KR102503577B1 - System and Method for managing energy - Google Patents

Abstract

The present specification includes a data collection device for collecting environmental data of a plurality of energy usage devices; and an energy usage prediction device that predicts future energy usage of the plurality of energy usage devices based on the environmental data collected by the data collection device. At this time, the energy usage prediction device includes a pre-processing unit for generating new environmental data by pre-processing the environmental data collected by the data collection device according to cluster information; A correlation determination unit that determines whether a correlation between the new environmental data and one or more environmental data pools included in the environmental data pool set is equal to or greater than a first threshold correlation, and a plurality of energy using devices according to the determination result of the correlation determination unit. and an artificial intelligence model set management unit that manages an artificial intelligence model set including one or more artificial intelligence models predicting future energy usage.

Description

Energy management system and method {System and Method for managing energy}

The present invention relates to energy management systems and methods.

Recently, with the rapid increase in energy demand, the phenomenon of power shortage is aggravating. In order to solve this power shortage, as additional power generation and transmission and distribution facilities are installed, social costs are rapidly increasing, and the expansion of power supply is also being delayed. Accordingly, the government is also shifting from a supply-oriented energy policy to a demand-oriented energy policy.

Demand management of electricity is a method to meet stable electricity demand at minimum cost through changes in consumer's electricity usage pattern. Demand management of electricity can be divided into demand response and energy efficiency improvement. If such electricity demand management is applied to buildings, homes, and factories, it is judged that the effect of energy demand management can be significant.

In order to achieve such an energy demand management effect, the energy management system (EMS, Energy Management System) collects energy consumption information and status information of multiple energy-using devices (ex. lighting, electric vehicles, air conditioners, refrigerators) that use energy. It is necessary to collect and accurately predict the future energy consumption of multiple energy consuming devices based on this.

In order for the energy management system to predict future energy consumption as accurately as possible, it is necessary to secure many types of data/long-term data, and also to efficiently learn/relearn a model that predicts future energy consumption using the obtained data. there is

Embodiments provide an energy management system and method capable of efficiently performing re-learning of an artificial intelligence model for predicting future usage of a plurality of energy use devices.

In addition, the embodiments provide an energy management system and method capable of increasing the accuracy of an artificial intelligence model for predicting future usage of a plurality of energy use devices.

However, the technical problem to be achieved by the present embodiment is not limited to the technical problem as described above, and other technical problems may exist.

One embodiment includes a data collection device that collects environmental data of a plurality of energy usage devices; and an energy usage prediction device that predicts future energy usage of the plurality of energy usage devices based on the environmental data collected by the data collection device. At this time, the energy usage prediction device includes a pre-processing unit for generating new environmental data by pre-processing the environmental data collected by the data collection device according to cluster information; A correlation determination unit that determines whether a correlation between the new environmental data and one or more environmental data pools included in the environmental data pool set is equal to or greater than a first threshold correlation, and a plurality of energy using devices according to the determination result of the correlation determination unit. and an artificial intelligence model set management unit that manages an artificial intelligence model set including one or more artificial intelligence models predicting future energy usage.

Another embodiment is an energy management method performed by an energy management system, comprising: a data collection step of collecting environmental data of a plurality of energy using devices; and an energy usage prediction step of predicting future energy usage of a plurality of energy usage devices based on the environmental data collected in the data collection step. At this time, the energy usage estimation step may include a preprocessing step of generating new environmental data by preprocessing the environmental data collected in the data collection step according to cluster information; a correlation determination step of determining whether a correlation between the new environment data and one or more environmental data pools included in the environment data pool set is greater than or equal to a first threshold correlation; and an artificial intelligence model set management step of managing an artificial intelligence model set including one or more artificial intelligence models for predicting future energy usage of a plurality of energy use devices according to a result of the correlation determining step.

According to the energy management system and method according to the embodiments, re-learning of an artificial intelligence model for predicting future usage of a plurality of energy using devices can be efficiently performed, and future usage of a plurality of energy using devices can be predicted. It is possible to increase the accuracy of the artificial intelligence model for

