Power estimation system and method

Abstract

This system: extracts, from factor data and power system data of a point in a power system, sample data that is data including a value for a date and time at which a value of a predetermined factor related to a second component is less than or equal to a predetermined threshold; and generates, from the sample data, a first component estimation model for estimating a first component. The system: calculates a first estimated value of the second component from the power system data, the factor data, and the first component estimation model; and generates a second component model for estimating the second component, from the first estimated value of the second component and data of the predetermined factor related to the second component. The system calculates a second estimated value of the first component and a second estimated value of the second component, from: the power system data; the data of the predetermined factor related to the second component; and the second component estimation model.

Description

Power estimation system and method

[0001] The present invention relates generally to techniques for estimating power.

[0002] In the field of electric power, there are attempts to reduce transmission losses by controlling equipment in a power transmission system and / or a power distribution system (hereinafter referred to as the "power system") based on values ​​estimated from the active power and / or reactive power (hereinafter referred to as the "power demand") of tens to thousands of substations every tens of minutes to several hours. Therefore, it is important to estimate the power demand with high accuracy in order to control the equipment appropriately.

[0003] Electric power companies and other entities that operate power systems measure and understand the demand for electricity supplied from the power system at certain points in the power system, such as the demand end of a substation. Meanwhile, when photovoltaic (PV) power generation facilities or wind power generation facilities are connected to the power system, electricity generated by the PV and the wind power generation facilities flows into the power system. Therefore, electricity demand contains two or more components, such as a PV output component and a load component. Among the multiple components, some components are strongly correlated with certain factors, while other components are not strongly correlated with the factors.

[0004] Since each component, such as the PV output component and the load component, has a different tendency to fluctuate, it is important to estimate each component separately in order to estimate the power demand with high accuracy.

[0005] Furthermore, the multiple components that make up electricity demand can be considered to be composed of a second component that shows a relatively strong correlation with a single factor (e.g., the strongest correlation with the single factor) and a first component other than the second component. The "first component" is, for example, a load component or a component that is a mixture of the load component and the second component. The "load component" is the active power or reactive power consumed by the operation of a load such as an electrical appliance, and shows a correlation with factors such as temperature and the type of calendar day (e.g., weekday or holiday). The "second component" is, for example, a PV output component that shows a strong correlation with solar radiation (an example of a single factor) or a wind power output component that shows a strong correlation with wind speed (an example of a single factor). Because the second component is strongly correlated with a single factor, using information on a single factor that shows a strong correlation tends to enable more accurate estimation of the second component than the first component. The "PV output component" is the active or reactive power of the power generated by the PV, and the "wind power output component" is the active or reactive power of the power generated by the wind power generation facility. Examples of the first component and second component include the following (x) to (z): (x) Active power of the load component: Active power of the PV output component (y) Active power of the load component: Active power of the wind power generation component (z) Mixed component of the active power of the load component and the active power of the wind power generation component: Active power of the PV output component

[0006] Although electric power companies can measure the electricity demand that is the composite of each component, it is difficult to directly grasp each component, such as the PV output component and the load component, so it is necessary to estimate each component from the measured electricity demand.

[0007] Known technologies for estimating components such as PV output and load included in measured power demand are disclosed in Japanese Patent Laid-Open No. 2014-200167 (Patent Document 1) and Japanese Patent Laid-Open No. 2016-197948 (Patent Document 2).

[0008] Patent Literature 1 discloses a technology for generating a conversion formula (reactive power conversion formula) for deriving active power from reactive power based on measured active power, reactive power, and solar radiation, and estimating the active power output by PV and the active power consumed by a load using the generated reactive power conversion formula. Patent Literature 2 discloses a technology for estimating a conversion formula (solar radiation conversion formula) for deriving PV output from solar radiation based on measured active power and solar radiation, and estimating the active power output by PV and the active power consumed by a load using the generated solar radiation conversion formula.

[0009] JP 2014-200167 A JP 2016-197948 A

[0010] In Patent Document 1, when there is no PV output, the PV output and actual load are estimated based on the assumption that there is a linear relationship between the measured active power and reactive power. However, in reality, the value of the measured reactive power fluctuates depending on the settings of equipment that adjusts reactive power, such as a reactive power compensator, so this assumption does not hold true and the estimation accuracy deteriorates.

