JP7352385B2 - Power management system, power management method and power management program - Google Patents

Description

The present invention relates to a power management system, a power management method, and a power management program that manage power in a plurality of user systems provided for each power consumption unit.

In recent years, as power transmission systems, in addition to conventional power transmission networks, systems called smart grids (next-generation power transmission networks) and smart communities that can control and optimize the flow of power from both the supply and demand sides have become popular. It is being done. BACKGROUND ART Conventionally, a power trading system is known in which surplus power at each consumer can be shared among a plurality of users through such a smart grid (for example, see Patent Document 1). In such smart grids, monthly meter reading operations can be automated and electricity usage status in HEMS (Home Energy Management System) can be monitored through smart meters that measure power consumption and power generation at each customer. managed.

The above-mentioned conventional power transmission systems and smart grids predict the future amount of power consumed by homes and business equipment, etc., and display the predicted power consumption on a monitor to visualize it, thereby encouraging consumers to save electricity. Development of systems that automatically control home appliances and other electrical equipment to optimize usage conditions is underway. By predicting future power consumption, it becomes possible to perform so-called peak cutting to avoid instantaneous or local increases in power consumption. On the other hand, as a method for adjusting power supply and demand, a method of combining renewable power sources such as solar power generation and wind power generation, centered on thermal power plants, as a distributed power source is becoming popular.

By the way, when a renewable power source is used as a distributed power source, for example, in a solar power generation device, output fluctuations occur due to short-term weather changes such as small clouds passing in front of the sun. As one of the power supply and demand adjustment technologies to compensate for fluctuations in the power supply and demand balance, there is a technology that utilizes distributed energy storage such as "storage batteries" that are interconnected under the power distribution network of the power system (for example, see Patent Document 1) ).

It is practical to reflect fluctuations in power supply due to changes in weather over a short period of time in the operation schedule (control information) for secondary batteries created by the power system control device, which is a host device, based on the power supply schedule. Therefore, in the system disclosed in Patent Document 1, by adjusting the time interval at which the battery control device obtains operation control information from an external device and the operation time interval for operation control processing for the battery, This makes it possible to adjust battery operation in response to changes in the actual power system status while making it less susceptible to communication failures.

JP 2018-130021 Publication

However, in the system disclosed in Patent Document 1 mentioned above, changes in the state of the power system at multiple consumers are collected and analyzed, and an operation schedule (control information) for each consumer is output. It is difficult to respond in real time to fluctuations that occur individually and in the short term at each consumer, and changes in the state of the power system have unique characteristics for each consumer. The problem with control is that it is difficult to provide detailed responses. As a result, in the system disclosed in Patent Document 1 mentioned above, power control is performed with some leeway in order to cover state changes unique to each consumer that occur individually and in a short period of time. Therefore, there is a need for wasted power supply or storage/discharge, and there is a certain limit to the effective use of energy resources.

Therefore, in view of the problems of the prior art, it is an object of the present invention to improve the accuracy of predicting future power consumption in small-scale facilities such as individual homes or small buildings on the local side in power control using smart grids. The purpose of the present invention is to provide a power management system, a power management method, and a power management program that allow each customer to operate independently and cover state changes unique to each customer that occur individually and in a short period of time.

In order to solve the above-mentioned problems, the present invention is a power management system that manages the power consumption of electrical equipment in a consumer, which is a power consumption unit, and is installed in a consumer and is a power management system that manages the power consumption of electrical equipment in a consumer, which is a unit of electricity consumption. A performance data generation unit that measures the amount of power and generates performance data, and a performance data generation unit that is installed on the electrical equipment side of the consumer rather than the performance data generation unit, and that analyzes temporal fluctuations in total power consumption measured for each consumer. , an individual equipment estimator that estimates the individual equipment operating within the consumer and its individual power consumption, and records the power status of the individual equipment in chronological order based on the estimation results from the individual equipment estimator. The data recording unit records the estimated history information recorded by the data recording unit, and the estimated history information recorded by the data recording unit is used to individually calculate the total power consumption measured in real time from the total power consumption in real time at the start of prediction. An individual power prediction unit that estimates the number of operating devices and predicts the transition of individual power consumption during the prediction period from the start of the prediction using power status parameters for the period corresponding to the prediction period; and a power prediction unit that predicts a change in total power consumption in a prediction period based on a predicted change in individual power consumption in the prediction period.

The present invention also provides a power management method for managing the power consumption of electrical equipment within a consumer, which is a power consumption unit, comprising:
(1) A performance data generation step in which a performance data generation unit provided within the consumer measures the amount of electricity generated or consumed within the consumer and generates performance data;
(2) A data recording unit installed on the electrical equipment side of the consumer rather than the actual data generation unit operates within the consumer by analyzing temporal fluctuations in total power consumption measured for each consumer. a data recording step of estimating the individual equipment and its individual power consumption, and recording estimated history information in which the state of the power supply of the individual equipment is recorded in chronological order based on the estimation result;
(3) The individual power prediction unit uses the estimated history information recorded in the data recording step to calculate the operation of the individual equipment at the start of prediction from the total power consumption in real time, based on the total power consumption measured in real time. an individual power prediction step of estimating the number of units and predicting the transition of individual power consumption in the prediction period from the prediction start point using the power state parameter of the period corresponding to the prediction period;
(4) A total power prediction step in which the power prediction unit predicts a change in total power consumption in a prediction period based on a change in individual power consumption in the prediction period predicted in the individual power prediction step.