1 is a configuration diagram of an energy management system according to an embodiment.
FIG. 2 is a configuration diagram of the energy usage estimation device of FIG. 1 .
3 is a flowchart illustrating an example of an operation of an energy usage estimation apparatus for selecting an environmental data pool to compare correlation with new environmental data.
4 is a flowchart illustrating an example of an energy consumption estimation apparatus operating according to a correlation between new environment data and a first environment data pool.
FIG. 5 is a conceptual diagram illustrating an operation of re-learning a first artificial intelligence model based on a first environmental data pool by the energy usage prediction device in FIG. 4 .
6 is a flowchart illustrating an example of an operation of determining when the energy usage prediction device relearns the first artificial intelligence model.
7 is a flowchart illustrating an example of an operation of an energy usage estimation apparatus according to a correlation between new environmental data and an environmental data pool included in an environmental data pool set.
8 is a flowchart illustrating an example of an operation of the apparatus for predicting energy usage according to the number of environmental data having a correlation with new environmental data equal to or greater than a second threshold correlation.
FIG. 9 is a conceptual diagram illustrating an operation of generating a second environment data pool and a second artificial intelligence model by the energy usage prediction device in FIG. 8 .
10 is a flowchart illustrating another example of an operation of an energy usage estimation apparatus according to a correlation between new environmental data and an environmental data pool included in an environmental data pool set.
FIG. 11 is a conceptual diagram illustrating an operation of generating a third environment data pool and a third artificial intelligence model of the energy usage prediction device in FIG. 10 .
12 is a flowchart of an energy management method according to an embodiment.
13 is a flowchart illustrating an energy usage estimation step of FIG. 12 .

Hereinafter, some embodiments of the present invention will be described in detail through exemplary drawings. In adding reference numerals to components of each drawing, it should be noted that the same components have the same numerals as much as possible even if they are displayed on different drawings. In addition, in describing the present invention, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present invention

In this case, the detailed description is omitted.

In addition, in describing the components of the present invention, terms such as first, second, A, B, (a), and (b) may be used. These terms are only used to distinguish the component from other components, and the nature, order, or order of the corresponding component is not limited by the term. When an element is described as being “connected,” “coupled to,” or “connected” to another element, that element is directly connected or connectable to the other element, but there is another element between the elements. It will be understood that elements may be “connected”, “coupled” or “connected”.

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

Referring to FIG. 1 , an energy management system 100 may include a data collection device 110 and an energy usage estimation device 120 .

The data collection device 110 may collect environmental data of a plurality of energy use devices.

An energy use device is a device that uses (consumes) energy and may also be referred to as a load. For example, the energy use device may be a cooling/heating air conditioner, lighting, electric heat, electric vehicle, etc. that uses energy in a home, commercial facility, factory, or the like.

The environmental data of the energy consuming device may include energy usage and state information (eg, internal/external environmental information of the energy consuming device, energy charge information, weather information, day of the week information).

For example, the data collection device 110 may collect environmental data of a plurality of energy use devices from environmental sensors, energy devices (loads), external systems, distributed power sources, electric vehicles (EVs), and the like. The data collection device 110 may collect information such as indoor/outdoor temperature, humidity, illuminance, and solar radiation from environmental sensors. The data collection device 110 may collect information such as electric heat usage from energy devices (loads). The data collection device 110 may collect energy cost (eg, power rate), meteorological data, and the like from an external system. The data collection device 110 may collect the amount of energy supplied from a distributed power source (eg, a solar power generator or a battery). The data collection device 110 may collect information about the amount of charge or discharge of the battery of the electric vehicle. The data collection device 110 may collect the number of occupants in the room.

The energy usage estimation device 120 may predict future energy usage of a plurality of energy usage devices based on the environmental data collected by the data collection device 110 . Hereinafter, a specific configuration of the energy usage prediction device 120 will be described with reference to FIG. 2 .

FIG. 2 is a configuration diagram of the energy usage estimation device of FIG. 1 .

Referring to FIG. 2 , the energy usage prediction device 120 may include a pre-processing unit 121, a correlation determination unit 122, and an artificial intelligence model set management unit 123.

The pre-processing unit 121 may generate new environmental data by pre-processing the environmental data collected by the data collection device 110 according to the cluster information. The new environmental data may include both energy consumption and state information included in the environmental data collected by the data collection device 110 for each cluster information.

At this time, the pre-processing unit 121 may also perform general pre-processing operations (ex. missing value, NaN processing, resampling, type conversion, interpolation) in the process of generating new environment data.

The cluster information may include information of a unit of period for collecting environmental data (eg, 1 minute/5 minutes/1 hour) and the number of times environmental data is collected (eg, 10 times/20 times/30 times). For example, the cluster information may indicate that environmental data is collected for 60 times per minute (in this case, environmental data is collected for a total of 1 hour (60 times * 1 minute)). As another example, the community information may be environmental data collected for 24 times in units of 5 minutes (in this case, environmental data was collected for a total of 2 hours (24 times * 5 minutes)).

Depending on how cluster information is set, the accuracy of future energy consumption predicted by an artificial intelligence model to be described later may vary. Cluster information may be determined in various ways according to the accuracy of future energy usage prediction or the efficiency of time/cost required for predicting energy usage. For example, as the period of collecting environmental data in the cluster information becomes shorter, future energy consumption can be predicted in a shorter time. Resources for estimating energy usage increase, which can increase overall cost.

