KR20180053453A - Energy management system, method for managing energy and method for predicting energy demand - Google Patents

Abstract

According to the present invention, an energy management system of a community with at least one building divided into one or more areas comprises: a community agent to manage building energy demand prediction data and building energy usage data for each building, and manage a community demand reaction incentive policy applied to the community; and a machine learning device to generate a building demand reaction incentive policy to be applied to each building by a community optimization machine learning model to include the building energy demand prediction data and the building energy usage data for each building and the community demand reaction incentive policy as input data and calculate a community demand management profit for the community to be the largest to send the building demand reaction incentive policy to the community agent. The community agent transmits the building demand reaction incentive polity to a building agent corresponding to each building.

Description

[0001] ENERGY MANAGEMENT SYSTEM, METHOD FOR MANAGING ENERGY AND METHOD FOR PREDICTING ENERGY DEMAND [0002]

The present invention relates to energy management techniques.

Recently, the energy shortage is worsening due to the rapid increase in energy demand. In order to overcome this power shortage, social costs are rising rapidly due to the addition of power generation and transmission and distribution facilities, and the supply of electric power is delayed. As a result, the government is also shifting its energy policy from supply-oriented to demand-oriented.

Demand management of electricity is a way to satisfy consumers' electricity demand patterns with minimizing costs by changing the pattern of power usage. Demand management of electricity can be divided into demand reaction and energy efficiency improvement. This power demand management can be effective when applied to buildings, homes, and factories.

Recently, various smart grid technologies such as renewable energy such as solar power, LED lighting, ESS (Enery Storage System), electric vehicle and smit meter have been introduced into buildings, Market demand for BEMS (Building Energy Management System), HEMS (Home Energy Management System) and FEMS (Factory Management System) technologies is increasing.

However, the conventional BEMS, HEMS, and FEMS have adopted a centralized integrated control method, which can not reflect the environmental difference in the detailed area.

For example, in applying the demand response incentive policy, conventional technologies have applied a single demand response incentive policy to the entire area (entire community area or entire building area). However, since there is an environmental difference in each sub-district and the tendency of users residing in the sub-district is also different, it is difficult to achieve high overall performance with a single demand response incentive policy. In addition, when the system increases the impossibility and responds to the demand reaction, there is a problem that user convenience is lowered.

In addition, since the conventional BEMS, HEMS, and FEMS adopt a centralized integrated control method, there is a problem that they can not flexibly cope with a failure of some of the components, and there is a problem that the system needs to be modified every time a new configuration is added.

In addition, this centralized integrated control method manages and controls the entire configuration with a single EMS algorithm, which controls different blocks of power consumption patterns with a single EMS (Energy Management System) algorithm, There was a problem that energy supply and demand could not be optimized for each pattern.

In addition, such a centralized integrated control method has a problem in that it can not effectively manage the local load variation caused by the building area.

In addition, this centralized integrated control method has a problem that the maintenance cost increases because the entire EMS needs to be corrected and recompiled every time a device is added or removed.

In addition, such a centralized integrated control method has a problem in that it does not sufficiently reflect the occupant status, energy use pattern, and convenience of residents.

In this context, an object of the present invention is, in one aspect, to provide an Energy Management System (EMS) technology that reflects the diverse (energy) environment of each building zone.

An object of the present invention is to provide an EMS technique which, on the other side, does not deteriorate the reliability of the entire system even if some failures occur.

It is an object of the present invention to provide, in yet another aspect, an EMS technology capable of plug & play of a device.

It is another object of the present invention to provide an EMS technique that reflects occupant status, energy usage pattern, and convenience of a resident.

It is an object of the present invention to provide an energy management system technology which, on the other side, allows the user convenience and energy cost to be determined optimally for each zone of the building.

It is an object of the present invention to overcome the problems of the conventional top-down approach (centralized integrated control method) and, in another aspect, to solve the problem of the energy management policy of the whole area by reflecting the independent energy management policy in each zone And to provide a technique for determining the quality of the image.

In order to achieve the above object, in one aspect, the present invention provides an energy management system for a community in which at least one building is subdivided into at least one zone, wherein the building energy demand forecast data for each building and the building energy A community agent for managing usage data and managing community demand response incentive policies applied to the community; And community optimization machine learning that includes the building energy demand forecast data, the building energy consumption data, and the community demand reaction incentive policy for each building as input data and calculates the community demand management profit for the community to be maximum And generating a building demand response incentive policy to be applied to each building through a model, and transmitting the building demand response incentive policy to the community agent, wherein the community agent transmits the building demand response incentive policy to a building agent corresponding to each building Energy management system.

In another aspect, the present invention provides a method of managing energy in a community in which at least one building is subdivided into at least one zone, the method comprising: managing building energy demand forecast data and building energy usage data for each building; Managing a community demand response incentive policy applied to the community demand response incentive policy; And community optimization machine learning that includes the building energy demand forecast data, the building energy consumption data, and the community demand reaction incentive policy for each building as input data and calculates the community demand management profit for the community to be maximum And generating a building demand response incentive policy to be applied to each building through a model.

As described above, according to the present invention, the EMS device can reflect the various (energy) environments according to the zones of the building, and even if some failures occur, the reliability of the entire system is not degraded, Plug & Play), and it is possible to reflect the residence status, energy use pattern, and convenience of residents. Further, according to the present invention, it is possible to provide an energy management system technology that enables optimal determination of user convenience and energy cost for each building area. According to the present invention, the problem of the conventional top-down method (centralized integrated control method) is solved and the energy management policy of the entire area can be determined by reflecting the independent energy management policy in each zone .

1 is a conceptual diagram of an energy management system according to an embodiment.
2 is a diagram showing a hierarchical structure of a community.
3 is a system configuration diagram of a community level.
4 is a system configuration diagram of a building level.
5 is a system block diagram of a zone level.
6 is a flowchart of an energy demand forecasting method according to an embodiment.
FIG. 7 is a block diagram of a first machine learning model for predicting occupancy.
FIG. 8 is a flowchart of a method for predicting occupancy number for each zone.
9 is a block diagram of a second machine learning model for predicting the energy demand by zone.
10 is a flow chart of a method for predicting energy demand by zone.
11 is a block diagram of a third machine learning model for predicting the energy demand of each building.
12 is a flowchart of a method for estimating energy demand for each building.
13 is a block diagram of a fourth machine learning model for forecasting community energy demand.
14 is a flowchart of a community energy demand forecasting method.
15 is a flow diagram of an energy optimization method according to one embodiment.
16 is a diagram illustrating a process of determining a community demand response incentive policy in the energy management system according to an exemplary embodiment.
17 is a diagram for explaining a process of optimizing at the community level according to an embodiment.
18 is a block diagram of a community optimized machine learning model according to an embodiment.
19 is a flow diagram of a method for generating a community optimal control scenario in accordance with one embodiment.
20 is a diagram for explaining a process of optimizing at a building level according to an embodiment.
21 is a block diagram of a building optimization machine learning model according to an embodiment.
22 is a flow diagram of a method for generating a building optimal control scenario in accordance with one embodiment.
23 is a block diagram of a zone optimization program for generating an optimal control scenario for each zone according to an embodiment.
24 is a flowchart of a method for generating an optimal control scenario for each zone in accordance with one embodiment.
25 is a flowchart of an energy management method according to another embodiment.
26 is a flowchart of a method of predicting energy demand according to yet another embodiment.
27 is an internal configuration diagram of xEMA.
FIG. 28 is a detailed configuration diagram of the layer management apparatus in FIG. 27;