[0011] In addition, in Patent Document 2, learning data for a regression equation is generated from measured active power and solar radiation, parameters of a solar radiation conversion equation are calculated using the regression equation, and a model for estimating the active power of PV output is estimated. However, since the measured active power includes a load component, errors occur in the estimated parameters, deteriorating the estimation accuracy.

[0012] The present invention has been made in consideration of the above points, and has an object to accurately estimate each component contained in measured power demand.

[0013] The system includes power system data and factor data for a point in a power system. The power system data includes a time series of measured values ​​of power demand at the point. The factor data includes a time series of factor values ​​for a predetermined factor, of which a first component and a second component are included in the power demand at the point and whose correlation with the second component is stronger than that with the first component. The system extracts sample data from the power system data and factor data for the point, the sample data including values ​​for dates and times when the value of the predetermined factor related to the second component is equal to or less than a predetermined threshold, and generates a first component estimation model that estimates the first component from the sample data. The system calculates a first estimate of the second component from the power system data, the factor data, and the first component estimation model, and generates a second component model that estimates the second component from the data of the predetermined factor related to the second component and the first estimate of the second component. The system calculates a second estimate of the first component and a second estimate of the second component from the power system data, the data of the predetermined factor related to the second component, and the second component estimation model.

[0014] According to the present invention, each component contained in the measured power demand can be estimated with high accuracy.

[0015] Problems, configurations, and effects other than those described above will become apparent from the following description of the embodiments.

[0016] FIG. 1 is an overall configuration diagram of a grid management system. FIG. 2 is a conceptual diagram showing a data flow of the grid management system. FIG. 3 is a functional configuration diagram of each device constituting a power estimation system. FIG. 4 is a functional block diagram showing an embodiment of a first component model identification process. FIG. 5 is a conceptual diagram showing an embodiment of power grid data. FIG. 6 is a conceptual diagram showing an embodiment of factor data. FIG. 7 is a conceptual diagram showing an embodiment of power estimation data. FIG. 8 is a flowchart showing the processing procedure of the power estimation system. FIG. 9 is a conceptual diagram showing the effects of Comparative Example 1, Comparative Example 2 and this embodiment. FIG. 10 is a functional block diagram showing an embodiment of a first component model identification process. FIG. 11 is a functional block diagram showing an embodiment of a power estimation system. FIG. 12 is a schematic diagram of the processing procedure of the power estimation system.

[0017] In the following description, an "interface apparatus" may be one or more interface devices. The one or more interface devices may be at least one of the following: - One or more I / O (Input / Output) interface devices. The I / O (Input / Output) interface device is an interface device for at least one of an I / O device and a remote display computer. The I / O interface device for the display computer may be a communication interface device. The at least one I / O device may be a user interface device, for example, either an input device such as a keyboard and a pointing device, or an output device such as a display device. - One or more communication interface devices. The one or more communication interface devices may be one or more homogeneous communication interface devices (e.g., one or more NICs (Network Interface Cards)) or two or more heterogeneous communication interface devices (e.g., a NIC and an HBA (Host Bus Adapter)).

[0018] In the following description, the term "memory" refers to one or more memory devices, typically a primary storage device. At least one memory device in the memory may be a volatile memory device or a non-volatile memory device.

[0019] In the following description, a "persistent storage device" refers to one or more persistent storage devices. A persistent storage device is typically a non-volatile storage device (e.g., an auxiliary storage device), and specifically, for example, a hard disk drive (HDD) or a solid state drive (SSD).

[0020] In the following description, the term "storage device" may refer to at least one of memory and persistent storage device.

[0021] In the following description, a "processor" refers to one or more processor devices. The at least one processor device is typically a microprocessor device such as a CPU (Central Processing Unit), but may also be other types of processor devices such as a GPU (Graphics Processing Unit). The at least one processor device may be a single-core or multi-core. The at least one processor device may also be a processor core. The at least one processor device may also be a processor device in a broader sense, such as a hardware circuit that performs part or all of the processing (for example, an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit)).

[0022] In the following description, functions may be described using the expression "yyy unit." However, the functions may be realized by one or more computer programs executed by a processor, by one or more hardware circuits (e.g., FPGAs or ASICs), or by a combination thereof. When a function is realized by a program executed by a processor, the specified processing is performed using a storage device and / or an interface device, etc., as appropriate, and therefore the function may be considered to be at least a part of the processor. Processing described using a function as the subject may be processing performed by a processor or a device having the processor. A program may be installed from a program source. The program source may be, for example, a program distribution computer or a computer-readable recording medium (e.g., a non-transitory recording medium). The description of each function is an example, and multiple functions may be combined into one function, or one function may be divided into multiple functions.