In the above invention, the actual power consumption storage section stores the actual power consumption information regarding the actually consumed power, the actual power consumption information stored in the actual power consumption storage section, and the estimated history information recorded by the data recording section. The power prediction unit further includes a learning unit that causes the artificial intelligence of the power prediction unit to learn as training data .

In the above invention, the individual equipment estimating unit analyzes the power waveform and its temporal changes, and extracts the characteristics of frequency components and power fluctuation patterns, thereby identifying the operating individual equipment and its power consumption. It is preferable to estimate the power consumption.
Note that the above-described power management system and power management method according to the present invention can be realized by executing the power management program of the present invention written in a predetermined language on a computer. That is, the program of the present invention can be installed on an IC chip or memory device of a general-purpose computer such as a mobile terminal device, a smartphone, a wearable terminal, a mobile PC or other information processing terminal, a personal computer or a server computer, and executed on the CPU. Accordingly, it is possible to construct a system having each of the above-mentioned functions and implement the method according to the present invention.

Further, the power management program of the present invention can be distributed, for example, through a communication line, and can also be transferred as a package application that runs on a stand-alone computer by recording it on a computer-readable recording medium. can do. Specifically, the information can be recorded on various recording media such as a magnetic recording medium such as a flexible disk or a cassette tape, an optical disc such as a CD-ROM or a DVD-ROM, or a RAM card. According to the computer-readable recording medium on which this program is recorded, it becomes possible to easily implement the above-mentioned system and method using a general-purpose computer or a special-purpose computer, and it also becomes possible to store, transport, and store the program. Installation can be done easily.

According to the present invention, in power control using a smart grid, the accuracy of predicting the future power consumed by small units of facilities such as individual homes or small buildings is independently performed on the local side, and each demand It can cover changes in conditions unique to each consumer that occur individually and in a short period of time in a house.

1 is a conceptual diagram showing the overall configuration of a power management system according to an embodiment. 1 is a block diagram showing the equipment configuration of a user system according to an embodiment. FIG. FIG. 2 is a block diagram showing the internal configuration of a power control terminal according to an embodiment. FIG. 2 is a block diagram showing modules related to power management built in the power control terminal according to the embodiment. FIG. 2 is a block diagram showing an internal configuration of a learning section according to an embodiment. It is an explanatory diagram showing an outline of learning processing concerning an embodiment. FIG. 2 is a block diagram showing the internal configuration of a management server according to an embodiment. FIG. 3 is a flow diagram showing the operation of the power management system according to the embodiment.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of a power management system according to the present invention will be described in detail below with reference to the accompanying drawings. The embodiments described below are intended to exemplify a device etc. for embodying the technical idea of the present invention, and the technical idea of the present invention is based on the material, shape, structure, etc. of each component. The arrangement, etc. is not specified as below. The technical idea of this invention can be modified in various ways within the scope of the claims.

(Overview of power management system)
FIG. 1 is a diagram showing a network configuration of a power management system 1 according to this embodiment. The power management system according to this embodiment is a system that manages power in a plurality of user systems 4, 4, etc., which controls and manages power for each power consumption unit, and is installed in each user system 4, 4, etc. The smart meter 41 is a performance data generation unit, and a management server 2 is connected to the smart meter 41 via the Internet, a telephone line, a dedicated line, or the like.

In the power management system 1, each smart meter 41 measures the amount of power generated or consumed by each consumer during each power use period in each user system to generate performance data D1, and the management server 2 side stores the performance data. Power consumption within the user systems 4, 4, . . . is managed based on D1. As shown in FIG. 1, in this embodiment, the power management results and prediction results of each user system 4, 4, . . . are made available on the management server 2 side. In the power management system 1, a management server 2 and power control terminals 40 provided in each facility (power plant, PPS, consumer, power prosumer, etc.) are interconnected via a communication network 3. In each consumer, the smart meter 41 of each user system 4 is connected to the external power system.

The power control terminal 40 is composed of, for example, an information processing terminal equipped with a CPU, and is a device that centrally controls the power equipment of each facility such as a power plant, a PPS, a power prosumer, an aggregator, etc. in addition to each consumer. It is installed in the user system closer to the electrical equipment (load) than the smart meter 41, which is the performance data generation unit, and is connected to both the smart meter 41 and the distribution board 45 in the user system. The target equipment to be controlled by this power control terminal 40 includes a smart meter 41, a storage battery 42, a PV (Photovoltaics) 43, etc. included in a user system 4 installed in a facility such as a consumer or a power prosumer. This includes devices that manage power generation, storage, and power consumption.