Hereinafter, some embodiments of the present invention will be described in detail with reference to exemplary drawings. It should be noted that, in adding reference numerals to the constituent elements of the drawings, the same constituent elements are denoted by the same reference symbols as possible even if they are shown in different drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

In describing the components of the present invention, terms such as first, second, A, B, (a), and (b) may be used. These terms are intended to distinguish the constituent elements from other constituent elements, and the terms do not limit the nature, order or order of the constituent elements. When a component is described as being "connected", "coupled", or "connected" to another component, the component may be directly connected to or connected to the other component, It should be understood that an element may be "connected," "coupled," or "connected."

1 is a conceptual diagram of an energy management system according to an embodiment.

Referring to FIG. 1, the energy management system 100 includes a community system and a cloud system.

The community system includes at least one agent device 110, and at least one machine learning device 120 is included in the cloud system.

The agent apparatus 110 generates measurement data from various sensors and meters located in the space of the community, and transmits the measurement data to the machine learning apparatus 120 through the network. The agent device 110 includes a UI (User Interface) device, and transmits the configuration information acquired from the user to the machine learning device 120 through the UI device.

The machine learning apparatus 120 generates predictive data and optimization data through a machine learning model using measurement data and setting information received from the agent apparatus 110 as input data, and transmits the predictive data and the optimization data to the agent apparatus 110.

The agent apparatus 110 manages the energy of the community by using the predictive data and the optimization data received from the machine learning apparatus 120. [

In the community, the agent devices 110 can be hierarchically arranged, and the agent devices 110 disposed at each layer can independently transmit and receive information while communicating with the machine learning device 120.

The hierarchical structure of the community will be described with reference to Figs. 2 to 5. Fig.

2 is a diagram showing a hierarchical structure of a community.

2, a community energy management agent (CEMA) is located at the highest community level and at least one building energy management agent (BEMA) is connected to the community agent (CEMA) at the following building level Located. At least one zone agent (ZEMA) is connected to each building agent (BEMA) at the next zone level, and each zone agent (ZEMA) is located at the lowest sensor network level Energy devices - such as HAVC devices, lighting devices, and personalization devices - such as a personal computer (PC) - meters, sensors, and so on.

3 is a system configuration diagram of a community level.

Referring to FIG. 3, at least one building 310 is located in the community. The community may include community devices not belonging to the building 310, for example, a community may include a community load 320 - a streetlight, a traffic signal system, and so on. The community may include a community distributed power source 330, a community ESS (Energy Storage System) 340, a community EV (electric vehicle) charging station 350, etc. as a community device.

The community may include all of the above-described community devices-community load 320, community distributed power 330, community ESS 340, community EV charging station 350, or may include only at least one community device have.

The community has a community agent (CEMA) that manages the energy of the community as a whole. The community agent (CEMA) is associated with the building agent (BEMA) located in each building (310), and supervises each building agent (BEMA). The community agent CEMA is connected to the community devices 320, 330, 340, and 350 that do not belong to the respective buildings and acquires the state information of the community devices 320, 330, 340, and 350, And controls each of the community devices 320, 330, 340, and 350.

4 is a system configuration diagram of a building level.

Referring to FIG. 4, each building is subdivided into at least one zone 410.

The person building the energy management system according to one embodiment can subdivide each building into a plurality of thermal zones. Here, the thermal zone may mean a zone in which a thermal energy device - for example, a cooling / heating device - is independently controlled. Different thermal zones can be controlled in different thermal states. For example, the first thermal zone may be controlled at a room temperature of 23 degrees, and the second thermal zone may be controlled at a room temperature of 28 degrees.

The building may include building devices that do not belong to each zone 410. For example, the building may include a building load 420 - an elevator load, etc., a building distributed power source 430, a building ESS 440, a building EV charging station 450, and the like.

The building has a building agent (BEMA) that manages the energy of the entire building. The building agent (BEMA) is associated with a zone agent (ZEMA) located in each zone 410, and administers each zone agent (ZEMA). In addition, the building agent BEMA is connected to the building devices 420, 430, 440, and 450 that are not included in each zone, and acquires status information of the respective building devices 420, 430, 440, and 450, And controls each of the building devices 420, 430, 440, and 450.

5 is a system block diagram of a zone level.

Referring to FIG. 5, energy zones 312 and 314 and personalization equipment 316 may be located in each zone.

The personalization device 316 is an electric device that reflects the characteristics of the individual. For example, a personal computer (PC), a desk stand illumination, and the like belong to the personalization device 316. The energy devices 312 and 314 are electric devices not corresponding to the personalization device 316 and are mainly electric devices for controlling the environment of each zone. For example, a heating, ventilation, and Ari condition (HVAC) device 312 and a lighting device 314 belong to an energy device.

A sensor network can be configured with a plurality of sensors positioned in each zone. For example, a temperature sensor 520, a CO2 sensor 530, a humidity sensor 540, an illuminance sensor 550, and the like may be located in each zone.

The zone is located with the Zone Agent (ZEMA), which manages the energy of the entire area. The zone agent ZEMA may then obtain environmental data from the sensors 520, 530, 540, 550 and obtain the appliance energy usage data from the energy devices 312, 314 and the personalization device 316.

The energy management system uses the hierarchical structure of the community to predict energy demand.

6 is a flowchart of an energy demand forecasting method according to an embodiment.

Referring to FIG. 6, energy demand is predicted for each zone (S602).

The zone agent (ZEMA) located in each zone transmits the environmental data acquired from the sensor network and the energy consumption data of the equipment obtained from the electric equipment to the machine learning device and receives the zone energy demand forecast data.

When energy demand is predicted for each district, energy demand is predicted for each building (S604).

The building agent (BEMA) located in each building transmits the zone energy demand forecast data for each zone and the status information of the building devices not belonging to each zone to the machine learning device and receives the building energy demand forecast data.

Once energy demand is predicted for each building, the energy demand for the entire community is then predicted (S606).