[0023] An embodiment of the present invention will be described in detail below with reference to the drawings. In the following description, a "date and time" may refer to a point in time or a period of time having a certain length of time.

[0024] In this embodiment, for the sake of concrete explanation, the first component to be estimated is the active power of the load (hereinafter referred to as the load active power), and the second component is the active power of the PV output (hereinafter referred to as the PV output active power). However, in the present invention, the first component does not have to be limited to the active power of the load, and the second component does not have to be limited to the active power of the PV output. (1) Overall Configuration

[0025] FIG. 1 shows the overall configuration of a power grid management system 1 according to this embodiment.

[0026] The power grid management system 1 is a system for calculating an estimated value of power demand at a predetermined future date and time (hereinafter referred to as a predicted power demand value) based on factor data including observed data and predicted data of factors and power grid data including observed values ​​of power demand values ​​at each point of estimation, and for controlling power grid equipment 41 based on the predicted power demand value. The power grid management system 1 is composed of a power operator 2, a power grid 4, a factor information provider 5, and various devices owned by each of them. Furthermore, communication paths 60, 61, and 62 are communication paths (typically communication networks) that connect various devices such as a WAN (Wide Area Network) and a LAN (Local Area Network) so that they can communicate with each other.

[0027] In the power operator 2, a power estimation system 3 calculates a predicted value of power demand at each point for several hours ahead, for example, every hour, based on power system data and factor data, determines control details for the power system equipment 41 based on the predicted value, and controls the power system equipment 41 through a power system equipment management device 22. The power operator 2 includes a power estimation data management device 30 that manages data for calculating the predicted value of power demand, such as power system data and factor data, a power estimation device 31 that calculates the predicted value of power demand, a power system control calculation device 21 that calculates control details for the power system equipment 41, a power system equipment management device 22 that manages and controls the power system equipment 41, and an information input / output terminal 20 for exchanging data with these devices. The power system data includes system observation data and system attribute data. The system observation data includes measurement data (voltage value, current value, power value, frequency, etc.) at each point of the power system 4 measured by a measurement device 40 of the power system 4. Note that data on the power demand at each point to be estimated is included in the power system observation data. The system attribute data includes, for example, information indicating functional specifications of the system equipment 41 of the power system 4, system configuration information indicating the connection relationships of the system equipment 41, geographical location information of the system equipment 41, planning information on future setting states and control values ​​of the system equipment 41, and planning information on future system configuration. Note that controlling the system equipment 41 is an operation of adjusting the output voltage by changing the winding ratio of the transformer, for example, by changing the tap position of the transformer.

[0028] The power grid 4 is a facility that transmits power generated by power generation facilities to load facilities via power distribution facilities. The power grid 4 is composed of system facilities 41 that supply, distribute, and consume power, and measurement devices 40 that measure the status of each point in the power grid 4. The system facilities 41 include at least power generation facilities, power distribution facilities, and load facilities. The power generation facilities are facilities that supply power, and include, for example, power generation equipment that generates power using thermal power generation, hydroelectric power generation, nuclear power generation, solar power generation, wind power generation, etc., power generation control devices that control the power generation equipment, and storage batteries that store energy and supply power as needed. The power distribution facilities include various facilities related to power transmission and distribution, such as overhead power lines, underground power lines, transformers, circuit breakers, disconnecting switches, and phase modifying equipment. The load facilities are facilities that consume power, and include, for example, household electrical appliances, industrial electrical appliances, and storage batteries. The measuring device 40 is a device that measures measurement data at each point in the power system 4, and includes, for example, a voltage measuring device, a current measuring device, a power measuring device, a frequency measuring device, and the like.

