KR20230093678A - Apparatus and method for managing energy of building using reinforcement learning - Google Patents

Abstract

In the method for managing the energy of a building through reinforcement learning, according to an embodiment of the present invention, (a) collecting first state data generated during a specific period in the past from a plurality of sensors disposed in a building; inputting state 1 data into an LSTM model and outputting predicted state data for a predetermined period from now; (b) Reinforcement that sets the first state data and predicted state data as state values, outputs building management data as action values, and provides compensation as the expected energy consumption is lower when the indoor temperature of the building is a specific temperature value learning a learning-based building management model; And (c) inputting the second state data received in real time from the sensor into the building management model and outputting the building management data to operate the air conditioning and heating devices provided in the building; State data and predicted state data are data of indoor and outdoor temperatures and energy usage for a predetermined period of time, and building management data include data on power level fluctuations supplied to air conditioning and heating devices.

Description

Apparatus and method for managing energy of a building using reinforcement learning {APPARATUS AND METHOD FOR MANAGING ENERGY OF BUILDING USING REINFORCEMENT LEARNING}

The present invention relates to a building energy management system (BEMS) that manages energy used inside a building using reinforcement learning.

As construction technology has developed compared to the past, the size of buildings is also rapidly increasing compared to the past. In this situation, the problem of maintaining a pleasant condition inside the building, which did not exist in the past, has emerged. For example, cooling or heating can be a condition for maintaining a comfortable condition inside a building.

Meanwhile, as buildings account for about 20% of the total energy consumption in Korea, they are emerging as an important field in achieving the national greenhouse gas reduction target.

Accordingly, the concept of a building energy management system (BEMS) was developed, and BEMS refers to a system that improves energy use efficiency by collecting and analyzing various information of energy management facilities in a building in real time.

However, the existing BEMS has a problem in that it is difficult to achieve energy optimization of a building because it simply performs energy consumption monitoring or device efficiency analysis without automatic control or optimization function. In addition, since it is difficult to consider various factors that greatly affect energy control, such as the external environment, practical energy optimization has many problems.

In order to solve the above problems, an object of the present invention is to implement a system for implementing optimization of energy used in a building using artificial intelligence.

Specifically, optimal results for external factors that affect building conditions and energy use are derived through artificial intelligence, so that optimal energy saving methods can be proposed.

In addition, it is possible to implement energy optimization by deriving an optimal temperature suitable for user's convenience as well as reducing energy consumption.

In the method for managing the energy of a building through artificial intelligence, performed by a server, according to an embodiment of the present invention for achieving the above technical problem, (a) from a plurality of sensors disposed in the building collecting first state data generated during a specific period of time, inputting the first state data into an LSTM model, and outputting predicted state data for a predetermined period from now; (b) Reinforcement that sets the first state data and predicted state data as state values, outputs building management data as action values, and provides compensation as the expected energy consumption is lower when the indoor temperature of the building is a specific temperature value learning a learning-based building management model; And (c) inputting the second state data received in real time from the sensor into the building management model and outputting the building management data to operate the air conditioning and heating devices provided in the building; The state data and the predicted state data are data of the indoor and outdoor temperatures of the building and energy consumption for a predetermined time, and the building management data may include data on power level fluctuations supplied to air conditioning and heating devices.

In addition, the LSTM model may generate predicted state data by calculating an expected indoor temperature for various energy usage based on the first state data.

In addition, the indoor temperature calculated according to the size of each energy consumption is corrected based on the outdoor temperature and climate, and a correction value according to the weather outside the building and the season may be applied.

In addition, the feasibility of energy usage may be determined based on the indoor/outdoor temperature for the section in which energy usage has increased rapidly among the first state data, and a correction value for the section in which energy usage has rapidly increased may be applied according to the validity.

In addition, the building management model may generate building management data so as to change the indoor temperature closest to a specific temperature while calculating the minimum energy consumption for real-time indoor temperature fluctuations.

In addition, when a plurality of air conditioning and heating devices are provided for each area of a building, the building management data generated by the building management model can individually adjust the power level applied to the air conditioning and heating devices for each area of the building.