The Community Agent (CEMA) transmits the building energy demand forecast data for each building and the status information of the community devices not belonging to each building to the machine learning device and receives the community energy demand forecast data.

This hierarchical structure of the energy management system enables forecasting of energy demand for each district.

We will explain how to predict energy demand at each level.

FIG. 7 is a block diagram of a first machine learning model for estimating the number of occupants, and FIG. 8 is a flowchart of a method for estimating the occupancy number for each section.

The energy management system can predict the number of occupants in each zone before anticipating the energy demand for the zone. The energy management system can predict the number of occupants first, thereby improving user convenience and energy efficiency for residents in each area.

Referring to FIG. 7, the first machine learning model for estimating the number of occupants can include temperature data, CO2 data, and personalization device energy usage data as input data. Temperature data, CO2 data, and personalization appliance energy usage data are all variables related to attendance. For example, an increase in the number of occupants in each zone leads to a higher CO2 concentration. The first machine learning model learns the relationship between the concentration of CO2 and the number of occupants, and predicts the number of occupants when the CO2 data is input at the prediction stage. Personalization energy consumption data can also be closely correlated with the number of occupants. For example, if your PC is in a high energy state, it may mean that the person who runs the PC is in that area. The first machine learning model learns the relationship between the energy consumption data of the personalization machine and the number of occupants, and when the personalization machine energy consumption data is input at the prediction stage, the number of occupants can be predicted through the data.

The first machine learning model may further use illumination data or wireless communication with the user terminal (e.g., Bluetooth, Wi-Fi, etc., which are near-field communication) as input data. To generate wireless communication data, a wireless communication device-for example, a Bluetooth device-may be located in each zone. Such a wireless communication device can determine whether a user is located indoors through a wireless communication with a user terminal (e.g., a mobile phone) and generate the wireless communication data.

The first machine learning model may use both temperature data, CO2 data, and personalization device energy usage data as input data, but depending on the embodiment, only some of the data may be used as input data. For example, the first machine learning model can calculate the number of lost workers using CO2 data as input data. In addition, according to the embodiment, temperature data and device energy usage data for the personalization device may be further included as input data.

The first machine learning model can compare the occupancy number prediction data with the actual data to generate error data and change the parameters and structure in the first machine learning model so that the error value in the error data becomes smaller. At this time, each zone may not include a sensor for measuring the number of occupants. In such a situation, the first machine learning model can utilize other information in order to acquire actual data. For example, a zone agent (ZEMA) may include a UI device, and the user input information may be periodically used as the actual data of the first machine learning model through the UI device.

The first machine learning model can generate not only the number of occupants but also occupancy patterns. The occupancy pattern may be a value representing the occupancy in accordance with the passage of time, and the first machine learning model can also generate this occupancy pattern. The machine learning apparatus can generate occupant estimation information including occupant number and occupant pattern through the first machine learning model, and generate the occupancy energy demand forecast data using the occupant estimation information. Hereinafter, the number of occupants, which is an example of the occupant estimating information, is described, but another occupant pattern may be used.

Referring to FIG. 8, the zone agent ZEMA transmits the first environment data to the machine learning device 120 (S802). Here, the first environmental data may be environmental data having a high degree of association with the number of occupants among the environmental data acquired from the sensor network installed in each zone, for example, temperature data and CO2 data.

The zone agent (ZEMA) transmits the energy usage data for the personalization device to the machine learning device 120 (S804).

The machine learning apparatus 120 calculates the number of occupants for each zone through the first machine learning model including the first environment data and the energy usage data for the personalization apparatus as input data, generates the occupant prediction data, (ZEMA) (S806).

If the number of residents is predicted, then the district energy demand forecast data is generated.

FIG. 9 is a block diagram of a second machine learning model for predicting energy demand by zone, and FIG. 10 is a flowchart of a method of predicting energy demand by zone.

Referring to FIG. 9, the second machine learning model for estimating the energy demand of each zone includes input data, occupancy data, environmental data, equipment energy consumption data of the electric equipment (energy equipment and personalization equipment) . Here, the physical information for each zone may include position information of each zone, area information of each zone, window information of each zone, and outer wall information of each zone. The device energy usage data may include real-time energy usage and energy usage pattern information. The environmental data may include temperature data, humidity data, illuminance data, CO2 data, and the like.

The second machine learning model learns the relationship between the input data and the zone energy usage, and generates the zone energy demand forecast data when the input data is input in the prediction step.

The second machine learning model may compare the zone energy demand forecast data with the actual data to generate error data and change the parameters and structure in the second machine learning model such that the error value in the error data is smaller. At this time, the actual data can be obtained through the device energy usage data.

Referring to FIG. 10, the machine learning apparatus 120 may estimate the number of occupants and generate the occupancy number prediction data as described with reference to FIGS. 7 to 8 (S1002)

The zone agent ZEMA manages the physical information about the zone, and periodically or aperiodically transmits the physical information to the machine learning device 120 (S1004).

Then, the zone agent ZEMA transmits the environmental data (S1006), and transmits the energy consumption data of the appliance to the electric appliance (S1008).

The machine learning apparatus 120 generates zone energy demand forecast data for each zone through a second machine learning model that includes occupancy number, environmental data, appliance energy usage data, and physical information as input data, (S1010).

The Zone Agent (ZEMA) manages the energy of each zone using the zone energy demand forecast data received for each zone.

The physical information for each zone may further include location information for each zone. The machine learning apparatus further acquires ambient data for the position of each zone from the zone agent (ZEMA) or another device (e.g., weather server), and further includes ambient data as input data in the second machine learning model To generate zone energy demand forecast data.

The zone agent ZEMA may store the environmental data, the energy usage data of the equipment, and the physical information in a local database (DATABASE), and periodically transmit the data stored in the local DB to the cloud DB associated with the machine learning apparatus 120 . The machine learning apparatus 120 may generate data on occupancy number or zone energy demand using data stored in the cloud DB.

The energy management system can collect energy demand forecasts for each zone and collect energy demand forecasts for the buildings.

FIG. 11 is a block diagram of a third machine learning model for estimating energy demand for each building, and FIG. 12 is a flowchart of a method for predicting energy demand for each building.

Referring to FIG. 11, a third machine learning model for predicting the energy demand of each building may include zone energy demand forecast data, state information on the building apparatus, and physical information of the building as input data. Here, the physical information of the building may include location information of each building, area information of each building, window information of each building, and outer wall information of each building. The state information on the building apparatus may include information such as, for example, the amount of power generated by the building distributed power source, the amount of charged building ESS, and the amount of power supplied to the electric vehicle connected to the building EV charging station.

The third machine learning model learns the relationship between input data and building energy consumption, and generates building energy demand forecast data when input data is input in the prediction step.

