of completion (B), area (C), building location (D), occupancy schedule (E), and number of occupants (E) is required for the machine learning model. Used as a model input factor.

Next, the 2D drawing creation module 120 creates a 2D modeling drawing by the user, sets color information for each room on the 2D drawing, and sets the floor height for each room and window height information for each window. Input is received (S02).

Accordingly, the created 2D modeling drawing information may include shape information representing each room and each window for the corresponding building, color information for each use for each room, floor height information for each room, and window height information for each window.

Next, the extraction module 130 extracts coordinates for each room through shape information of the 2D modeling drawing information, and based on the extracted coordinate information, floor height information, and window height information, the gross area (G), orientation ( H), window area ratio (I), long and short constipation (J) are calculated, and the number of toilets (K), number of open doors (L), and number of balconies (M) are calculated through color information, and the total area (G) and circumference ( H), seven learning model extraction factors consisting of window area ratio (I), long and short constipation (J), number of toilets (K), number of open rooms (L), and number of balconies (M) can be obtained (S103).

Next, the energy requirement prediction module 150 inputs six learning models from the building basic information input module 110 (building use (A), completion year (B), area (C), building location (D), Receiving schedule (E), number of occupants (E), and extracting seven learning models from the extraction module 130 (Gross Area (G), Perimeter (H), Window Area Ratio (I), Long-term Constipation (J)) , Receives the number of toilets (K), the number of rooms (L), the number of balconies (M)), receives energy requirement information for each variable according to the representative gross area from the energy requirement database 140 for each representative area variable, and then Based on 13 machine learning model factors including learning model input factors and 7 learning model extraction factors, it matches energy demand information for each representative gross area variable, and energy requirement information for representative total area variables matched with 13 machine learning model factors. Machine learning is performed to predict the energy demand for the building (S104). The energy requirement for the predicted building can be output to the deplay unit.

Although described with reference to the above embodiment, those skilled in the art can understand that the present invention can be variously modified and changed without departing from the spirit and scope of the present invention as set forth in the claims below. There will be.

110...building basic information input module
120...2D drawing creation module
130...extraction module
132...Coordinate extraction unit
134...Factor
150...Representative energy requirement database by gross area variable
150...Energy requirement prediction module
152...Machine Learning Executive

Claims (5)

A building basic information input module that receives a learning model input factor consisting of basic building information about a building from a user;
Provides a 2D modeling creation tool so that users can create 2D modeling drawings, and a color setting unit that sets color information for each room in each 2D modeling drawing created numerically. It is configured to include an input height height input unit and a window height input unit for numerically inputting window height information installed in a room, and 2D modeling drawing information including color information, the floor height height information, and the window height information for each room use. 2D drawing creation module configured to generate;
An extraction module for extracting learning model extraction factors based on the 2D modeling drawing information;
An energy requirement database for each variable according to the representative total area, in which energy requirement information for each variable according to the representative total area is stored;
Constructing a machine learning model factor including the learning model input factor received from the building basic information input module and the learning model extraction factor received from the extraction module, and energy for each variable according to the learning model factor and the representative gross area. It includes; energy requirements prediction module that estimates and estimates the energy demand for the building by matching the required information and performing machine learning based on the variable energy demand information according to the representative gross area matched with the learning model factor.
The extraction module,
A coordinate extraction unit for extracting coordinate values for each room and each window from the 2D modeling drawing information created through the 2D drawing creation module;
Based on the coordinate information extracted from the coordinate extracting unit, the floor height information, and the window height information, calculate the total area, circumference, window area ratio, and long and short constipation, and calculate the number of toilets, the number of rooms, and the number of balconies through the color information. It is configured to be, including the factor of calculating the learning model extraction factors consisting of the total area, circumference, window area ratio, pros and cons ratio, number of toilets, number of rooms, number of balcoty;
The learning model input factors include building use, completion year, area, building location, occupancy schedule, and number of occupants.
The learning model extraction factor is a machine learning-based energy demand prediction system using a two-dimensional drawing, characterized in that it consists of a total area, circumference, window area ratio, long and short constipation, number of toilets, number of rooms, and number of balconies.
delete delete a) the building basic information input module inputting a learning model input factor consisting of building use, completion year, area, building location, occupancy schedule, and number of occupants as basic building information for the building;
b) the 2D drawing creation module creates a 2D modeling drawing by the user, sets color information for each application on the created 2D drawing, and receives floor height information of each room and window height information of each window;
c) The extraction module extracts the coordinates for each room through the shape information of the 2D modeling drawing information, and calculates the total area, circumference, window area ratio, and long-term constipation based on the extracted coordinate information, floor height information, and window height information. And calculating the number of toilets, the number of rooms, and the number of balconies through the color information to obtain a learning model extraction factor consisting of the total area, circumference, window area ratio, long and short constipation, number of toilets, number of rooms, and number of balconies;
d) The learning model input factor and the learning model extraction factor are configured as machine learning model factors necessary for machine learning, matching the machine learning model factor and energy requirement information for representative gross area variables, and matching the machine learning model factor. A machine learning-based energy demand prediction method using a two-dimensional drawing, characterized in that it is configured to include; performing a machine learning with energy requirements information for each representative gross area variable to estimate and estimate the energy demand for the building.




delete

Images