In addition, in step (c), if the specific temperature is set differently according to the change of time, the building management data generated by the building management model adjusts the power level applied to the air conditioning and heating device according to the change of time according to the specific temperature. can change

In addition, step (b) further sets light management data including on/off times of a plurality of lights installed inside the building as a state value, and step (c) sets multiple lights through building management data. The on/off of the lights can be further adjusted at specific times.

Also, after the step (c), the LSTM model and the building management model may be updated based on the second state data and the building management data.

In addition, the LSTM model may be corrected based on an error generated by comparing the second predicted state data calculated by inputting the second state data into the LSTM model and the building management data.

In addition, even though the temperature of the building is adjusted based on the building management data after step (c), if the energy consumption or the real-time internal temperature of the building is out of the preset threshold range, it is determined that a problem has occurred in the air conditioning and heating device, A warning alarm may be provided to the set manager terminal.

In addition, in the device for managing the energy of a building through artificial intelligence, a program that provides a method for managing energy in a building through artificial intelligence is stored in a memory and the program stored in the memory is executed to save the energy of the building through artificial intelligence. Includes a processor that provides a management method, wherein the processor collects first state data generated during a specific period in the past from a plurality of sensors disposed in the building, inputs the first state data into an LSTM model, and inputs the first state data to a predetermined period from the present Output predicted state data for the future, set the first state data and predicted state data as state values, and output building management data as action values, but when the indoor temperature of the building is a specific temperature value, the expected energy consumption is low. It trains a building management model based on reinforcement learning to provide recording compensation, inputs the second state data received in real time from the sensor into the building management model, and outputs the building management data to control the air conditioning and heating devices provided in the building. operation, the first and second state data and the predicted state data are data of the indoor and outdoor temperature of the building and energy consumption used for a predetermined time, and the building management data includes data on power level fluctuations supplied to air conditioning and heating devices It may be a device that

In addition, it may be a computer readable storage medium on which a program for performing the method of managing energy of a building through artificial intelligence according to claim 1 is recorded.

According to one embodiment of the present invention, it is possible to implement a system for optimizing energy used in a building using artificial intelligence.

Specifically, optimal results for external factors that affect building conditions and energy use are derived through artificial intelligence, so that optimal energy saving methods can be proposed.

In addition, it is possible to implement energy optimization by deriving an optimal temperature suitable for user's convenience as well as reducing energy consumption.

1 is a diagram showing the configuration of a system for managing energy of a building through artificial intelligence according to an embodiment of the present invention.
2 is a diagram showing the configuration of a server according to an embodiment of the present invention.
3 is an operational flowchart illustrating a process of managing energy of a building through artificial intelligence according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail so that those skilled in the art can easily practice the present invention with reference to the accompanying drawings. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected" but also the case where it is "electrically connected" with another element interposed therebetween. . In addition, when a part "includes" a certain component, this means that it may further include other components, not excluding other components, unless otherwise stated, and one or more other characteristics. However, it should be understood that it does not preclude the possibility of existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

The following examples are detailed descriptions for better understanding of the present invention, and do not limit the scope of the present invention. Therefore, inventions of the same scope that perform the same functions as the present invention will also fall within the scope of the present invention.

Throughout the specification, 'server 100' may refer to a device that manages energy of a building through artificial intelligence.

1 is a diagram showing the configuration of a system for managing energy of a building through artificial intelligence according to an embodiment of the present invention.

Referring to FIG. 1 , according to an embodiment of the present invention, a system may include a server 100 and a building sensor 200 . At this time, each device may be interconnected wired or wirelessly through a communication network.

According to an embodiment of the present invention, the server 100 may collect state data from the building sensors 200 disposed in the building and input them to an LSTM model to calculate predicted state data.

At this time, the state data and the predicted state data may mean data of the indoor and outdoor temperatures of the building and the amount of energy used for a predetermined time from the building sensor 200 . That is, the external temperature of the building, the temperature inside the building formed through cooling or heating, and the amount of energy used to maintain the temperature inside the building may be included. At this time, the energy may generally be electricity, but since other energy sources may be developed later, the scope of the invention is not limited by defining the type of energy in the present invention.