The third machine learning model can compare the building energy demand forecast data with the actual data to generate error data and change the parameters and structure in the third machine learning model so that the error value in the error data becomes smaller. At this time, the actual data can be obtained through energy usage data of each zone and energy usage data or energy supply data of the building apparatus.

Referring to FIG. 12, the machine learning apparatus 120 generates energy demand forecast data for each zone (S1202).

The building agent BEMA manages physical information on the building, and periodically or aperiodically transmits the physical information to the machine learning device 120 (S1204).

Then, the building agent BEMA acquires state information on at least one of the building load, the distributed building power of the building, the building ESS, and the building EV charging station, and transmits the state information to the machine learning apparatus 120 (S1206).

Then, the machine learning apparatus 120 generates a third machine learning model including, for each building, a zone energy demand forecast data for each zone, state information about the building apparatus, and physical information about each building as input data, Energy demand forecast data may be generated and transmitted to the building agent (BEMA) (S1208).

The energy management system can collect energy demand forecasts for each building and collect energy demand forecasts for the community.

FIG. 13 is a block diagram of a fourth machine learning model for community energy demand forecasting, and FIG. 14 is a flowchart of a community energy demand forecasting method.

Referring to FIG. 13, the fourth machine learning model for predicting the community energy demand may include building energy demand forecast data, state information on the community device, and physical information of the community as input data. Here, the physical information of the community may include location information of the community and the like. The state information on the community device may include, for example, information such as a power generation amount of the community distributed power source, a charging amount of the community ESS, and an electric power amount supplied to the electric vehicle connected to the community EV charging station.

The fourth machine learning model learns the relationship between input data and community energy usage, and generates community energy demand forecast data when input data is input in the prediction step.

The fourth machine learning model can compare the community energy demand forecast data with the actual data to generate error data and change the parameters and structure in the fourth machine learning model so that the error value in the error data becomes smaller. At this time, the actual data may be obtained through energy consumption data of each building and energy usage data or energy supply data of the community device.

Referring to FIG. 14, the machine learning apparatus 120 generates energy demand forecast data for each building (S1402).

Then, the community agent (CEMA) manages the physical information about the community and periodically or aperiodically transmits the physical information to the machine learning device 120 (S1404).

Then, the community agent CEMA obtains state information on at least one community device among the community load, the community distributed power, the community ESS, and the community EV charging station and transmits the state information to the machine learning device 120 (S1406).

Then, the machine learning apparatus 120 receives the community energy demand forecast data through the fourth machine learning model including the building energy demand forecast data for each building, the state information about the community apparatus, and the physical information about the community as input data And transmits the generated message to the community agent (CEMA) (S1408).

The energy management system can generate optimal control scenarios and energy management policies for each level using energy demand forecast data.

15 is a flow diagram of an energy optimization method according to one embodiment.

Referring to FIG. 15, the energy management system optimizes energy at the community level (S1502).

The community level may include a community device that does not belong to each building. In the community level optimization step, an optimal control scenario for this community device may be created. When creating an optimal control scenario for a community device, the energy management system can create an optimal control scenario so that the energy cost is minimized. For example, an energy management system can use energy demand incentives and real-time energy pricing information to control the energy usage of community devices over time or as a whole, thereby minimizing energy costs. Specifically, the energy management system schedules the use of community devices at low energy prices through real-time energy pricing information and also schedules the use of community devices to reduce energy usage during times of demand response incentives .

The energy management system can generate an optimal control scenario for direct control of the community device and an energy management policy to induce the maximum value of the specific objective function for the belonging buildings to be transmitted to the building agent of each building . Here, the energy management policy refers to a policy for guiding the value of a specific objective function to be maximized rather than a mandatory control signal. For example, an energy management policy may include a policy to guide the value of a particular objective function, such as a demand response incentive policy, to a maximum value, rather than a mandatory control signal such as load blocking or load scheduling.

The community agent can create an energy management policy that maximizes this objective function while using an objective function that maximizes the aggregate demand response incentive of the community. For example, the entity managing the community agent may be an energy demand management company that recruits and operates buildings that will participate in the demand response in the community unit. Such an energy demand management company may be a company established to maximize the total demand incentive of the community.

Alternatively, the energy demand management company may be a company established for the purpose of maximizing the profit through participating in the demand reaction. In this case, the objective function used by the community agent may be that the total amount of the demand management profit of the community becomes the maximum. Demand management revenues can be the amount that excludes building demand response incentives that redistribute community demand response incentives from the power trading market to each building. If the building demand reaction incentive is set too high, the revenue of the community demand response incentive may decrease. On the contrary, if the building demand response incentive is set too low, the demand response adaptation index of each building may be lowered and the profit of the community demand response incentive may decrease have. An energy management system can create an energy management policy based on these factors.

When the energy management policy is transmitted to the building agent, the energy management system optimizes the energy at the building level (S1504).

Building levels may include building units that do not belong to each zone, and in the building level optimization stage, an optimal control scenario for this building unit may be created. When creating an optimal control scenario for a building device, the energy management system can create an optimal control scenario so that the energy cost is minimized. For example, an energy management system can utilize demand response incentives and real-time energy pricing information to control the energy usage of a building device over time or as a whole, thereby minimizing energy costs. Specifically, the energy management system can schedule the use of building devices at low energy prices through real-time energy pricing information and also schedule the use of building devices to reduce energy usage during times of demand response incentives .

Here, the demand response incentive may be included in the energy management policy received from the community agent. Since the building agent generates the optimal control scenario considering the energy management policy received from the community agent, the building agent can maintain the indirect control relationship with the community agent although it is not directly controlled by the community agent.

The energy management system can generate an optimal control scenario for direct control of the building system and generate an energy management policy that guar- antees that the value of the specific objective function is maximized for the belonging zones and can be transmitted to the zone agent of each zone . Here, the energy management policy refers to a policy for guiding the value of a specific objective function to be maximized rather than a mandatory control signal. For example, an energy management policy may include a policy to guide the value of a particular objective function, such as a demand response incentive policy, to a maximum value, rather than a mandatory control signal such as load blocking or load scheduling.

The building agent can create an energy management policy that minimizes this objective function while using the objective function to minimize the energy cost of the building.

Energy management policies that a community agent sends to each building agent or energy management policies that a building agent sends to each zone agent can be set differently for each building agent or for each zone agent. For example, the building agent may differentiate the first energy management policy for the first zone agent and the second energy management policy for the second zone agent differently. When the demand response incentive policy is used in the energy management policy, the demand response load capacity capable of receiving the demand response incentive is determined. Therefore, the demand reaction load capacity is properly allocated to each building or each district, It is necessary to provide appropriate incentives for To this end, the community agent creates energy management policies differently for each building agent, so that the maximum demand management profit can be generated through each building agent. In addition, the building agent generates energy management policies differently for each zone agent so that the energy cost of the entire building is minimized.