[0029] The factor information provider 5 is a business operator that provides factor data including past observation data and future prediction data of factors. The factor information provider 5 includes a factor data distribution device 50 for distributing the factor data. The factor data may be time-series data including observed values ​​at each point in time in the past and / or estimated values ​​(predicted values) at each point in time in the future. For example, the factor data includes the following data (1) and / or (2). The unit of time may be any unit such as week, day, hour, or minute. (1) Data having a certain periodicity similar to the periodicity of electricity demand, such as meteorological data such as temperature, humidity, cloud cover, and solar radiation, and calendar data such as days of the week and holidays. (2) Data having a small periodicity but correlated with electricity demand, such as data on events such as sporting events, mobile data such as GPS for cars and mobile devices, data on consumers such as factory operation status, and data on social activities such as industrial trends.

[0030] Next, the power grid management system 1 according to this embodiment will be described with reference to FIG.

[0031] The measuring device 40 measures measurement data at each point in the power system 4 and transmits the data to the system equipment management device 22 .

[0032] The power system equipment management device 22 transmits power system data 3006, which includes system observation data including measurement data of each point in the power system 4 received from the measurement device 40 and system attribute data held by the power system equipment management device 22, to the power estimation system 3. The power system data 3006 may include system observation data of each substation.

[0033] The factor data distribution device 50 transmits factor data 3007 including observed data and estimated data of factors to the power estimation system 3 .

[0034] The power estimation system 3 calculates a predicted value of the power demand at each point based on the power system data 3006 received from the power system equipment management device 22 and the factor data 3007 received from the factor data distribution device 50, and transmits data including the calculated predicted value to the system control calculation device 21.

[0035] The system control calculation device 21 generates control data for the system equipment 41 based on the data received from the power estimation device 31 (data including predicted values ​​of power demand at each point) and transmits the control data to the system equipment management device 22.

[0036] The system equipment management device 22 controls the system equipment 41 based on the control data of the system equipment 41 received from the system control arithmetic device 21 .

[0037] The power estimation system 3 is a system that calculates a predicted value of power demand at each point based on the power system data 3006 and the factor data 3007, and outputs the predicted value of power demand at each point to the system control calculation device 21. As described above, the power estimation system 3 is made up of the power estimation data management device 30 and the power estimation device 31. The power estimation system 3 may be a physical computer system (one or more computers), or may be a logical computer system based on a physical computer system (for example, a cloud computing system on a cloud platform).

[0038] The power estimation data management device 30 stores the power system data 3006 received from the power system equipment management device 22 and the factor data 3007 received from the factor data distribution device 50, and transmits these data 3006 and 3007 to the power estimation device 31. The power system data 3006 includes time series data of measured load active power, and the factor data 3007 includes time series data of factor values ​​of factors that have a strong correlation with PV output active power.

[0039] The power estimation device 31 calculates a predicted value of the power demand at each point based on the power system data 3006 and factor data 3007 received from the power estimation data management device 30, and transmits data including the predicted value to the system control calculation device 21. The power estimation device 31 may be integrated with the power estimation data management device 30.

[0040] The power estimation device 31 has functions such as a first component model estimation unit 3106 , a first estimated value calculation unit 3107 , a second component model estimation unit 3108 , a second estimated value calculation unit 3109 , and a prediction processing unit 3110 .

[0041] The first component model estimation unit 3106 identifies (generates) each load model that estimates the load active power at each location at a predetermined date and time, based on the power system data 3006, the factor data 3007, and threshold data 3111 in which a predetermined threshold is set in advance. Each load model (load model for each location in the power system (e.g., a substation)) identified by the first component model estimation unit 3106 is input to the first estimate value calculation unit 3107. The load model is an example of a first component estimation model.

[0042] The first estimate value calculation unit 3107 calculates, from each load model identified by the first component model estimation unit 3106 and the power system data 3006, an estimate value of the PV output active power (first estimate value of the PV output active power) included in the observed value of the power demand at a predetermined estimation target date and time at each point included in the power system data 3006. The first estimate value of the PV output active power calculated by the first estimate value calculation unit 3107 is input to the second component model estimation unit 3108.

[0043] The second component model estimation unit 3108 identifies (generates) each PV output model that estimates the PV output active power at a predetermined date and time for each location, based on the first estimated value of the PV output active power calculated by the first estimated value calculation unit 3107 and time series data of the amount of solar radiation (an example of a factor value of a factor that is strongly correlated with the PV output active power) included in the factor data 3007. Each PV output model (a PV output model for each location in the power system (e.g., a substation)) identified by the second component model estimation unit 3108 is input to the second estimated value calculation unit 3109. The PV output model is an example of a second component estimation model.