For example, the cluster information may vary depending on when the energy usage prediction device 120 predicts the energy usage. As another example, the cluster information may vary according to a point in time when a charge for used energy is calculated.

For example, if energy bills are calculated based on instantaneous values extracted in 15-minute increments and billing is accumulated in 1-hour increments, cluster information is information indicating that environmental data was collected for 4 times in 15-minute increments can In this case, the time point at which the future energy usage amount is predicted may be a time point after a multiple value (eg, 24 hours) by 1 hour from the present.

The correlation determination unit 122 determines whether the correlation between the new environmental data generated by the preprocessing unit 121 and one or more environmental data pools (DBPs) included in the environmental data pool set 124 is greater than or equal to a first threshold correlation. can judge In this case, the first threshold correlation may be a preset value (ex. 0.8).

In this case, the environment data pool DBP is a data pool that stores environment data pre-collected by the data collection device 110 . Each environmental data pool (DBP) may store one or more environmental data of the aforementioned cluster information unit. The degree of correlation between the environmental data pool DBP and the new environmental data may mean the degree of correlation between previously collected environmental data stored in the environmental data pool DBP and new environmental data.

Meanwhile, in FIG. 2 , the case where the environment data pool set 124 is stored inside the energy usage prediction device 120 is described as an example, but the environment data pool set 124 is located outside the energy usage prediction device 120. It may be stored in a separate data storage area.

The artificial intelligence model set management unit 123 may manage the artificial intelligence model set 125 according to the determination result of the correlation determination unit 122 . The artificial intelligence model set 125 may include one or more artificial intelligence models (MDLs) that predict future energy usage of a plurality of energy using devices.

The artificial intelligence model (MDL) may receive all or part of environmental data of a plurality of energy using devices and output information on future energy consumption of the plurality of energy using devices. The artificial intelligence model (MDL) is, for example, a logistic regression, a support vector machine (SVM), a machine learning model such as a random forest, or a convolutional neural network (CNN), It may be a deep learning model such as RNN (Recurrent Neural Networks) or LSTM (Long Short-term memory model). The artificial intelligence model (MDL) may correspond to at least one of environmental data pools (DBPs) included in the environmental data pool set 124 .

On the other hand, the above-described correlation determination unit 122 compares the correlation between each environmental data pool (DBP) and new environmental data for all environmental data pools (DBPs) included in the environmental data pool set 124, for example. can

However, in this case, since the correlation with the new environmental data must be calculated for all environmental data pools (DBPs), a problem in that the time required for determining the correlation may increase.

Accordingly, the correlation determination unit 122 selects an environmental data pool (DBP) from among the environmental data pools (DBP) included in the environmental data pool set 124, and represents the selected environmental data pool (DBP) for each cluster information. By extracting a value and comparing it with a representative value for each cluster information of the new environmental data, it is possible to compare the correlation between the selected environmental data pool (DBP) and the new environmental data. Through this, the correlation determination unit 122 can reduce the time required to determine the correlation.

For example, the correlation determination unit 122 compares a representative value for each cluster information of each environmental data pool (DBP) included in the environmental data pool set 124 with a representative value for each cluster information of the new environment data, A pool of environmental data can be determined against which data and correlations can be compared. Hereinafter, this will be described in detail in FIG. 3 .

3 is a flowchart illustrating an example of an operation of an energy usage estimation apparatus for selecting an environmental data pool to compare correlation with new environmental data.

Referring to FIG. 3 , the correlation determination unit 122 of the energy usage prediction device 120 determines between a representative value for each cluster information of each environmental data pool included in the environmental data pool set and a representative value for each cluster information of new environmental data. Correlations may be compared (S310).

The correlation determination unit 122 determines whether a correlation between a representative value for each cluster information of the environmental data pool and a representative value for each cluster information of the new environmental data is equal to or less than a set threshold correlation (S320). If the correlation difference is less than or equal to the threshold correlation (S320-Y), the correlation determination unit 122 may add new environmental data to the selected environmental data pool (S330).

On the other hand, if the difference in correlation exceeds the threshold correlation (S320-N), whether the correlation determination unit 122 can create a new environment data pool instead of adding new environment data to the selected environment data pool. It can be determined (S340).

In this case, the representative value may be, for example, an average or a median representing characteristics of all data.

Hereinafter, when comparing the degree of correlation between the new environment data and the first environment data pool, which is one of the environment data pools included in the environment data pool set, an operation according to the degree of correlation will be described through a flowchart of FIG. 4 .

4 is a flowchart illustrating an example of an energy consumption estimation apparatus operating according to a correlation between new environment data and a first environment data pool.

Referring to FIG. 4 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 may determine a correlation between new environment data and the first environment data pool (S410).