Note that various devices to be controlled by this power control terminal 40 can be omitted as necessary. For example, the power consumption of the user systems 4, 4... is measured by the smart meter 41, but some consumers have power generation equipment and power storage equipment, while others have either power generation equipment or power storage equipment. There are also smart meters 41 that are equipped with only a smart meter 41 without any power generation or power storage equipment, and which only consume electricity. Electricity prosumers are also in the position of consuming electricity, but they can also be equipped with solar power generation and storage batteries and also be in the position of supplying electricity.

The communication network 3 is an IP network using the communication protocol TCP/IP such as the Internet, and includes various communication lines (telephone line, ISDN line, ADSL line, public line such as optical line, private line, WCDMA (registered trademark) In addition to third generation (3G) communication methods such as CDMA2000, fourth generation (4G) communication methods such as LTE, and fifth generation (5G) and later communication methods, Wifi (registered trademark), Bluetooth ( It is a distributed communication network constructed by interconnecting wireless communication networks such as (registered trademark). This IP network includes LANs such as intranets (corporate networks) and home networks based on 10BASE-T and 100BASE-TX.

(Configuration of each device)
Next, the configuration of each device will be explained. The term "module" used in the explanation refers to a functional unit that is composed of hardware such as devices and equipment, software that has the function, or a combination of these, and is used to achieve a predetermined operation. .

(1) User system 4
The user system 4 is the entire power equipment owned by consumers and power prosumers, and is also a unit of power consumption. A consumer is a contract unit related to the power equipment that receives and uses electricity, and includes high-voltage large-scale consumers with a contracted power of 500kW or more, high-voltage small-scale consumers with a contracted power of 50kW or more and less than 500kW, and less than 50kW such as general households. This includes low-voltage consumers. Further, the user system 4 may include power generation and power storage equipment. Examples of power generation equipment include solar power generation and wind power generation. This user system 4 includes a power control terminal 40 and a smart meter 41 as a performance data generation section. Furthermore, equipment that consumes power includes not only various home appliances, factory equipment, and office equipment, but also control devices in general such as power control devices (IoT devices).

The power control terminal 40 installed at each customer is installed closer to the electrical equipment (storage battery 42, PV 43, and other loads 441 to 44n) than the smart meter 41, which is the performance data generation unit, in the user system 4 of the customer. , it is possible to obtain the electric current, voltage, power waveform, frequency, etc. of the electric power connected to the distribution board 45 and distributed within the user system 4, and also to obtain the electric current, voltage, power waveform, frequency, etc. of each electric appliance and the user system 4 (customer). This is a device that actually controls the power generation and charging/discharging of the PV 43 and storage battery 42 located inside. Specifically, the power control terminal 40 is an information processing terminal equipped with a communication function and a CPU, and various functions can be implemented by installing an OS, firmware, and various application software. Once installed and executed, it functions as a power management unit. In addition to a personal computer, this information processing terminal can be realized by, for example, a smartphone or a dedicated device with specialized functions, and includes a tablet PC, a mobile computer, and a mobile phone.

The smart meter 41 is a performance data generation unit that comprehensively manages power generation, storage, and power consumption within the user system, which is a demand unit, and measures the power consumption of each customer within the user system 4 of the customer. In addition, it also controls and manages other equipment in the user system, such as storage and power generation by storage batteries and solar power generation, and measures the amount of power generated, stored, or consumed by consumers during each period of power use, and records actual results. D1 is generated and periodically sent to the management server 2 via the power control terminal 40. This performance data D1 is transmitted to the management server 2 through the communication network 3, telephone line, dedicated line, etc. In this embodiment, the smart meter 41 is used as the performance data generation unit, but the present invention is not limited to this. For example, the smart meter 41 is used as the performance data generation unit. This includes all electronic devices that are equipped with a control device such as a power control device (IoT device) and have a function of transmitting their own status as performance data to a communication network.

(2) Configuration of management server 2 The management server 2 is a server device managed and operated by a power management service provider, and as shown in FIG. , an external information database 21a, a user database 21b, a track record management database 21c, a learning information database 21d, an external information management section 24, and a data management section 26.

The communication interface 23 is a module that transmits and receives data to and from other communication devices through the communication network 3. In this embodiment, in order to provide this service, the communication interface 23 is a module that transmits and receives data to and from other communication devices, and in order to provide this service, it connects each power control terminal 40, smart meter 41, and external information source. 5.

The authentication unit 22 is a computer or software having the function of verifying the legitimacy of an accesser related to power management, and executes authentication processing based on a user ID that identifies the user. In this embodiment, the user ID and password are acquired from the accesser's terminal device through the communication network 3, and checked against the user database 21b to determine whether the accesser has the right or not, and whether the accesser is a subscriber. Check to see if it is.

The learning execution unit 25 is a module that executes AI machine learning provided in each user system 4 through the communication network 3, and in this embodiment includes a learning history management unit 25a and a system cooperation unit 25b.

The learning history management unit 25a is a module that generates learning history data that is a history of learning performed on each user system 4 side. The system cooperation unit 25b is a module that cooperates with the power control terminal 40 of each user system 4 to advance machine learning processing. The learning execution unit 25 manages machine learning using AI by coordinating with each power control terminal 40 through the system coordination unit 25b.