When the energy management policy is sent to the zone agent, the energy management system optimizes energy at the zone level (S1506).

Since there is no lower level at the zone level, it may be important to create an optimal control scenario for all controllable devices (loads). On the other hand, since the energy management system generates the control scenario considering not only the energy cost but also the convenience of users residing in each zone, different levels and objective functions may be different.

The energy management system can generate an optimal control scenario by considering not only the energy management policy but also user setting information, indoor / outdoor environment information, energy usage pattern information per device, and occupant information, etc., when generating the optimal control scenario of the zone . Such information can be used to consider user convenience.

The following is an example of a case where demand response incentive policy is used as an energy management policy for convenience of understanding.

16 is a diagram illustrating a process of determining a community demand response incentive policy in the energy management system according to an exemplary embodiment.

Referring to FIG. 16, each zone agent (ZEMA) calculates zone energy demand forecast data and zone demand response load capacity and transmits them to the building agent (BEMA). Each Zone Agent (ZEMA) may not transfer zone demand response load capacity to the building agent (BEMA). The building agent (BEMA) may also manage load capacity estimates that are expected to participate in the demand response in each zone.

The building agent (BEMA) collects the zone energy demand forecast data and adds the energy demand forecast data for the building units not belonging to each zone to generate the building energy demand forecast data. Then, the building agent (BEMA) collects the zone demand response load capacity, and calculates the demand response load capacity of the building apparatus not belonging to each zone to generate the building demand response load capacity. Then, the building agent (BEMA) transmits the building energy demand forecast data and the building demand response load capacity to the community agent (CEMA). The building agent (BEMA) may not transmit the building demand response load capacity to the community agent (CEMA). The Community Agent (CEMA) can estimate and manage the load capacity expected to participate in the demand response in each building.

The Community Agent (CEMA) aggregates the building energy demand forecast data and adds the energy demand forecast data for the community devices that do not belong to each building to generate the community energy demand forecast data. Then, the community agent (CEMA) collects the building demand response load capacity, and calculates the demand response load capacity for the community devices not belonging to each building to generate the community demand response load capacity. The community demand response load capacity may be a value estimated through a machine learning model. In an embodiment where the community agent (CEMA) does not receive the load capacity to participate in the demand response from each building agent (BEMA), the community agent (CEMA) can estimate and manage the community demand response load capacity.

The Community Agent (CEMA) can bid on the demand response system using community energy demand forecast data and community demand response load capacity, and receive community demand response incentive policies as a result of the bidding.

Community demand response incentive policies can be demand response load capacity and pricing policies. For example, a community demand response incentive policy may be a pricing policy for the load capacity participating in the demand response and its load capacity. As a more specific example, the community demand response incentive policy may be the size of the load capacity and the price per KW of that load capacity. At this time, the price can be set differently in units of time. Building demand response incentive policies or zone demand response incentive policies can also be pricing policies for demand response load capacity and its demand response load capacity.

17 is a diagram for explaining a process of optimizing at the community level according to an embodiment.

Referring to FIG. 17, a machine learning device may generate an optimal control scenario for a community device through a community optimized machine learning model.

The community agent can acquire state information on at least one of community loads, community distributed power, community energy storage system (ESS), and community electric EV charging station, which are not belonging to the building, to the machine learning device.

The machine learning device may then generate an optimal control scenario for the community device using the state information for the community device and the community demand response incentive policy. At this time, the machine learning apparatus can generate an optimal control scenario so that the energy cost of the community apparatus is minimized.

Once the optimal control scenario generated at the machine learning device is sent to the community agent, the community agent can control the community devices according to this optimal control scenario.

On the other hand, the machine learning device has a community-optimized machine learning model that includes building energy demand forecast data, building energy consumption data, and community demand response incentive policy for each building as input data, You can create a policy. The building demand response incentive policy generated at this time may vary from building to building.

The community-optimized machine learning model can generate building demand response incentive policies so that certain optimization functions are maximized or minimized. The community optimized machine learning model can perform the machine learning according to the error data based on the difference between the predicted value and the measured value for the optimization function.

For example, the optimization function may be a function of community demand management revenues. At this time, the community optimization machine learning model can generate a building demand response incentive policy for each building to maximize the community demand management profit. However, demand response may not proceed in the direction expected by each building. In this case, the community optimized machine learning model can determine that the internal parameters are not optimized, and can perform the machine learning according to the error data based on the difference between the predicted value and the measured value.

As an example, the machine learning device may generate a predicted value for the community demand management revenue through a community optimized machine learning model and receive an actual value for the community demand management revenue from the community agent. Then, the machine learning apparatus can learn a community optimized machine learning model using error data corresponding to the difference between the predicted value and the measured value.

As another example, the machine learning device may generate a predicted value for the demand management profit of the community view for each building through the community optimized machine learning model, and receive the actual value of the demand response incentive profit for each building from the community agent. Then, the machine learning apparatus can learn a community optimized machine learning model using error data corresponding to the difference between the predicted value and the measured value.

The community-optimized machine learning model can calculate demand response load capacity of each building as a building demand response incentive policy. In this embodiment, the machine learning apparatus calculates the demand response load capacity And the community optimized machine learning model can be learned by using the error data according to the difference of the load capacity actually involved in the demand response for each building.

The community-optimized machine learning model can also generate demand response adaptation indices for each building. This demand response adaptation index can be used as an internal parameter of the community-optimized machine learning model and can be used to calculate the demand response load capacity and energy price to be allocated to each building.

On the other hand, the demand response adaptation index can be used for machine learning. The community optimized machine learning model further generates a demand response adaptive index prediction value for each building, and the machine learning apparatus can learn a community optimized machine learning model using error data according to the difference between the demand response adaptive index predicted value and the measured value .

18 is a block diagram of a community optimized machine learning model according to an embodiment.

Referring to FIG. 18, the community-optimized machine learning model includes real-time energy price information, a demand response incentive policy, a CO2 reduction incentive policy, state information on a community device, energy usage data for each building, .

The machine learning device may then generate a community device optimal control scenario through a community optimized machine learning model. And the machine learning device can generate building demand reaction incentive policy for each building through the community optimized machine learning model. Depending on the embodiment, the machine learning device may further generate a CO2 reduction incentive policy for each building.

The community optimization machine learning model can generate output values such that the optimization function is maximized or minimized.

Here, the incentive due to CO2 reduction can be calculated as a function of the renewable energy usage or the electric charge of the electric vehicle, and i, k, j can be determined by the choice of the demand management operator or the selection of the energy management system.

The community-optimized machine learning model can generate an optimization function value as an output and use it for machine learning by comparing it with measured data. In addition, the community optimization machine learning model may further include the demand response adaptation index for each building as input data. The demand response adaptation index can be generated by the community agent and transmitted to the machine learning device.