The artificial intelligence model set management unit 123 determines whether a correlation between the new environment data and the first environment data pool is equal to or greater than a first threshold correlation (S420).

If the correlation between the new environment data and the first environment data pool is equal to or greater than the first threshold correlation (S420-Y), the artificial intelligence model set management unit 123 adds the new environment data to the first environment data pool ( S430), when the set re-learning condition is satisfied, the first artificial intelligence model corresponding to the first environment data pool may be re-learned (S440). In order to increase the accuracy of the first artificial intelligence model corresponding to the first environment data pool, the artificial intelligence model set management unit 123 retrains the first artificial intelligence model with new environment data having a high correlation with the first environment data pool. can be used for

On the other hand, when the correlation between the new environment data and the first environment data pool is less than the first threshold correlation (S420-N), the artificial intelligence model set management unit 123 adds the new environment data to the first environment data pool. do not run

FIG. 5 is a conceptual diagram illustrating an operation of re-learning a first artificial intelligence model based on a first environmental data pool by the energy usage prediction device in FIG. 4 .

Referring to FIG. 5 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 adds new environment data (NEW_DATA) to the first environment data pool (DBP_1), and the new environment data (NEW_DATA) is added. The first artificial intelligence model MDL_1 may be re-learned using the first environment data pool DBP_1.

At this time, the time point at which the artificial intelligence model set management unit 123 relearns the first artificial intelligence model MDL_1 using the first environment data pool DBP_1 to which the new environment data NEW_DATA is added varies. can be determined The artificial intelligence model set management unit 123 may increase accuracy by re-learning the first artificial intelligence model (MDL_1), but if re-learning of the first artificial intelligence model (MDL_1) is performed too often, re-learning may occur. This is because the cost of doing so may increase.

For example, the artificial intelligence model set management unit 123 may relearn the first artificial intelligence model MDL_1 at preset intervals. As another example, the artificial intelligence model set management unit 123 may re-learn the first artificial intelligence model MDL_1 when the size of data stored in the first environment data pool DBP_1 is greater than or equal to a preset critical data size.

As another example, the artificial intelligence model set management unit 123 may re-learn the first artificial intelligence model MDL_1 when the first artificial intelligence model MDL_1 satisfies a specific condition.

Hereinafter, an example of an operation of determining when the artificial intelligence model set management unit 123 relearns the first artificial intelligence model MDL_1 will be described with reference to FIG. 6 .

6 is a flowchart illustrating an example of an operation of determining when the energy usage prediction device relearns the first artificial intelligence model.

Referring to FIG. 6 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 determines whether the accuracy (MDL_1) of the first artificial intelligence model is less than a critical accuracy (S610). At this time, the accuracy of the first artificial intelligence model (MDL_1) can be calculated through the difference between the future energy consumption predicted by the first artificial intelligence model (MDL_1) and the energy consumption actually used by a plurality of energy consuming devices and the cost incurred accordingly. there is.

If the accuracy of the first artificial intelligence model MDL_1 is less than the critical accuracy (S610-Y), the artificial intelligence model set management unit 123 determines whether the size of the first environment data pool DBP_1 is greater than or equal to the critical data size (S610-Y). S620).

If the size of the first environment data pool DBP_1 is equal to or greater than the critical data size (S620-Y), the artificial intelligence model set management unit 123 performs the first artificial intelligence model MDL_1 based on the first environment data pool DBP_1. ) can be relearned (S630). That is, when re-learning is necessary because the accuracy of the first artificial intelligence model (MDL_1) is low, and the size of the environmental data that can be used for re-learning is sufficient, the artificial intelligence model set management unit 123 performs the first artificial intelligence model (MDL_1) Re-learning can be performed.

On the other hand, if the accuracy of the first artificial intelligence model (MDL_1) is greater than or equal to the critical accuracy (S610-N) or the size of the first environment data pool (DBP_1) is less than the critical data size (S620-N), the artificial intelligence model set management unit (123) does not relearn the first artificial intelligence model (MDL_1). This is because the current situation means that the accuracy of the first artificial intelligence model MDL_1 does not need to be improved or the accuracy of the first artificial intelligence model MDL_1 cannot be improved through relearning.

In this way, by performing the re-learning of the artificial intelligence model only at a specific time point, the artificial intelligence model set management unit 123 can efficiently re-learn the artificial intelligence model by preventing excessive costs for re-learning the artificial intelligence model. there is.

In FIGS. 4 to 6, the case of re-learning a previously generated artificial intelligence model has been described. 7 describes a case of generating a new artificial intelligence model instead of re-learning a previously generated artificial intelligence model.

7 is a flowchart illustrating an example of an operation of an energy usage estimation apparatus according to a correlation between new environmental data and an environmental data pool included in an environmental data pool set.