In addition, the server 100 may teach the first state data and the predicted state data generated through the LSTM model to each preset building management model. At this time, the building management model is a type of artificial intelligence, and a model having the characteristics of supervised learning, unsupervised learning, or reinforcement learning can be applied, and in this specification, the invention will be described based on reinforcement learning for smooth explanation.

Through this, the server 100 inputs the state data received from the building sensor 200 through the learned building management model, calculates the optimal building management data, and operates the air conditioning and heating devices provided in the building. to be

That is, in the building management model, the building management data may include energy usage and temperature control data inside the building. That is, data on how to perform heating and cooling inside the building and energy consumption therefor may be included.

According to an embodiment of the present invention, the building sensor 200 is generally a temperature sensor for measuring the temperature inside and outside the building and a temperature sensor for measuring the energy used by various devices (air conditioning or heating devices, etc.) provided in the building. A sensor (eg, an electric meter, etc.) may be applicable.

Meanwhile, the communication network serves to connect the server 100 and the building sensor 200. That is, the communication network refers to a communication network that provides an access path so that the server 100 can transmit and receive data after accessing the building sensor 200 . Communication networks include, for example, wired networks such as LANs (Local Area Networks), WANs (Wide Area Networks), MANs (Metropolitan Area Networks), and ISDNs (Integrated Service Digital Networks), wireless LANs, wireless networks such as CDMA, Bluetooth, and satellite communication. However, the scope of the present invention is not limited thereto.

2 is a diagram showing the configuration of a server according to an embodiment of the present invention.

Referring to FIG. 2 , the main server 100 according to an embodiment of the present invention includes a communication module 110, a memory 120, a processor 130, and a database 140.

In detail, the communication module 110 provides a communication interface necessary to provide a transmission/reception signal between the server 100 and the building sensor 200 in the form of packet data in conjunction with a communication network. Furthermore, the communication module 110 may serve to receive a data request from the building sensor 200 and transmit data as a response thereto.

Here, the communication module 110 may be a device including hardware and software necessary for transmitting and receiving a signal such as a control signal or a data signal with another network device through a wired or wireless connection.

The memory 120 stores a program for managing energy of a building through artificial intelligence. Also, it performs a function of temporarily or permanently storing data processed by the processor 130 . Here, the memory 120 may include magnetic storage media or flash storage media, but the scope of the present invention is not limited thereto.

The processor 130, as a kind of central processing unit, controls the entire process of managing energy of a building through artificial intelligence. Each step performed by the processor 130 will be described later with reference to FIG. 3 .

Here, the processor 130 may include all types of devices capable of processing data, such as a processor. Here, a 'processor' may refer to a data processing device embedded in hardware having a physically structured circuit to perform functions expressed by codes or instructions included in a program, for example. As an example of such a data processing device built into hardware, a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application-specific integrated (ASIC) circuit), field programmable gate array (FPGA), etc., but the scope of the present invention is not limited thereto.

The database 140 may store state data collected from the building sensor 200 and information generated from the LSTM and building management model.

Although not shown in FIG. 2, some of the state data and data on information generated from the LSTM and building management model may be stored in a database (not shown) physically or conceptually separated from the database 140.

3 is an operational flowchart illustrating a process of managing energy of a building through artificial intelligence according to an embodiment of the present invention.

Referring to FIG. 3, the server 100 collects first state data generated during a specific period in the past from a plurality of building sensors 200 disposed in a building, inputs the first state data into an LSTM model, and Prediction state data for a predetermined period may be output (S110).

At this time, as described above, the first and second state data (the second state data may be state data measured in real time) and the predicted state data are data of the indoor and outdoor temperature of the building and energy consumption used for a predetermined time. can be

That is, the LSTM model may generate predicted state data by calculating expected indoor temperature against various energy usage based on the first state data.