[0044] The second estimate value calculation unit 3109 calculates an estimate of the load active power (second estimate of the load active power) and an estimate of the PV output active power (second estimate of the PV output active power) included in the observed value of the power demand at a predetermined estimation target date and time at each point included in the power system data 3006, from each PV output model identified by the second component model estimation unit 3108, the power system data 3006, and the factor data 3007. Furthermore, the second estimate value calculation unit 3109 generates power estimation data 3112 including the second estimate of the load active power and the second estimate of the PV output active power. The power estimation data 3112 generated by the second estimate value calculation unit 3109 is input to the prediction processing unit 3110.

[0045] The prediction processing unit 3110 identifies (generates) a model for predicting the load active power and the PV output active power at a predetermined prediction date and time from the power estimation data 3112 generated by the second estimate value calculation unit 3109 and the factor data 3007, and calculates the predicted values ​​of the load active power and the PV output active power at the predetermined prediction date and time. Furthermore, the prediction processing unit 3110 calculates a predicted value of the power demand at the predetermined prediction date and time from the calculated predicted values ​​of the load active power and the PV output active power, and transmits prediction result data, which is data including the predicted value of the power demand, to the system control computing device 21. The system control computing device 21 is a device that generates control data for the system equipment 41 based on the predicted values ​​of the power demand at each point in the power system, and transmits the generated control data to the system equipment management device 22 that controls the system equipment 41 based on the control data. The system control computing device 21 generates the control data based on the prediction result data from the prediction processing unit 3110. In this way, the grid equipment 41 is controlled based on the prediction result data output by the power estimation system 3. This realizes voltage optimization of the power grid 4, resulting in the effect of suppressing transmission loss, and is expected to be incorporated into practical applications such as suppressing transmission loss for power transmission and distribution companies. (2) Internal Configuration

[0046] FIG. 3 shows the functional configuration of each device that constitutes the power estimation system 3.

[0047] The power estimation data management device 30 is configured from an information processing device such as a personal computer or a server computer, and includes a CPU (Central Processing Unit) 3001 that controls the overall operation of the power estimation data management device 30, an input device 3002, an output device 3003, a communication device 3004, and a storage device 3005. The CPU 3001 is connected to the devices 3002 to 3005 so as to be able to communicate with them.

[0048] The input device 3002 is, for example, a keyboard or a mouse, and the output device 3003 is, for example, a display or a printer. The communication device 3004 is equipped with a NIC (Network Interface Card) for connecting to a wireless LAN or a wired LAN. The storage device 3005 includes memories such as RAM (Random Access Memory) and ROM (Read Only Memory).

[0049] The storage device 3005 stores power system data 3006 and factor data 3007. The power system data 3006 is data (e.g., a database) including system observation data including data such as observed values ​​of power demand at each point, and system attribute data including attribute data held by the power operator 2, such as system configuration information indicating the connection relationships of the system equipment 41.

[0050] The factor data 3007 is data (for example, a database) that stores data that indicates a correlation with power demand, such as weather data and calendar date data.

[0051] The power estimation device 31 is composed of an information processing device such as a personal computer or a server computer, and includes a CPU 3101 that controls the overall operation of the power estimation device 31, an input device 3102, an output device 3103, a communication device 3104, and a storage device 3105. The CPU 3101 is communicably connected to the devices 3102 to 3105. The communication device 3104 is an example of an interface device. The input device 3102 and the output device 3103 may be an information input / output terminal 20 (for example, a remote client computer). The CPU 3101 is an example of a processor.

[0052] The input device 3102 is configured by, for example, a keyboard or a mouse, and the output device 3103 is configured by, for example, a display or a printer. The communication device 3104 is configured by including a NIC for connecting to a wireless LAN or a wired LAN. The storage device 3105 includes memories such as RAM and ROM.

[0053] The storage device 3105 stores various computer programs for realizing processing units (functions) such as a first component model estimation unit 3106, a first estimated value calculation unit 3107, a second component model estimation unit 3108, a second estimated value calculation unit 3109, and a prediction processing unit 3110. The various computer programs are read into the CPU 3101 and executed, thereby realizing the processing units 3106 to 3110.

[0054] The storage device 3105 also stores threshold data 3111 and power estimation data 3112 .