Referring to FIG. 7 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 may determine a correlation between new environment data and an environment data pool included in the environment data pool set 124 (S710). ).

The artificial intelligence model set management unit 123 determines whether a correlation between the new environment data and all environment data pools included in the environment data pool set 124 is less than a first threshold correlation (S720).

If the correlation between the new environment data and all environmental data pools included in the environment data pool set is less than the first threshold correlation (S720-Y), the artificial intelligence model set management unit 123 sets the new environment data and the environment data pool set. It may be determined whether a new second environment data pool including some of the environment data included in is capable of being created (S730). That is, when there is no environment data pool to which new environment data can be added among previously created environment data pools, the artificial intelligence model set management unit 123 determines whether a new environment data pool can be created, and if necessary You can create a new environment data pool and create a new artificial intelligence model for the newly created environment data pool.

Hereinafter, in FIG. 8 , the operation of the energy usage estimation apparatus 120 according to whether the second environment data pool can be generated, determined in step S730 , will be described through a flowchart.

8 is a flowchart illustrating an example of an operation of the apparatus for predicting energy usage according to the number of environmental data having a correlation with new environmental data equal to or greater than a second threshold correlation.

Referring to FIG. 8 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 searches environment data stored in an environment data pool included in the environment data pool set 124, and among them, new environment data and It is possible to determine the number of environmental data in which the degree of correlation of is greater than or equal to the second threshold degree of correlation (S810).

The artificial intelligence model set management unit 123 determines that the number of environment data whose correlation with new environment data is equal to or greater than the second critical level among the environmental data included in any one of the environmental data pools included in the environmental data pool set is critical environment data. It is determined whether the number is greater than or equal to (S820).

If the number of environmental data included in any one of the environmental data pools included in the environmental data pool set 124 has a correlation with new environmental data equal to or greater than the second threshold correlation, the number of environmental data is equal to or greater than the critical number (S820). -Y), the artificial intelligence model set management unit 123 determines that the amount of environment data is sufficient to create a new second environment data pool, creates a new second environment data pool, and adds it to the environment data pool set. It can be done (S830).

In this case, new environment data and environment data having a correlation degree greater than or equal to a second threshold correlation degree with the new environment data may be added to the second environment data pool.

The artificial intelligence model set management unit 123 may create a second artificial intelligence model corresponding to the generated second environment data pool and add it to the artificial intelligence model set (S840).

On the other hand, when the number of environmental data whose correlation with the new environmental data is equal to or greater than the second critical correlation among the environmental data included in any one of the environmental data pools included in the environmental data pool set 124 is less than the critical environmental data number ( S820-N), the artificial intelligence model set management unit 123 determines that the amount of environment data is not sufficient to create a new second environment data pool, and converts the new environment data to the environment data pool included in the environment data pool set 124. can be added to any of them.

For example, the artificial intelligence model set management unit 123 assigns the new environment data to an environment data pool having the highest correlation with the new environment data or a preset default environment data pool among the environment data pools included in the environment data pool set 124. can be added (S850). Then, the artificial intelligence model set management unit 123 may re-learn the artificial intelligence model corresponding to the environment data pool to which the new environment data is added (S860).

In this case, the default environment data pool may be, for example, an environment data pool generated first among the environment data pools included in the environment data pool set 124 or an environment data pool corresponding to the most accurate artificial intelligence model.

FIG. 9 is a conceptual diagram illustrating an operation of generating a second environment data pool and a second artificial intelligence model by the energy usage prediction device in FIG. 8 .

Referring to FIG. 9 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 has a correlation between new environment data (NEW_DATA) and new environment data among environment data included in the environment data pool set 124. A new second environment data pool DBP_2 may be created by collecting environment data having a second threshold correlation or higher. In this case, the value of the second critical correlation may be the same as or different from the aforementioned first critical correlation.

Also, the artificial intelligence model set management unit 123 may generate a second artificial intelligence model MDL_2 corresponding to the new second environment data pool DBP_2. The second artificial intelligence model MDL_2 may be learned through the second environment data pool DBP_2. The second artificial intelligence model MDL_2 may be included in the aforementioned artificial intelligence model set 125.

In the above, the case where the energy usage prediction device 120 adds new environmental data to a previously created environmental data pool or to a newly created environmental data pool has been described.

Hereinafter, a case where the energy usage prediction device 120 merges a plurality of environmental data pools will be described.

10 is a flowchart illustrating another example of an operation of an energy usage estimation apparatus according to a correlation between new environmental data and an environmental data pool included in an environmental data pool set.

Referring to FIG. 10 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 may determine a correlation between new environmental data and each environmental data pool included in the environmental data pool set 124. (S1010).