At this time, as an optional embodiment, the server 100 corrects the indoor temperature calculated according to the size of each energy usage based on the outdoor temperature and climate, and may apply correction values according to weather and seasons outside the building. That is, since the amount of energy used to adjust the indoor temperature on a sunny day and a rainy day, or in summer and winter, respectively, may be different, the server 100 applies the correction value for these situations to the LSTM model to perform learning. can

In addition, as another optional embodiment, the server 100 determines the feasibility of the energy usage based on the indoor and outdoor temperatures for the section in which the energy usage rapidly increased among the first state data, and determines the validity of the energy usage for the section in which the energy usage rapidly increased according to the validity. can be applied For example, the server 100 recognizes a special situation, such as a case where only energy consumption increases in comparison to the temperature rise inside the building due to a failure of a heater installed in the building, and applies the correction value to the LSTM model to perform learning. there is.

Next, the server 100 sets the first state data and the predicted state data as state values and outputs the building management data as action values. A reinforcement learning-based building management model to be provided may be trained (S120). Reinforcement learning is a learning model that allows machine learning to be performed by providing a reward when an agent takes a specific action when the state value is set in the environment of a building.

In this case, the building management data may include change data of the power level supplied to the air conditioning and heating devices. That is, it may correspond to data for determining how much energy to supply to drive an air conditioner or a heating device for temperature management inside a building.

That is, even if the indoor temperature of the building is 23˚C, when the air conditioning and heating devices in the hallway and some offices are operated among the plurality of air conditioning and heating devices, and when the air conditioning and heating devices in all offices except the hallway are operated If the same indoor temperature is displayed, output the information on the side that consumes the lowest power level among the two cases (EX. Operate the air conditioning unit in offices 1, 2 and hallway 1 to 23˚C from 1:00 to 5:00 PM) A building management model can be trained to do this.

Since the present invention is a technology for managing building energy with optimized power, information on which air conditioning and heating devices are operated at what temperature and at what time of day in order to efficiently maintain building temperature with the least amount of power. is derived through machine learning, and based on this, it is possible to operate the building's air conditioning and heating devices.

As mentioned above, the building management model may be configured as one type of a preset reinforcement learning model. Therefore, the building management model generates building management data so that the indoor temperature is changed to the indoor temperature closest to a specific temperature while calculating the minimum energy consumption for real-time indoor temperature fluctuations.

That is, in step S120, in the course of learning the building management model based on the first state data, the predicted state data, and the building management data, building management data having an optimal energy consumption may be learned as a reward.

Finally, the server 100 may input the second state data received in real time from the building sensor 200 to the building management model, output the building management data, and operate the air conditioning and heating devices provided in the building (S130). ).

As an optional embodiment, if a plurality of air conditioning and heating devices are provided for each area of the building, the building management data generated by the building management model can individually adjust the power level applied to the air conditioning and heating devices for each area of the building. there is. This is because the temperature of each area of the building is not the same, so the temperature to be controlled by the air conditioning and heating devices installed in each area must be different from each other.

In addition, as another optional embodiment, when a specific temperature is set differently according to a change in time, the building management data generated by the building management model compares the power level applied to the air conditioning and heating device according to the change in time to a specific temperature can change accordingly. For example, it may be to prevent energy from being wasted by forcibly setting a temperature even in a space that is not used by people during work hours.

Meanwhile, in step S120, the server 100 may further set lighting management data including on/off times of a plurality of lights installed inside the building as a state value.

Through this, in step S130, the server 100 can further adjust the on/off of a plurality of lights at specific times through the building management data. Through this, it is possible to control energy by turning off lights that may be turned on unnecessarily.

Also, after step S130, the server 100 may update the LSTM model and the building management model based on the second state data and the building management data. At this time, the server 100 may correct the LSTM model based on an error generated by comparing the second predicted state data calculated by inputting the second state data into the LSTM model and the building management data.

On the other hand, as an optional embodiment, the server 100 determines that a problem has occurred in the air conditioning and heating device when the energy usage or real-time internal temperature is out of a predetermined threshold range even though the temperature of the building is adjusted based on the building management data. , a warning alarm may be provided to a preset manager terminal.

The above description of the present invention is for illustrative purposes, and those skilled in the art can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, the embodiments described above should be understood as illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and equivalent concepts should be construed as being included in the scope of the present invention. do.