The external information management unit 24 is a module that collects information from external information sources 5 distributed over the communication network 3. Specifically, the external information management section 24 includes an information collection section 24a, a correlation extraction section 24b, and a correlation information provision section 24c.

The external information database 21a is a storage device that classifies and stores collected external information, and stores each external information in association with additional information such as its type, time information, and keywords. The user database 21b is a storage device that accumulates information regarding users of each consumer and businesses such as aggregators.

The track record management database 21c is a storage device that collects, accumulates, and manages track record data from parties involved in the delivery and reception of power, such as power plants, consumers, and aggregators. Each performance data received from each smart meter is accumulated in this performance management database and used for machine learning in the learning execution unit 25. The learning information database 21d is a storage device that records machine learning results for each customer.

The data management unit 26 is a module that generates training data for learning by collecting and analyzing performance data D1 and estimated history D2 from each customer, and the analysis results by this data management unit 26 are used for external information management. The correlation information analyzed by the unit 24 is input to the learning execution unit 25 and used for machine learning. Specifically, the data management section 26 includes a performance data collection section 26a and an estimated history collection section 26b.

(3) Power control terminal 40
Specifically, the power control terminal 40 includes a CPU 402, a memory 403, an input interface 404, a storage 401, an output interface 405, and a communication interface 406, as shown in FIG. Note that in this embodiment, each of these devices is connected via a CPU bus 400, and data can be exchanged with each other.

The memory 403 and the storage 401 are storage devices that store data in a recording medium and read out the stored data in response to requests from each device, such as a hard disk drive (HDD), solid state drive (SSD), It can be configured using a memory card or the like. In particular, in this embodiment, the storage 401 functions as a data recording unit that records estimation history information D2 that records the power status of individual equipment in chronological order based on the estimation result by the operating electric equipment identification unit 402d, which is an individual equipment estimation unit. It also functions as an actual power consumption storage unit that stores, inside the consumer, actual power consumption information D5 regarding the power actually consumed within the consumer.

The input interface 404 is a module that receives control signals from each piece of equipment installed in the user system, and the received control signals are transmitted to the CPU 402 and processed by the OS and each application. On the other hand, the output interface 405 is a module that outputs control signals to each facility installed in the user system. Each equipment installed in such a user system differs depending on the type of consumer or prosumer. For example, in a consumer, power consumption is measured by a smart meter 41, and power generation and storage is measured by solar power generation and Some have both power storage equipment, some have either solar power generation or storage battery equipment, and some have neither power generation nor power storage equipment. Furthermore, in the prosumer, power consumption is measured by a smart meter 41, and control signals for photovoltaic power generation (PV) 42 and storage battery 42 are input and output.

The communication interface 406 is a module that sends and receives data to and from other communication devices, and the communication methods include, for example, public lines such as telephone lines, ISDN lines, ADSL lines, and optical lines, private lines, and WCDMA (registered trademark). In addition to third generation (3G) communication methods such as CDMA2000, fourth generation (4G) communication methods such as LTE, and fifth generation (5G) and later communication methods, Wifi (registered trademark), Bluetooth ( wireless communication networks such as (registered trademark). The communication interface 406 is connected to the smart meter 41

The CPU 402 is a device that performs various arithmetic operations necessary to control each section, and virtually constructs various modules on the CPU 11 by executing various programs. An OS (Operating System) is started and executed on this CPU 402, and the basic functions of each power control terminal 40 are managed and controlled by this OS. Furthermore, various applications can be executed on this OS, and by executing the OS program on the CPU 402, various functional modules are virtually constructed on the CPU.

In this embodiment, by executing browser software on the CPU 402, it is possible to view information on the system and input information through the browser software. To be more specific, this browser software is a module for viewing web pages, and downloads HTML (HyperText Markup Language) files, image files, music files, etc. from the management server 2 through the communication network 3, and analyzes the layout. to display and play. This browser software also allows users to send data to a web server using forms, and to run application software written in JavaScript (registered trademark), Flash, Java (registered trademark), etc. Through this browser software, each user can use the power management service provided by the management server 2.

As shown in FIG. 4, the CPU 402 includes a total power usage acquisition unit 402a and power waveform information as an actual power consumption management module for detecting the operating status of the loads 441 to 44n in the consumer through the input interface 404. It includes an acquisition section 402b and a time variation calculation section 402c. Further, as a module for identifying electrical equipment in operation within a consumer and predicting power consumption, it includes an operating electrical equipment identification unit 402d, a likelihood estimation unit 402e, and a power prediction unit 402f. . Furthermore, a learning unit 402g is provided as a module for causing the power prediction unit 402f to perform machine learning. A charging/discharging control unit 402h is provided as a module that actually controls power generation and charging/discharging of the PV 43 and the storage battery 42 located in the consumer.