19 is a flow diagram of a method for generating a community optimal control scenario in accordance with one embodiment.

The Community Agent (CEMA) can manage demand response load capacity for the community and receive real-time energy pricing information, demand response incentive policies and CO2 reduction incentive policies from the power trading market. Here, the demand response incentive policy and the CO2 reduction incentive policy may be the value received by the community agent (CEMA) as a result of the bidding and the bid response by the community agent (CEMA) using the demand response load capacity for the community data and the community energy demand . The Community Agent (CEMA) can periodically or non-periodically update the demand response incentive policy and the CO2 reduction incentive policy through these bids.

Referring to FIG. 19, the machine learning device 120 generates community energy demand forecast data and transmits it to the community agent (CEMA) (S1902).

The community agent (CEMA) transmits the community energy demand forecast data to the power trading market (S1904), and transmits the demand response load capacity, which can participate in the demand response, to the power trading market in the community (S1906).

The power trading market - a server that manages the power trading market in hardware - can transmit the demand response incentive policy and the CO2 reduction incentive policy to the community agent (C901) in response to this information.

Demand Response The incentive policy can include information such as load capacity, time to participate in the demand response, and incentives when participating in the demand response. In addition, the CO2 reduction incentive policy may include incentive information for each new renewable energy.

The power trading market may transmit the real-time energy price information to the community agent (S1910).

The community agent (CEMA) transmits real-time energy price information to the machine learning device 120 (S1912), and transmits the demand response incentive policy and the CO2 reduction incentive policy (S1914).

The community agent CEMA transmits the state information on the community device to the machine learning device 120 (S1916), and transmits the energy usage data for each building to the machine learning device 120 (S1920).

The machine learning apparatus 120 is a community optimization apparatus that includes real-time energy price information, a demand response incentive policy, a CO2 reduction incentive policy, state information on a community device, energy usage data for each building, A demand response incentive policy for each building, a CO2 reduction incentive policy for each building, and a control scenario for the community device can be generated through the learning model and transmitted to the community agent (S1922, S1924).

The community agent (CEMA) can control the community devices according to the control scenario for the community device, and can transmit demand response incentive policies and CO2 reduction incentive policies for each building to each building agent (BEMA).

Meanwhile, the building agent manages the building demand response load capacity including the building energy demand forecast data including the zone energy demand forecast data for each zone and the zone demand response load capacity of each zone, and the building demand response Manage incentive policies.

The machine learning device is then applied to each zone through a building-optimized machine learning model that includes zone energy demand forecast data for each zone and building demand response incentive policies as input data and calculates energy costs for the building to be minimized Zone demand response incentive policies can be created and sent to building agents.

The building agent can then transmit the zone demand response incentive policy to the zone agent corresponding to each zone.

20 is a diagram for explaining a process of optimizing at a building level according to an embodiment.

Referring to FIG. 20, a machine learning apparatus can generate an optimal control scenario for a building apparatus through a building optimization machine learning model.

The building agent can acquire status information on at least one of the building load, the distributed building power of the building, the building ESS, and the building EV charging station that does not belong to the zone, and can transmit the status information to the machine learning device.

Then, the machine learning apparatus can generate the optimal control scenario for the building apparatus using the state information on the building apparatus and the building demand response incentive policy. At this time, the machine learning apparatus can generate an optimal control scenario so that the energy cost of the building apparatus is minimized.

Once the optimal control scenario generated in the machine learning device is sent to the building agent, the building agent can control the building devices according to this optimal control scenario.

On the other hand, the machine learning device has a zone-specific demand response incentive to be applied to each zone through a building-optimized machine learning model that includes zone energy demand forecast data, zone energy usage data for each zone, and building demand response incentive policy as input data You can create a policy. The zone demand response incentive policy generated at this time may vary from district to district.

The building optimization machine learning model can generate the zone demand response incentive policy so that the specific optimization function is maximized or minimized. And, the building optimization machine learning model can perform the machine learning according to the error data by the difference between the predicted value and the measured value for the optimization function.

For example, the optimization function may be a function of building energy cost. At this time, the building optimization machine learning model can generate the zone demand response incentive policy for each zone so that the building energy cost is minimized. However, demand response may not proceed in the direction expected by each zone. In this case, the building optimization machine learning model can determine that the internal parameters are not optimized, and can perform the machine learning based on the error data due to the difference between the predicted value and the measured value.

21 is a block diagram of a building optimization machine learning model according to an embodiment.

21, the building optimization machine learning model includes real-time energy price information, demand reaction incentive policy and CO2 reduction incentive policy for each building, status information for each building apparatus, energy usage data for each zone, and building energy Demand prediction data can be included as input data.

And the machine learning device can generate the demand response incentive policy for each zone, the CO2 reduction incentive policy for each zone and the control scenario for each building device through the building optimization machine learning model.

Here, the building optimization machine learning model can generate output values such that the second optimization function is minimized.

The second optimization function is a function that minimizes the energy consumption of the entire building, maximizes the incentive according to the demand response, and maximizes the incentive due to the CO2 reduction.

22 is a flow diagram of a method for generating a building optimal control scenario in accordance with one embodiment.

Referring to FIG. 22, the machine learning apparatus 120 transmits a demand response incentive policy and a CO2 reduction incentive policy for each building to a community agent (S 2202), and the community agent (CEMA) The demand response incentive policy and the CO2 reduction incentive policy can be transmitted (S2204).

The building agent (BEMA) provides real-time energy price information, demand reaction incentive policy and CO2 reduction incentive policy for each building, state information of the building apparatus, demand response load information of the building, (S2206, S2208, S2210, S2212, S2214).

In addition, the machine learning apparatus 120 is provided with real-time energy price information, demand reaction incentive policy and CO2 reduction incentive policy for each building, status information about each building apparatus, energy usage data for each zone, The incentive policy for each zone, the CO2 reduction incentive policy for each zone, and the control scenario for each building device can be generated and transmitted to the building agent (BEMA) through the building-optimized machine learning model including the input data , S2218).

The machine learning apparatus 120 calculates a demand response load capacity and a demand response adaptation index for each zone and generates a demand response incentive policy and a CO2 reduction incentive policy for each zone using demand response load capacity and demand response adaptation index for each zone You may.

FIG. 23 is a block diagram of a zone optimization program that generates an optimal control scenario for each zone according to an embodiment; and FIG. 24 is a flowchart of a method for generating an optimal control scenario for each zone according to an embodiment.

Zone agent (ZEMA) can manage demand response load information for each zone, receive demand response incentive policies and CO2 reduction incentive policies for each zone from the building agent (BEMA).