[0055] The threshold data 3111 is data (for example, a database) including thresholds related to predetermined values ​​for extracting data used in estimating each load model in the first component model estimation unit 3106. For example, when extracting data on a day with low solar radiation, a predetermined value of solar radiation serving as a reference for extraction is stored as the threshold.

[0056] The power estimation data 3112 is data (for example, a database) including the estimated values ​​of the PV output active power and the estimated values ​​of the load active power at a predetermined target estimation date and time at each point, which are the estimation results of the second estimate value calculation unit 3109. (3) Details of each processing unit

[0057] Hereinafter, details of the embodiments of each processing unit of the power estimation device 31 will be explained using the drawings. Note that the power estimation device 31 calculates a predicted value of power demand at one or more points, but for simplicity, the case where a predicted value of power demand at one point is calculated will be explained here. (3-1) First component model estimation unit 3106

[0058] FIG. 4 shows a detailed embodiment of the first component model estimator 3106 in the power estimation system 3 .

[0059] The first component model estimation unit 3106 in this embodiment is made up of a sample extraction unit 31061 and a first component model identification processing unit 31062 .

[0060] The sample extraction unit 31061 extracts date and time data that can be considered as load active power from each of the data in the power system data 3006 and the factor data 3007. Specifically, for example, the sample extraction unit 31061 compares the solar radiation value correlated with the power demand at the target point included in the factor data 3007 with a predetermined threshold in the threshold data 3111, and identifies a time range in which the solar radiation value is equal to or less than the predetermined threshold (hereinafter referred to as the extracted time range). Furthermore, the sample extraction unit 31061 extracts data within the extracted time range of the observed value of the power demand at the target point in the power system data 3006 and data within the extracted time range of the factor data 3007, and outputs the extracted data to the first component model identification processing unit 31062.

[0061] The processing of the sample extracting unit 31061 will be described in more detail with reference to FIGS.

[0062] FIG. 5 shows a conceptual diagram of power system data 3006. Specifically, power system data 3006 is composed of a column 30061 for an identifier of each location, a column 30062 for the observation date and time of the estimation target, and a column 30063 for the observed value of power demand for each predetermined time range. Therefore, in the case of FIG. 5, each row is a single piece of data (sample) in which the observed value of power demand is divided at a time granularity of one hour. According to the first row, at the location with location ID "S0001", the observed value of power demand at 0:00 on December 10, 2020 is "0.9".

[0063] FIG. 6 shows a conceptual diagram of factor data 3007. The factor data 3007 is composed of a column 30071 for identifying each location, a column 30072 for the date and time of the factor data, and a column 30073 showing the value of each factor. Therefore, in the case of FIG. 6 , each row is a single piece of data (sample) in which the values ​​of each factor are divided at an hourly time granularity. According to the first row, at the location with location ID "S0001," the solar radiation value was "100" and the temperature value was "20" at 0:00 on December 10, 2020. For the observed data in the factor data 3007, the observation time is stored in the factor data date and time column 30072, and the observed value is stored in the column 30073 showing the value of each factor. For the predicted data in the factor data, the predicted time is stored in the factor data date and time column 30072, and the predicted value is stored in the column 30073 showing the value of each factor. Note that only some of the solar radiation and temperature data from the factor data are shown here.

[0064] Here, the predetermined threshold of the threshold data 3111 is assumed to be "150". The sample extraction unit 31061 extracts samples whose solar radiation values ​​in column 30073 of FIG. 6 are less than or equal to "150" (for example, the samples in the first row of FIG. 6 ) from the factor data 3007. Furthermore, the sample extraction unit 31061 sets the value in the date and time column 30072 of the extracted samples as the extraction time range. Next, the sample extraction unit 31061 extracts samples within the extraction time range from the power system data 3006 in the date and time column 30062 of FIG. 5. The sample extraction unit 31061 outputs the data of the extraction result to the first component model identification processing unit 31062.