The total power consumption acquisition unit 402a is a module that is connected to the distribution board 45 via the input interface 404 and measures and acquires the current, voltage, etc. of power related to the power distributed within the user system 4, and the power waveform. The information acquisition unit 402b is a module that measures and acquires the waveform and frequency of power related to the power distributed within the user system 4 through the distribution board 45. The temporal variation calculation unit 402c is a module that calculates the temporal variation in the total power consumption measured by the total power consumption acquisition unit 402a. The total power usage acquisition unit 402a, the power waveform information acquisition unit 402b, and the time fluctuation calculation unit 402c constitute an actual power consumption management module that is an actual power consumption storage unit. Various information obtained by the unit 402b and the time variation calculation unit 402c are accumulated in the storage 401 as actual power consumption information D5 regarding the actually consumed power. This actual power consumption information D5 is used together with the estimated history information D2 as training data for learning the artificial intelligence of the power prediction unit 402f.

The operating electrical equipment identification unit 402d is installed closer to the electrical equipment (load) than the smart meter 41, which is the performance data generation unit, within the consumer, and is measured for each consumer by the smart meter 41, and calculated by the time fluctuation calculation unit 402c. This module functions as an individual equipment estimator that analyzes the temporal fluctuations in the total power consumption and estimates the individual equipment operating within the customer and the individual power consumption that is the power consumption thereof. Further, the operating electrical equipment identifying unit 402d stores an estimation history D2, which is a history of the generated estimation results, in the storage 401 in association with information representing a specific period. The likelihood estimated by the operating electrical equipment identification unit 402d is verified by the likelihood estimation unit 402e.

Here, the virtual individual devices may be each of the electrical devices actually existing in the house, or may be virtual electrical devices corresponding to the magnitude of the estimated decomposed power. For example, if the total power consumption measured by the total power measurement device 300 increases by 50W in a certain period, it is analyzed that an electrical device (there is no need to specify the specific device) with a power consumption of 50W has been operated. do. Further, if the total power consumption measured by the smart meter 41 decreases by 100W in a certain period, it is analyzed that the electrical equipment having the power consumption of 100W has stopped. That is, the virtual individual devices in this case refer to electrical devices with an estimated decomposition power of 50 W, and electrical devices with an estimated decomposition power of 100 W (hereinafter, these will be referred to as individual devices [50 W], individual devices [100 W], etc.). .

In addition, the specific period targeted for analysis is a period that covers the usage pattern of electrical equipment in individual homes (daily usage pattern, weekly usage pattern, etc.), and for example, 3 to 7 days per season. Examples include making it a daily cycle. In addition, the predetermined time unit for calculating power status parameters is one that divides one day into multiple time periods, such as morning, noon, and evening, taking into account the daily power usage pattern in the household. , night and four time periods. Note that this is just an example. For example, the predetermined time unit may be one hour or may be a shorter time unit. Alternatively, the predetermined time unit may be longer than one day, for example, several days. In short, the predetermined time unit can be determined depending on how much granularity one wants to predict future total power consumption.

Further, the operating electrical equipment identification unit 402d estimates the power-on state of each individual equipment using the equipment list T1 stored in the storage 401 with respect to the total power consumption acquired by the total power consumption acquisition unit 402a. In addition, the operating electrical equipment identification unit 402d applies the equipment list to the total power consumption acquired by the total power usage acquisition unit 402a, thereby determining the individual equipment at the time of prediction start from the total power consumption in real time. Estimate the number of machines in operation.

The power prediction unit 402f is a module that uses the learning data generated by the learning unit 402g to predict the power consumption of individual devices (hereinafter referred to as individual power consumption) in a desired future period (prediction period).

Specifically, the power prediction unit 402f obtains a device list T1 by estimating virtual individual devices and their individual power consumption, and estimates the number of operating individual devices using this device list T1. Then, based on this estimation result, a power state parameter representing the power state of the individual device is calculated in a predetermined time unit, thereby generating learning data regarding the power consumption of the individual device. Further, the power prediction unit 402f uses the estimated history information D2 recorded in the storage 401 to calculate the number of operating individual devices at the time of starting prediction from the total power consumption measured in real time. It also functions as an individual power prediction unit that predicts the transition of individual power consumption during the prediction period from the prediction start point using the power state parameters of the period corresponding to the prediction period.

Furthermore, the power prediction unit 402f uses a time fluctuation calculation unit 402c to calculate time fluctuations in the total power consumption and power waveforms measured in real time by the total power consumption acquisition unit 402a and the power waveform information acquisition unit 402b. Let them calculate. Then, by applying the device list T1 (virtual individual devices and their individual power consumption) stored in the storage 401 for the period corresponding to the prediction period, the total power consumption of the individual devices at the start of the prediction is calculated based on the total power consumption in real time. Estimate the number of machines in operation. Then, the power prediction unit 402f uses the power state parameter for a period corresponding to the prediction period to predict the transition of individual power consumption during the prediction period from the prediction start point.

Furthermore, in the present embodiment, the power prediction unit 402f predicts the transition in the total power consumption during the prediction period based on the transition in the individual power consumption during the prediction period predicted by the operating electrical equipment identification unit 402d. Specifically, the power prediction unit 402f sums up the individual power consumption at each time in the prediction period predicted by the operating electrical equipment identification unit 402d, thereby calculating the change in the total power consumption in the prediction period. Predict.