The artificial intelligence model set management unit 123 determines whether the number of environmental data pools having a correlation with new environment data equal to or greater than a first threshold correlation is two or more among the environmental data pools included in the environmental data pool set 124. (S1020).

The artificial intelligence model set management unit 123 determines that the correlation with the new environmental data is the first threshold when the number of environmental data pools having a correlation with the new environmental data equal to or greater than the first threshold is two or more (S1020-Y). Two or more environment data pools having a degree of correlation or higher may be merged into one third environment data pool (S1030). The artificial intelligence model set management unit 123 may generate a third artificial intelligence model corresponding to the third environment data pool (S1040). The artificial intelligence model set management unit 123 may reduce the number of environmental data pools and artificial intelligence models managed by merging the plurality of environment data pools and the plurality of artificial intelligence models that can be merged into one. Accordingly, the artificial intelligence model set management unit 123 can more efficiently manage the environment data pool and the artificial intelligence model.

On the other hand, the artificial intelligence model set management unit 123 assigns the new environment data to the environment data pool set when the number of environment data pools having a correlation with the new environment data equal to or greater than the first threshold correlation is less than two (S1020-N). It may be added to any one of included environmental data pools or may be added to a newly created environmental data pool (S1050). In this case, since there is no effect of merging multiple environmental data pools and multiple artificial intelligence models into one, the artificial intelligence model set adds new environmental data to any one of the environmental data pools included in the environmental data pool set as described above. or added to the newly created environment data pool.

FIG. 11 is a conceptual diagram illustrating an operation of generating a third environment data pool and a third artificial intelligence model of the energy usage prediction device in FIG. 10 .

Referring to FIG. 11 , the artificial intelligence model set management unit 123 of the energy usage prediction device 120 is configured to use new environment data (NEW_DATA) and an environment data pool to be merged (correlation with the new environment data is the first threshold correlation). A third environment data pool (DBP_3) may be created by merging environment data pools having a size of 10 or more degrees).

Also, the artificial intelligence model set management unit 123 may newly generate a third artificial intelligence model MDL_3 corresponding to the third environment data pool DBP_3.

The third environment data pool DBP_3 may be added to the environment data pool set 124 and the third artificial intelligence model MDL_3 may be added to the artificial intelligence model set 125 .

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

The energy management method 1200 performed by the energy management system 100 may include a data collection step S1210 of collecting environmental data of a plurality of energy using devices.

The energy management method 1200 may include an energy usage estimation step (S1220) of predicting future energy usage of a plurality of energy using devices based on the environmental data collected in the data collection step (S1210).

13 is a flowchart illustrating an energy usage estimation step of FIG. 12 .

The energy usage estimation step (S1220) may include a preprocessing step (S1310) of generating new environmental data by preprocessing the environmental data collected in the above-described data collection step (S1210) according to cluster information.

The energy usage estimation step (S1220) may include a correlation determination step (S1320) of determining whether or not a correlation between the new environmental data and one or more environmental data pools included in the environmental data pool set is greater than or equal to a first threshold correlation. can

In the correlation determination step (S1320), for example, a representative value of each environmental data pool included in the environmental data pool set is compared with a representative value of the new environmental data to determine an environmental data pool to compare the correlation with the new environmental data. can In this case, the representative value may be an average or a median value.

And, in the energy usage prediction step (S1220), an artificial intelligence model set including one or more artificial intelligence models for predicting the future energy usage of a plurality of energy using devices according to the result of the correlation determination step (S1320) is managed. An intelligent model set management step (S1330) may be included.

In the artificial intelligence model set management step (S1330), for example, when the correlation between the new environment data and the first environment data pool included in the environment data pool set is equal to or greater than a first threshold correlation, the new environment data pool is stored in the first environment data pool. Environment data may be added and a first artificial intelligence model corresponding to a first environment data pool among artificial intelligence models included in the artificial intelligence model set may be re-learned. At this time, the first artificial intelligence model is selected when the accuracy of the future energy usage of the plurality of energy using devices predicted by the first artificial intelligence model is less than the critical accuracy and the size of environmental data included in the first environmental data pool is greater than or equal to the critical data size. , may be re-learned based on the first environment data pool.

In the artificial intelligence model set management step (S1330), as another example, when the correlation between the new environment data and all environment data pools included in the environment data pool set is less than the first threshold correlation, the new environment data and the environment data pool set are selected. Based on this, it may be determined whether a second environment data pool including some of the new environment data and environment data included in the environment data pool set can be generated.