[0065] Next, the first component model identification processing unit 31062 identifies a load model that estimates the load active power at a specified date and time from the data extracted by the sample extraction unit 31061, and outputs the identified load model to the first estimate value calculation unit 3107. The load model may be, for example, a regression model in which the objective variable is the observed value of the power demand value included in the extracted data and the explanatory variable is the observed value of the temperature included in the extracted data, or an average curve obtained by averaging the observed values ​​of the power demand value included in the extracted data for each time period. The sample extraction unit 31061 may use data that has been processed, such as standardized or normalized, on the data extracted by the sample extraction unit 31061 for identifying the load model. (3-2) First estimate value calculation unit 3107

[0066] The first estimate calculation unit 3107 receives the load model identified by the first component model estimation unit 3106, the power system data 3006, and the factor data 3007, inputs predetermined factor data (e.g., solar radiation factor data) required by the load model from the factor data 3007 to the load model, and calculates an estimate of the load active power (a first estimate of the load active power) for a predetermined estimation date and time from the output of the load model. Note that the first estimate calculation unit 3107 may correct the estimate as the output of the load model based on values ​​of observed values ​​of power demand in the power system data 3006 within a predetermined time range, and use the result as the first estimate of the load active power. The "correction" refers to a process of adding a statistical value (e.g., an average value) of the difference between the observed value of power demand and the output (estimate) of the load model for the predetermined time range (e.g., a time period with low solar radiation, such as before sunrise) to the output of the load model. This improves the accuracy of the portion of the time series represented by the time series data as the output of the load model (the time series of the load active power as the first estimated value) corresponding to the predetermined time range.

[0067] Next, the first estimate value calculation unit 3107 calculates the difference between the first estimate value of the load active power at a predetermined estimation date and time and the observed value of the power demand in the power system data 3006, and sets the calculation result as the estimate value of the PV output active power at the predetermined estimation date and time (first estimate value of the PV output active power). Explaining this in concrete terms, when the first estimate value of the load active power at a predetermined estimation date and time is "1.0" and the observed value of the power demand in the power system data 3006 is "0.9", the first estimate value of the PV output active power at that time is calculated as "1.0 - 0.9 = 0.1". The first estimate value calculation unit 3107 outputs the first estimate value of the PV output active power to the second component model estimation unit 3108 as data to be used for model identification of the PV output. (3-3) Second component model estimation unit 3108

[0068] The second component model estimating unit 3108 receives the first estimate of the PV output active power calculated by the first estimate calculating unit 3107 and the factor data 3007, and identifies a regression model (an example of a PV output model) in which the explanatory variable is the amount of solar radiation at a predetermined estimation time included in the factor data 3007 and the dependent variable is the first estimate of the PV output active power. Note that the second component model estimating unit 3108 may detect incompatible values ​​included in the first estimate of the PV output active power during the estimation period, perform a countermeasure such as deleting the detected incompatible values ​​from the first estimate of the PV output active power during the estimation period (in addition to deletion, for example, a process of lowering the weight of the incompatible values), and re-identify a regression model in which the first estimate of the PV output active power subjected to the countermeasure process is used as the dependent variable and the amount of solar radiation at a predetermined estimation time included in the factor data 3007 is used as the dependent variable. This is expected to improve the accuracy of the regression model. The "incompatible value" is a value (a value of the first estimated value of PV output active power) corresponding to a date and time when the error (modeling error) between the estimated value (estimated value of PV output active power) obtained by inputting solar radiation into the regression model identified from the original first estimated value of PV output active power and the first estimated value of PV output active power is equal to or greater than a predetermined reference value. The "predetermined reference value" is a threshold value of the modeling error, such as a value obtained by multiplying the standard deviation of the modeling error by a predetermined multiple.

[0069] The second component model estimating unit 3108 outputs the identified regression model as a PV output model to the second estimated value calculating unit 3109. (3-4) Second estimated value calculating unit 3109

[0070] The second estimate value calculation unit 3109 receives the PV output model identified by the second component model estimation unit 3108, the power system data 3006, and the factor data 3007. The second estimate value calculation unit 3109 inputs the amount of solar radiation at a predetermined estimation date and time from the factor data 3007 into the PV output model, and calculates an estimate of the PV output active power (a second estimate of the PV output active power) from the output of the PV output model. Next, the second estimate value calculation unit 3109 calculates the sum of the observed value of the power demand in the power system data 3006 at the predetermined estimation date and time and the second estimate of the PV output active power, and sets the calculation result as an estimate of the load active power at the predetermined estimation date and time (a second estimate of the load active power). To explain this as a specific example, when the observed value of power demand in the power system data 3006 at a certain time is "0.9" and the second estimated value of the PV output active power is "0.1," the second estimated value of the load active power at that time is calculated as "0.9 + 0.1 = 1.0." Furthermore, the second estimate value calculation unit 3109 outputs data including the second estimated value of the load active power and the second estimated value of the PV output active power to the prediction processing unit 3110 as power estimation data 3112.