The charge/discharge control unit 402h controls power generation and charging/discharging of the PV 43 and the storage battery 42 through the output interface 405, and also has a function of notifying the user of the total power consumption predicted by the power prediction unit 402f, for example. . Alternatively, when the total power consumption predicted by the power prediction unit 402f exceeds a threshold, a warning message may be notified including information on which time period and to what extent the threshold will be exceeded. .

(4) AI Learning Here, the learning section 402g will be explained in detail. 5 and 6 are block diagrams showing the configuration of the learning section 402g. The learning unit 402g is a module that allows the deep learning recognition unit 402i, which is the artificial intelligence of the power prediction unit 402f, to learn to make appropriate judgments. Teacher data is generated from the estimation history D2, external information D3, and correlation information D4, and the deep learning recognition unit 402i is caused to learn based on this teacher data.

The learning unit 402g serves as a comparison unit that compares the estimated history, which is the determination result by the deep learning recognition unit 402i, with the actual data regarding the input performance data D1 and estimated history D2 for the deep learning recognition unit 402i. Fulfill. Specifically, the learning unit 402g uses the judgment result D77 regarding the same event (target customer, equipment, time of occurrence, etc.) input as training data to the deep learning recognition unit 402i, and the actual control result data at that time. The validity of the various options selected by the power prediction unit 402f when predicting the power can be verified by comparing the results with D1 and checking whether the comparison results match, and if so, which option was incorrect. It is verified inductively and fed back to the power prediction unit 402f.

The deep learning recognition unit 402i is a module that performs determination using so-called deep learning, and is used for functional verification as learning data (teacher data) for automatically setting various options for power management. Specifically, the deep learning recognition unit 402i analyzes the correlation between each estimation history D2 and the actual data D1, as well as the external information D3 and the correlation information D4, according to a predetermined deep learning algorithm, and analyzes the deep The learning recognition result (price prediction AI model) is set in the power prediction unit 402f.

In this embodiment, the algorithm implemented in the deep learning recognition unit 402i is a learning and recognition system that includes a multi-layered neural network, particularly three or more layers, and imitates the mechanism of the human brain. When data such as images is input to this recognition system, the data is propagated in order from the first layer, and learning is repeated in order in each subsequent layer. In this process, the features inside the image are automatically calculated.

This feature quantity is an essential variable necessary for solving a problem, and a variable that characterizes a specific concept. The deep learning recognition unit 402i also inputs performance data D1, estimation history D2, external information D3, and correlation information D4, extracts a plurality of feature points from these data hierarchically, and calculates the extracted feature points. Recognize patterns using hierarchical combination patterns. An outline of this recognition process is shown in FIG. As shown in the figure, the recognition function module of the deep learning recognition unit 402i is a multi-class classifier, in which a plurality of events are set and a feature vector 702 is an object that includes a specific feature point from among a plurality of objects. (Here, for example, "on/off of home appliance A") is detected. This recognition function module has an input unit (input layer) 707, a first weighting factor 708, a hidden unit (hidden layer) 709, a second weighting factor 710, and an output unit (output layer) 711.

At this time, a plurality of feature vectors 702 are input to the input unit 707. A first weighting factor 708 weights the output from input unit 707 . The hidden unit 709 non-linearly transforms the linear combination of the output from the input unit 707 and the first weighting coefficient 708 . A second weighting factor 710 weights the output from hidden unit 709. The output unit 711 calculates the identification probability of each class (for example, equipment used, usage status, etc.). Although three output units 711 are shown here, the present invention is not limited to this. The number of output units 711 is the same as the number of events that the pattern identifier can detect. By increasing the number of output units 711, the number of events that can be detected by the event identifier, such as recognizable device types, increases.

Specifically, the learning section 402g includes an item extraction section 402j, a teacher data creation section 402k, and a keyword processing section 402l. The item extraction unit 402j is a module that performs segmentation of character strings and numerical values in implementation data, external information, etc. to be recognized, in order to perform deep learning recognition. Specifically, in order to perform deep learning recognition, it is necessary to extract specific items and related keywords from the implementation data and estimation information, and corresponds to various events that affect power management.

The keyword processing unit 402l is a module that associates each keyword with a specific information source (character string or numerical value divided into regions). This association involves annotating related information (metadata) to a specific keyword associated with a specific information source, and tagging the metadata using a description language such as XML. A variety of information is described by dividing it into "meaning of information" and "content of information."

(Operation of power management system)
By operating the power management system described above, the power management method of the present invention can be implemented. FIG. 8 is a flow diagram showing the operation of the power management system. Note that the processing procedure described below is only an example, and each process may be changed as much as possible. Further, regarding the processing procedure described below, steps can be omitted, replaced, or added as appropriate depending on the embodiment.