In this case, the artificial intelligence model set management step (S1330) includes the number of environmental data whose correlation with the new environment data is equal to or greater than the second threshold correlation among the environmental data included in any one of the environmental data pools included in the environmental data pool set. When is equal to or greater than the critical environment data number, a first environment data including new environment data and environmental data having a correlation with the new environment data of the second threshold correlation or higher among environmental data included in any one of the environment data pools included in the environment data pool set. 2 environment data pools may be created and added to the environment data pool set, and a second artificial intelligence model corresponding to the second environment data pool may be created and added to the artificial intelligence model set.

On the other hand, in the artificial intelligence model set management step (S1330), when the number of environment data having a correlation with the new environment data of the environment data included in the environment data pool set is equal to or greater than the second threshold correlation is less than the threshold number of environment data, Among the environment data pools included in the pool set, the new environment data may be added to an environment data pool having the highest correlation with the new environment data or a preset default environment data pool.

In the artificial intelligence model set management step (S1330), as another example, when the number of environmental data pools having a correlation with new environmental data equal to or greater than a first threshold correlation among the environmental data pools included in the environmental data pool set is plural, A plurality of environmental data pools having a correlation with the new environment data equal to or greater than a first threshold correlation may be merged into one third environment data pool, and a third artificial intelligence model corresponding to the third environment data pool may be generated.

In the above, even though all the components constituting the embodiment of the present invention have been described as being combined or operated as one, the present invention is not necessarily limited to these embodiments. That is, within the scope of the object of the present invention, all of the components may be selectively combined with one or more to operate.

Terms such as "comprise", "comprise" or "having" described above mean that the corresponding component may be inherent unless otherwise stated, and therefore do not exclude other components. It should be construed that it may further include other components. All terms, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs, unless defined otherwise. Commonly used terms, such as terms defined in a dictionary, should be interpreted as consistent with the meaning in the context of the related art, and unless explicitly defined in the present invention, they are not interpreted in an ideal or excessively formal meaning.

The above description is merely an example of the technical idea of the present invention, and various modifications and variations can be made to those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be construed according to the claims below, and all technical ideas within the equivalent range should be construed as being included in the scope of the present invention.

100: energy management system
110: data collection device
120: energy usage prediction device
121: pre-processing unit
122: correlation determination unit
123: artificial intelligence model set management unit

Claims (18)