100: server
200: building sensor

Claims (13)

In a method for managing energy of a building through reinforcement learning, performed by a server,
(a) Collecting first state data generated during a specific period in the past from a plurality of sensors disposed in the building, and inputting the first state data into an LSTM model to output predicted state data for a predetermined period from now step;
(b) setting the first state data and predicted state data as state values, outputting building management data as action values, and providing compensation as the expected energy consumption is lower when the indoor temperature of the building is a specific temperature value learning a reinforcement learning-based building management model; and
(c) inputting the second state data received in real time from the sensor to the building management model and outputting the building management data to operate air conditioning and heating devices provided in the building;
The first and second state data and predicted state data are data of indoor and outdoor temperatures of the building and energy consumption used for a predetermined time, and the building management data is data of power level change supplied to the air conditioning and heating device How to manage the energy of a building through reinforcement learning, which includes. According to claim 1,
Wherein the LSTM model generates the predicted state data by calculating the expected indoor temperature for various energy usage based on the first state data. According to claim 2,
The indoor temperature calculated according to the size of each energy consumption is corrected based on the outdoor temperature and climate, and the correction value according to the weather and season outside the building is applied through reinforcement learning. How to manage. According to claim 2,
Among the first state data, the validity of the energy usage is determined based on the indoor/outdoor temperature for the section in which the energy usage rapidly increased, and a correction value for the section in which the energy usage rapidly increased according to the validity is applied. How to manage energy in buildings through learning. According to claim 1,
The building management model calculates the minimum amount of energy used for real-time fluctuations in the indoor temperature, and generates the building management data so that the indoor temperature is changed to the indoor temperature closest to the specific temperature through reinforcement learning. How to manage your energy. According to claim 5,
The step (c) is
When a plurality of air conditioning and heating devices are provided for each area of the building, the building management data generated by the building management model individually controls the power level applied to the air conditioning and heating devices for each area of the building. How to manage the energy of a building through reinforcement learning. According to claim 5,
The step (c) is
If the specific temperature is set differently according to the change of time, the building management data generated by the building management model adjusts the power level applied to the air conditioning and heating device according to the change of time to the specific temperature. How to manage the energy of a building through reinforcement learning, which is to fluctuate. According to claim 1,
The step (b) is
Further setting lighting management data including on/off times of a plurality of lights installed inside the building as the state value,
The step (c) is
A method for managing energy of a building through reinforcement learning, further adjusting on/off of the plurality of lights at specific times through the building management data. According to claim 1,
After step (c) above
A method for managing energy of a building through reinforcement learning, wherein the LSTM model and the building management model are updated based on the second state data and the building management data. According to claim 9,
Building energy through reinforcement learning, which corrects the LSTM model based on an error generated by comparing the second predicted state data calculated by inputting the second state data into the LSTM model and the building management data How to manage. According to claim 1,
After step (c) above
Even though the temperature of the building has been adjusted based on the building management data, if the energy usage or the real-time internal temperature of the building is out of a predetermined threshold range, it is determined that a problem has occurred in the air conditioning and heating device, and a predetermined manager A method of managing the energy of a building through reinforcement learning, which is to provide a warning alarm to a terminal. In the device for managing the energy of a building through reinforcement learning,
A memory storing a program that provides a method for managing energy of a building through the reinforcement learning; and
A processor that executes a program stored in the memory to provide a method for managing energy of a building through the reinforcement learning,
The processor collects first state data generated during a specific period in the past from a plurality of sensors disposed in the building, inputs the first state data into an LSTM model, outputs predicted state data for a predetermined period after the present, , Reinforcement to set the first state data and predicted state data as state values, output building management data as action values, and provide compensation as the expected energy consumption is lower when the indoor temperature of the building is a specific temperature value Learning the learning-based building management model, inputting the second state data received in real time from the sensor to the building management model, and outputting the building management data to operate air conditioning and heating devices provided in the building,
The first and second state data and predicted state data are data of indoor and outdoor temperatures of the building and energy consumption used for a predetermined time, and the building management data is data of power level change supplied to the air conditioning and heating device A device for managing the energy of a building through reinforcement learning, which includes. A computer-readable storage medium on which a program for performing the method of managing energy of a building through reinforcement learning according to claim 1 is recorded.

Images