As shown in FIG. 8, the user system 4 on the consumer side constantly measures the power generated, stored, or consumed within the system (S101). In this power measurement, temporal fluctuations in the power waveform are also measured and recorded at any time. On the other hand, on the management server side, in addition to measuring the power consumption on the consumer side, external information is constantly collected and classified (S201). Then, periodically or as soon as a predetermined amount of information is accumulated, the collected and classified external information is provided to the power control terminal 40 on each customer side (S202). Upon receiving this external information, the consumer side analyzes the total power consumption and estimates the individual power consumption (S102).

Specifically, the operating electrical equipment identification unit 402d analyzes temporal fluctuations in the total power consumption measured for each consumer and identifies individual equipment operating within the user system 4 and its power consumption. Estimate power consumption. Using this estimation history information, the number of operating individual devices at the start of prediction is estimated from the total power consumption measured in real time. At this time, the operating electrical equipment identification unit 402d analyzes the power waveform and its temporal changes measured by the smart meter 41, and extracts the characteristics of the frequency components and power fluctuation patterns, thereby identifying the individual operating electrical equipment. Estimate the equipment, its individual power consumption, and its duration.

Next, using the power state parameters for the period corresponding to the prediction period, the transition of individual power consumption in the prediction period from the prediction start point is predicted, and the individual power consumption in the prediction period predicted by the operating electrical equipment identification unit 402d is predicted. The change in total power consumption during the prediction period is predicted based on the change in (S103). Furthermore, based on the estimation result by the operating electrical equipment specifying unit 402d, estimation history information is recorded in which the state of the power supply of the individual equipment is recorded in chronological order (S104).

Based on the transition of the total power consumption predicted in this way, DR control is executed to control power generation and charging/discharging of the storage battery in the user system (S105). This power control uses detailed power predictions to determine which devices will consume how much power and for how long, based on estimated characteristics of individual devices and past usage patterns. The charging and discharging of solar power and the on/off of solar power generation are controlled. In addition, in this power control, so-called "peak cut" is carried out to avoid instantaneously exceeding the contracted power amount, and depending on the estimated type of individual equipment, which equipment is Predict how much power will be used continuously, and if a peak occurs, stop storing power during that time or discharge the stored power, etc. Avoid high power consumption.

In addition, as part of this power control, for example, if an estimated electrical appliance is in use and a peak in the contracted power is likely to occur, it is possible to refrain from using the appliance or change the start time of the appliance. A recommendation message such as a recommendation may be output. This recommendation message may be output by displaying it on a display inside the consumer, or outputting it as voice from a smart speaker, or the like. Peak cuts can also be achieved by using a power control terminal to perform automatic control using storage batteries.

This control result is collected by the smart meter 41 as performance data D1, and estimated history information is accumulated in the storage 401 (S106), and these performance data and estimated history information are transmitted from each user system to the management server 2. (S107), and collected by the management server 2 (S203). The performance data and estimated history information collected here are compared with external information to extract correlations between them and generate correlation information (S204).

Then, based on the collected performance data, estimated history information, correlation information, and external information, teacher data is generated, and machine learning for learning the artificial intelligence of the power prediction unit 402f is executed (S108). The learning results by this machine learning are updated as a learning history (S205), and are provided to the power control terminal 40 of each customer as teacher data, and are reflected in the next power forecast (S109).

(action/effect)
According to the present embodiment described above, the time fluctuations in total power consumption and the power waveform (frequency) are locally measured and analyzed, such as the user system of a consumer, and the individual devices operating within the user system are analyzed. and its power consumption, and predict the subsequent total power consumption based on the characteristics of each individual device. Therefore, according to the present embodiment, power control using a smart grid and prediction accuracy of future power consumed by small units of facilities such as individual homes or small buildings can be performed locally using a power control terminal. This can be done independently to cover changes in conditions specific to each consumer that occur individually and in a short period of time.

In addition, regarding the estimation history of individual devices and the prediction of total power consumption, we also analyze the correlation with external information and extract the characteristics of frequency components and power fluctuation patterns. By estimating individual power consumption, which is electric power, and reflecting it in the next prediction process through machine learning, it is possible to realize individual power control according to the power consumption pattern unique to each consumer. Prediction accuracy can be improved.

D1...Actual data D2...Estimation history D3...External information D4...Correlation information D77...Judgment result 1...Power management system 2...Management server 3...Communication network 4...User system 5...External information source 11...CPU
21a...External information database 21b...User database 21c...Achievement management database 21d...Learning information database 22...Authentication section 23...Communication interface 24...External information management section 24a...Information collection section 24b...Correlation extraction section 24c...Correlation information provision section 25 ...Learning execution unit 25a...Learning history management unit 25b...System cooperation unit 26...Data management unit 26a...Actual data collection unit 26b...Estimation history collection unit 40...Power control terminal 41...Smart meter 42...Storage battery 43...PV
400...CPU bus 401...Storage 402...CPU
402a...Total power consumption acquisition unit 402b...Power waveform information acquisition unit 402c...Time fluctuation calculation unit 402d...Operating electrical equipment identification unit 402e...Likelihood estimation unit 402f...Power prediction unit 402g...Learning unit 402h...Charging/discharging control unit 402i ...Deep learning recognition unit 403...Memory 404...Input interface 405...Output interface 406...Communication interface 441-44n...Load