Referring to FIG. 23, a zone optimization program that generates an optimal control scenario for each zone includes real-time energy price information, a demand response incentive policy and a CO2 reduction incentive policy for each zone, occupancy information for each zone, energy devices operated in each zone, The energy consumption data (or energy usage pattern information of each device) of the personalization apparatus, and the regional energy demand forecast data for each zone may be included as input data.

As another example, the zone optimization program may include user setting information, indoor / outdoor environment information, energy usage pattern information for each device, occupant information, zone demand response load information, and zone energy demand forecast data as input data. The user setting information may include, for example, demand response priority setting information, conference schedule setting information, indoor setting information, and the like.

The machine learning apparatus can then generate an optimal control scenario for the energy and personalization equipment operating in each zone through a zone optimization program.

Further, the zone agent further manages the conference schedule setting information and the temperature or illumination setting frequency information of the energy device for each day, and the zone optimization program further includes conference schedule setting information and setting frequency information as input data, , And an optimal control scenario for indoor illumination and the like can be generated.

The zone optimization program may further include user setting information as input data. For example, the Zone Agent (ZEMA) includes a UI device, and may include information on priority of load devices to participate in a demand response, changeable meeting schedule information, desired indoor environment information (e.g., temperature, humidity, ) And input it to the zone optimization program as user setting information. Zone agent (ZEMA) without attendant input may generate user configuration information with default or previous settings.

The zone optimizer may create a control scenario so that the third optimization function is minimized (or maximized).

The comfort index can be a function of temperature, humidity, carbon dioxide concentration, fine dust concentration, and the number of energy device settings can be set, for example, by the setting of the HVAC The number of times of change or the number of times of changing the illuminance of the illuminating device. The proximity of the user-set value may be, for example, a close proximity between the room temperature set by the user and the actually measured room temperature. Alternatively, the proximity of the user-set value may be, for example, the proximity of the room temperature set by the user and the room temperature value to be controlled by the zone agent (ZEMA).

In addition, since the optimum control scenario includes the operation command value per unit, the zone agent (ZEMA) can control each device according to the operation command value. The control scenario for the building device and the control scenario for the community device can likewise include the operation command value, and each of the agents (BEMA, CEMA) can control the devices according to this operation command value.

Referring to FIG. 24, the machine learning apparatus 120 transmits a zone demand response incentive policy and a CO2 reduction incentive policy to the building agent (BEMA) (S2402). The building agent (BEMA) The demand response incentive policy and the CO2 reduction incentive policy can be transmitted (S2404).

Zone agent (ZEMA) provides real-time energy price information, demand response incentive policy and CO2 reduction incentive policy for each zone, number of occupants in each zone, energy energy consumption data of energy equipment and personalization equipment operating in each zone, Zone energy demand forecast data to the machine learning apparatus 120 (S2406, S2408, S2410).

Then, the machine learning apparatus can generate a control scenario for the energy device and the personalization apparatus operating in each zone through the zone optimization program and transmit the generated control scenario to the zone agent (ZEMA) (S2414).

25 is a flowchart of an energy management method according to another embodiment.

Referring to FIG. 25, an energy management system for managing energy of a community in which at least one building is divided into at least one or more zones acquires environmental data for each zone from a sensor network installed in each zone, The device energy usage data for the energy devices and the personalization devices operating in the zone can be obtained and the physical information for each zone can be obtained (S2502).

The energy management system generates occupancy information for each zone through a first machine learning model including temperature data and CO2 data and environmental energy usage data for the personalization apparatus among the environmental data as input data, The zone energy demand forecast data for each zone can be generated through a second machine learning model including the equipment energy usage data for the environmental data, the energy equipment, and the personalization equipment, and the physical information for each zone as input data (S2504 ). And, the energy management system can manage the energy of each zone by using the zone energy demand forecast data.

The energy management system acquires state information on at least one of the building load, the distributed building power, the building ESS (energy storage system), and the building EV (electric vehicle) charging station, Physical information about each building can be obtained (S2506).

In addition, the energy management system provides building energy demand through the third machine learning model, which includes, for each building, zone energy demand forecast data for each zone, state information about the building apparatus, and physical information about each building as input data, Prediction data can be generated (S2508).

The energy management system then obtains status information for at least one of the community load, the community distributed power, the community energy storage system (ESS), and the community EV (electric vehicle) charging station, (S2510). ≪ / RTI >

The energy management system can then generate community energy demand forecast data through a fourth machine learning model that includes building energy demand forecast data for each building, state information for community devices, and physical information about the community as input data (S2512).

The energy management system can then obtain demand response load information for the community and receive real-time energy price information, demand response incentive policies, and CO2 reduction incentive policies from the power management server (the server managing the power trading market) S2514).

And, the energy management system is a community-optimized machine learning system that includes real-time energy price information, demand response incentive policy, CO2 reduction incentive policy, status information on community devices, energy usage data for each building, The model can be used to generate a demand response incentive policy for each building, a CO2 reduction incentive policy for each building, and a control scenario for the community device (S2516).

And, the energy management system can obtain the demand reaction incentive policy and the CO2 reduction incentive policy (S2518).

The energy management system includes real-time energy price information, demand response incentive policy and CO2 reduction incentive policy for each building, status information for each building device, energy usage data for each zone, and building energy demand forecast data for each building, , The demand response incentive policy for each zone, the CO2 reduction incentive policy for each zone, and the control scenario for each building device can be generated (S2520).

And, the energy management system can acquire the demand reaction incentive policy and the CO2 reduction incentive policy (S2522).

The energy management system includes real-time energy price information, demand response incentive policy and CO2 reduction incentive policy for each zone, occupancy information for each zone, energy energy consumption data of energy equipment and personalization equipment operating in each zone, A zone optimization program that includes zone energy demand forecast data as input data may generate control scenarios for energy and personalization equipment operating in each zone (S2524).

26 is a flowchart of a method of predicting energy demand according to yet another embodiment.

Referring to FIG. 26, for a community in which at least one building is subdivided into at least one zone, the energy management system determines, for each zone through a first machine learning model that includes CO2 data for each zone as input data The number of occupants can be calculated (S2602).

Then, the energy management system can generate the zone energy demand forecast data for each zone through the second machine learning model including the number of occupants, the environmental data for each zone, and the energy usage data for the equipment as input data (S2604).

Then, the energy management system estimates the building energy demand forecast data for each building through the third machine learning model that includes the zone energy demand forecast data for each zone and the state information about the building devices not belonging to each zone as input data (S2606).

The energy management system generates the community energy demand forecast data for the community through the fourth machine learning model which includes the building energy demand forecast data for each building and the state information about the community devices not belonging to each building as input data (S2608).