a data collection device that collects environmental data of a plurality of energy usage devices; and
An energy usage prediction device for predicting future energy usage of the plurality of energy usage devices based on the environmental data collected by the data collection device;
The energy usage prediction device,
a pre-processing unit generating new environmental data by pre-processing the environmental data collected by the data collection device according to cluster information;
A correlation determination unit determining whether a correlation between the new environment data and one or more environmental data pools included in the environment data pool set is greater than or equal to a first threshold correlation; and
An artificial intelligence model set management unit for managing an artificial intelligence model set including one or more artificial intelligence models for predicting future energy consumption of the plurality of energy use devices according to the determination result of the correlation determination unit,
The environmental data includes energy usage and state information,
The cluster information indicates a unit of a period for collecting the environmental data and the number of times the environmental data is collected;
The artificial intelligence model set management unit,
When a correlation between the new environment data and a first environment data pool included in the environment data pool set is greater than or equal to the first threshold correlation,
The energy management system for adding the new environment data to the first environment data pool and relearning a first artificial intelligence model corresponding to the first environment data pool among artificial intelligence models included in the artificial intelligence model set.
According to claim 1,
The correlation determination unit,
By comparing the representative value for each cluster information of each environmental data pool included in the environmental data pool set with the representative value for each cluster information of the new environmental data, the correlation between the new environmental data and each environmental data pool is compared. energy management system.
◈Claim 3 was abandoned when the registration fee was paid.◈ According to claim 2,
The representative value is an average or median energy management system.
delete According to claim 1,
The artificial intelligence model set management unit,
When the accuracy of the future energy usage of the plurality of energy using devices predicted by the first artificial intelligence model is less than the critical accuracy and the size of the environmental data included in the first environmental data pool is greater than or equal to the critical data size, the first artificial intelligence model An energy management system for re-learning an intelligent model based on the first environmental data pool.
According to claim 1,
The artificial intelligence model set management unit,
When a correlation between the new environment data and all environmental data pools included in the environment data pool set is less than the first threshold correlation,
Based on the new environment data and the environment data pool set, it is determined whether a second environment data pool including some of the environment data included in the new environment data and the environment data pool set can be generated.
◈Claim 7 was abandoned when the registration fee was paid.◈ According to claim 6,
The artificial intelligence model set management unit,
When the number of environmental data included in any one of the environmental data pools included in the environmental data pool set having a correlation with the new environment data equal to or greater than a second threshold correlation is equal to or greater than the critical number of environmental data;
A second environmental data pool including environment data having a correlation with the new environment data of a second threshold correlation or higher among environment data included in any one of the new environment data and the environment data pools included in the environment data pool set. Create and add to the environment data pool set;
An energy management system for generating a second artificial intelligence model corresponding to the second environment data pool and adding it to the artificial intelligence model set.
◈Claim 8 was abandoned when the registration fee was paid.◈ According to claim 6,
The artificial intelligence model set management unit,
When the number of environment data having a correlation with the new environment data of the environment data included in the environment data pool set is equal to or greater than a second threshold correlation is less than the threshold number of environment data;
adding the new environment data to an environment data pool having the greatest correlation with the new environment data or a preset default environment data pool among the environment data pools included in the environment data pool set;
An energy management system for re-learning an artificial intelligence model corresponding to an environmental data pool to which the new environmental data is added.
According to claim 1,
The artificial intelligence model set management unit,
When the number of environment data pools having a degree of correlation with the new environment data equal to or greater than the first critical degree of correlation among the environmental data pools included in the environment data pool set is plural,
Energy management for merging a plurality of environmental data pools having a correlation with the new environmental data equal to or greater than a first critical correlation into one third environmental data pool, and generating a third artificial intelligence model corresponding to the third environmental data pool. system.
In the energy management method performed by the energy management system,
A data collection step of collecting environmental data of a plurality of energy usage devices; and
An energy usage prediction step of predicting future energy usage of the plurality of energy usage devices based on the environmental data collected in the data collection step;
The energy usage prediction step,
a pre-processing step of generating new environmental data by pre-processing the environmental data collected in the data collection step according to cluster information;
a correlation determination step of determining whether a correlation between the new environment data and one or more environment data pools included in the environment data pool set is greater than or equal to a first threshold correlation; and
An artificial intelligence model set management step of managing an artificial intelligence model set including one or more artificial intelligence models for predicting future energy usage of the plurality of energy use devices according to a result of the correlation determining step,
The environmental data includes energy usage and state information,
The cluster information indicates a unit of a period for collecting the environmental data and the number of times the environmental data is collected;
The artificial intelligence model set management step,
When a correlation between the new environment data and a first environment data pool included in the environment data pool set is greater than or equal to the first threshold correlation,
The energy management method of adding the new environment data to the first environment data pool and re-learning a first artificial intelligence model corresponding to the first environment data pool among artificial intelligence models included in the artificial intelligence model set.
According to claim 10,
In the correlation determination step,
By comparing the representative value for each cluster information of each environmental data pool included in the environmental data pool set with the representative value for each cluster information of the new environmental data, the correlation between the new environmental data and each environmental data pool is compared. How to manage energy.
◈Claim 12 was abandoned when the registration fee was paid.◈ According to claim 11,
The representative value is an average or median energy management method.
delete According to claim 10,
The first artificial intelligence model,
When the accuracy of the future energy usage of the plurality of energy usage devices predicted by the first artificial intelligence model is less than the critical accuracy and the size of the environmental data included in the first environment data pool is greater than or equal to the critical data size, the first environment data size is greater than the critical data size. An energy management method that is relearned based on a data pool.
According to claim 10,
The artificial intelligence model set management step,
When a correlation between the new environment data and all environmental data pools included in the environment data pool set is less than the first threshold correlation,
Based on the new environment data and the environment data pool set, it is determined whether a second environment data pool including some of the environment data included in the new environment data and the environment data pool set can be generated.
◈Claim 16 was abandoned when the registration fee was paid.◈ According to claim 15,
The artificial intelligence model set management step,
When the number of environmental data included in any one of the environmental data pools included in the environmental data pool set having a correlation with the new environment data equal to or greater than a second threshold correlation is equal to or greater than the critical number of environmental data;
A second environmental data pool including environment data having a correlation with the new environment data of a second threshold correlation or higher among environment data included in any one of the new environment data and the environment data pools included in the environment data pool set. Create and add to the environment data pool set;
An energy management method of generating a second artificial intelligence model corresponding to the second environment data pool and adding it to the artificial intelligence model set.
◈Claim 17 was abandoned when the registration fee was paid.◈ According to claim 15,
The artificial intelligence model set management step,
When the number of environment data having a correlation with the new environment data of the environment data included in the environment data pool set is equal to or greater than a second threshold correlation is less than the threshold number of environment data;
Adding the new environment data to an environment data pool having the highest correlation with the new environment data or a preset default environment data pool among the environment data pools included in the environment data pool set.
According to claim 10,
The artificial intelligence model set management step,
When the number of environment data pools having a degree of correlation with the new environment data equal to or greater than the first critical degree of correlation among the environmental data pools included in the environment data pool set is plural,
Energy for merging a plurality of environment data pools having a correlation with the new environment data equal to or greater than a first critical correlation into one third environment data pool, and generating a third artificial intelligence model corresponding to the third environment data pool management method.

Images