Claims (6)

A power management system that manages the power consumption of electrical equipment within a consumer, which is a unit of power consumption,
a performance data generation unit that is provided within the consumer and generates performance data by measuring the amount of electricity generated or consumed within the consumer;
In the consumer, the electrical equipment is installed closer to the electrical equipment than the performance data generation unit, and the temporal fluctuations in the total power consumption measured on a consumer basis are analyzed, and the individual equipment operating in the consumer and its an individual device estimation unit that estimates individual power consumption, which is power consumption;
a data recording unit that records estimation history information in which the state of the power supply of the individual equipment is recorded in chronological order based on the estimation result by the individual equipment estimation unit;
Using the estimated history information recorded by the data recording unit, the total power consumption measured in real time is estimated from the total power consumption in real time, and the number of individual devices in operation at the start of the prediction is calculated. an individual power prediction unit that predicts a transition in individual power consumption during the prediction period from the prediction start point using a power state parameter for a period corresponding to the period;
a power prediction unit that predicts a change in total power consumption in the prediction period based on a change in individual power consumption in the prediction period predicted by the individual power prediction unit ;
an actual power consumption storage unit that stores actual power consumption information regarding actually consumed power;
a learning unit that causes the artificial intelligence of the power prediction unit to learn using actual power consumption information stored in the actual power consumption storage unit and estimated history information recorded by the data recording unit as training data;
A power management system characterized by: The individual equipment estimating unit analyzes the power waveform and its temporal changes, and extracts the frequency components and the characteristics of the power fluctuation pattern, thereby estimating the individual power consumption of the operating individual equipment and its power consumption. 2. The power management system according to claim 1 , wherein the power management system estimates . A power management method for managing the power consumption of electrical equipment within a consumer, which is a unit of power consumption,
a performance data generation step in which a performance data generation unit provided within the consumer measures the amount of electricity generated or consumed within the consumer and generates performance data;
A data recording unit installed within the consumer on a side closer to the electrical equipment than the performance data generating unit operates within the consumer by analyzing temporal fluctuations in total power consumption measured on a consumer-by-consumer basis. a data recording step of estimating the individual equipment and its individual power consumption, and recording estimated history information in which the state of the power supply of the individual equipment is recorded in chronological order based on the estimation result;
The individual power prediction unit uses the estimated history information recorded in the data recording step to calculate the number of operating individual devices at the time of starting prediction from the total power consumption measured in real time, based on the total power consumption measured in real time. an individual power prediction step of estimating the individual power consumption during the prediction period from the prediction start point using the power state parameters of the period corresponding to the prediction period;
a total power prediction step in which a power prediction unit predicts a change in total power consumption in the prediction period based on a change in individual power consumption in the prediction period predicted in the individual power prediction step ;
an actual power consumption accumulation step in which an actual power consumption accumulation section accumulates actual power consumption information regarding actually consumed power;
a learning step in which the artificial intelligence of the power prediction unit is trained using the actual power consumption information accumulated in the actual power consumption accumulation step and the estimated history information recorded in the data recording step as training data;
A power management method comprising: In the data recording step, the individual device estimator analyzes the power waveform and its temporal changes, and extracts the characteristics of frequency components and power fluctuation patterns, thereby estimating the operating individual devices and their power consumption. 4. The power management method according to claim 3 , further comprising estimating individual power consumption. In a power management system that manages the power consumption of electrical equipment within a consumer, which is a unit of power consumption,
A computer installed on the electrical equipment side rather than a performance data generation unit that is installed in the consumer and measures the amount of electricity generated or consumed in the consumer and generates performance data,
In the consumer, the electrical equipment is installed closer to the electrical equipment than the performance data generation unit, and the temporal fluctuations in the total power consumption measured on a consumer basis are analyzed, and the individual equipment operating in the consumer and its an individual device estimation unit that estimates individual power consumption, which is power consumption;
a data recording unit that records estimation history information in which the state of the power supply of the individual equipment is recorded in chronological order based on the estimation result by the individual equipment estimation unit;
Using the estimated history information recorded by the data recording unit, the total power consumption measured in real time is estimated from the total power consumption in real time, and the number of individual devices in operation at the start of the prediction is calculated. an individual power prediction unit that predicts a transition in individual power consumption during the prediction period from the prediction start point using a power state parameter for a period corresponding to the period;
a power prediction unit that predicts a change in total power consumption in the prediction period based on a change in individual power consumption in the prediction period predicted by the individual power prediction unit ;
an actual power consumption storage unit that stores actual power consumption information regarding actually consumed power;
a learning unit that causes the artificial intelligence of the power prediction unit to learn using actual power consumption information stored in the actual power consumption storage unit and estimated history information recorded by the data recording unit as training data;
A power management program that is characterized by functioning as a power management program. The individual equipment estimating unit analyzes the power waveform and its temporal changes, and extracts the frequency components and the characteristics of the power fluctuation pattern, thereby estimating the individual power consumption of the operating individual equipment and its power consumption. 6. The power management program according to claim 5 , having a function of estimating .