Here, the first machine learning model may further include, as input data, temperature data for each zone and equipment energy usage data for personalization equipment located in each zone, and the second machine learning model may include physical information for each zone The third machine learning model further includes physical information about each building as input data, and the fourth machine learning model further includes physical information about the community as input data.

27 is an internal configuration diagram of xEMA.

Zone agent (ZEMA), building agent (BEMA), and community agent (CEMA) all have the same structure of xEMA. Each of the agents (ZEMA, BEMA, and CEMA) can set some configurations as active or non-active as necessary.

xEMA includes a local DB 2402 for storing data locally, a weather device 2404 for acquiring weather information or outdoor data, a demand reaction device 2406 for managing demand response information and processing a demand reaction command value, A real time price device 2408 for managing the energy price, a hierarchical management device 2410 for managing the hierarchical structure of the agent, a configuration management device 2412 for diagnosing the failure of the agent, a ML (Machine Learning device 2414, a sensor measurement device 2416 for acquiring environmental data from the sensor, an electric device measuring device 2418 for acquiring energy usage data from the electric device, An ESS control device 2422 for acquiring status information of the ESS and controlling the ESS, an EV charging station control device 2424 for acquiring status information of the EV charging station and controlling the EV charging station, A load management device 2426 for controlling and managing the UI, a UI device 2428 for providing a user interface, and the like.

FIG. 28 is a detailed configuration diagram of the layer management apparatus in FIG. 27;

The layer management apparatus 2410 includes a PEA module that manages node information of a parent EMS agent corresponding to an upper layer level, a CEA module that manages node information of a Child EMS agent that corresponds to a lower layer, and a Neighbor EMS agent An NEA module for managing node information, and an AR module for managing agent registration information.

2, when a building agent (BEMA) of a specific node fails, while managing the node information of a building agent (BEMA) corresponding to a lower level through the CEA module, the community agent (CEMA) Of the building agent (BEMA) in place of the function of the failed building agent (BEMA).

If the top-level Community Agent (CEMA) fails, one of the lower-level building agents (BEMA) can take over the Community Agent (CEMA) function.

At this time, the priorities are set in advance for other agents replacing the failed agent, so that the functions can be sequentially replaced according to this priority. This automatic failure recovery function is because each agent has the same structure and each agent operates in parallel without being operated in a sequential manner.

The embodiments of the present invention have been described above. According to this embodiment, the EMS device can reflect the various (energy) environment of each building area, and even if some failures occur, the reliability of the entire system is not degraded, and the plug & And it is possible to reflect the occupant status of the occupant, the pattern of energy use, and convenience.

In addition, according to this embodiment, it is possible to provide an energy management system technology for optimizing user convenience and energy cost for each building area. According to this embodiment, it is possible to solve the problem of the conventional top-down method (centralized integrated control method) and to determine the energy management policy of the entire area by reflecting the independent energy management policy in each zone do.

It is to be understood that the terms "comprises", "comprising", or "having" as used in the foregoing description mean that the constituent element can be implanted unless specifically stated to the contrary, But should be construed as further including other elements. All terms, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined. Commonly used terms, such as predefined terms, should be interpreted to be consistent with the contextual meanings of the related art, and are not to be construed as ideal or overly formal, unless expressly defined to the contrary.

The foregoing description is merely illustrative of the technical idea of the present invention, and various changes and modifications may be made by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be construed according to the following claims, and all technical ideas within the scope of equivalents should be construed as falling within the scope of the present invention.

Claims (10)

A community energy management system in which at least one building is subdivided into at least one zone,
A community agent that manages building energy demand forecast data and building energy usage data for each building and manages community demand response incentive policies applied to the community; And
A community optimization machine learning model that includes the building energy demand forecast data for each building, the building energy usage data, and the community demand response incentive policy as input data and calculates the community demand management profit for the community to be maximum And generating a building demand response incentive policy to be applied to each building through the network, and transmitting the building demand response incentive policy to the community agent,
And the community agent transmits the building demand response incentive policy to a building agent corresponding to each building. The method according to claim 1,
The machine learning apparatus generates a predicted value for the community demand management profit through the community optimized machine learning model, receives an actual value for the community demand management profit from the community agent, and calculates a difference between the predicted value and the measured value And learning the community optimized machine learning model using error data according to the community optimization machine learning model. The method according to claim 1,
Wherein the community agent bids the demand response system using the community energy demand forecast data and the community demand response load capacity and receives the community demand response incentive policy as a result of the bidding. The method according to claim 1,
The machine learning apparatus includes:
As an element of the community demand response incentive policy, energy management that learns the community-optimized machine learning model using error data according to the difference between the demand response load capacity calculated for each building and the load capacity actually participating in the demand response for each building system. The method according to claim 1,
The community optimized machine learning model further generates a demand response adaptation index prediction value for each building,
Wherein the machine learning apparatus learns the community optimized machine learning model using error data according to a difference between the demand response adaptation index predicted value and an actual measurement value. The method according to claim 1,
The building agent manages zone energy demand forecast data and zone energy usage data for each zone, manages the building demand response incentive policy applied to the building,
The machine learning apparatus includes a building optimization machine that includes the zone energy demand forecast data for each zone, the zone energy usage data, and the building demand response incentive policy as input data and calculates the energy cost for the building to be minimized A zone demand response incentive policy to be applied to each zone through a learning model is generated and transmitted to the building agent,
Wherein the building agent transmits the zone demand response incentive policy to a zone agent corresponding to each zone. The method according to claim 6,
The zone agent manages user setting information, indoor / outdoor environment information, energy usage pattern information for each device, and attendance information, manages the zone demand response incentive policy,
The machine learning apparatus includes the user setting information, the indoor / outdoor environment information, the energy usage pattern information for each device, and the occupancy information as input data, and the target function including the user convenience and energy cost reduction Generating an optimal control scenario for the intra-zone appliance through a zone optimization program that calculates the optimal control scenario, and transmitting the optimal control scenario to the zone agent. 8. The method of claim 7,
Wherein the user convenience is measured as a function of the number of energy device settings or the proximity of the energy device settings to the energy device measurements. CLAIMS What is claimed is: 1. A method of managing energy in a community in which at least one building is subdivided into at least one zone,
Managing building energy demand forecast data and building energy usage data for each building, and managing community demand response incentive policies applied to the community; And
A community optimization machine learning model that includes the building energy demand forecast data for each building, the building energy usage data, and the community demand response incentive policy as input data and calculates the community demand management profit for the community to be maximum To create a building demand response incentive policy for each building through
Lt; / RTI > 10. The method of claim 9,
Generating a predicted value for the community demand management profit through the community optimized machine learning model, receiving an actual value for the community demand management revenue, and receiving the actual value for the community demand management profit, using the error data according to the difference between the predicted value and the actual value, Learning the Optimized Machine Learning Model
Lt; / RTI >

Images