[0071] 7 shows a conceptual diagram of the power estimation data 3112. The power estimation data 3112 is composed of a column 31121 for an identifier of each point, a column 31122 for a date and time, a column 31123 for indicating an estimated value of the PV output active power, and a column 31124 for indicating an estimated value of the load active power. The second estimated value of the PV output active power at a predetermined target date and time estimated using the PV output model is written in each row of 31123. The second estimated value of the load active power at a predetermined target date and time is written in each row of 31124. (3-5) Prediction Processing Unit 3110

[0072] The prediction processing unit 3110 receives the power estimation data 3112 and the factor data 3007 output by the second estimate value calculation unit 3109. The prediction processing unit 3110 uses the power estimation data 3112 and the factor data 3007 to identify a load prediction model that predicts the load active power at a predetermined date and time and a PV output prediction model that predicts the PV output active power at a predetermined date and time. Note that the "load prediction model" is, for example, a regression model in which the estimated value of the load active power in the power estimation data 3112 is used as the dependent variable and one or more factor data in the factor data 3007 is used as the explanatory variable. The "PV output prediction model" is, for example, a regression model in which the estimated value of the PV output active power in the power estimation data 3112 is used as the dependent variable and one or more factor data in the factor data 3007 is used as the explanatory variable.

[0073] Next, the prediction processing unit 3110 calculates a predicted value of the load active power at the predetermined prediction target date and time by inputting the factor data at the predetermined prediction target date and time in the factor data 3007 into the identified load prediction model. Also, the prediction processing unit 3110 calculates a predicted value of the PV output active power at the predetermined prediction target date and time by inputting the factor data at the predetermined prediction target date and time in the factor data 3007 into the identified PV output prediction model.

[0074] Next, the prediction processing unit 3110 calculates a predicted value of the power demand for the predetermined prediction date and time by combining the predicted value of the load active power and the predicted value of the PV output active power for the predetermined prediction date and time. Note that the predicted value may be a value obtained by estimating a value at a certain time using a single value, or an interval estimated value obtained by estimating a range of possible values ​​for a certain time using an interval.

[0075] This completes the process of estimating power demand by the power estimating device 31 .

[0076] Furthermore, the power estimation device 31 transmits the predicted value of the power demand at each point to the system control calculation device 21 .

[0077] Next, based on the received predicted values ​​of power demand at each point, the system control computing device 21 generates an optimal voltage setting value at each point in the power system at a predetermined time in the future and a control instruction for each system equipment 41 to realize the optimal voltage setting value, and transmits control data including the control instruction for each system equipment 41 to the system equipment management device 22. Note that the "optimum voltage setting value" is, for example, a voltage value that maximizes the amount of reduction possible in transmission loss throughout the system, or a voltage value that suppresses failures of each device in the system equipment 41. The "control instruction for each system equipment 41" is, for example, tap changing of a transformer or increasing or suppressing reactive power output by a phase modifying equipment.

[0078] Next, the system equipment management device 22 operates the system equipment 41, for example, by changing the transformer tap from No. 1 to No. 2, based on the received control data for the system equipment 41. The system equipment 41 changes its operation through the operation, and a physical change occurs, for example, the output voltage of the transformer becomes larger than before the change. (4) Flowchart

[0079] Fig. 8 shows the processing procedure of the power estimation system. Fig. 12 is a schematic diagram of the processing procedure of the power estimation system. This processing is triggered when the power estimation device 31 receives an input operation from the power manager 2, and the power estimation device 31 executes the processing from step S101 to step S105.

[0080] First, the first component model estimation unit 3106 extracts sample data (composed of observed values ​​of power demand and factor data) for days when the amount of solar radiation at a target location is below a predetermined threshold from the power system data 3006 and the factor data 3007, and generates a load model that estimates the load active power at a predetermined date and time using the extracted sample data (S101). Specifically, the first component model estimation unit 3106 identifies one or more dates and times on which a factor value of a predetermined factor related to PV output active power (e.g., solar radiation) is below a predetermined threshold from the factor data 3007. The first component model estimation unit 3106 extracts measurement sample data including measurement values ​​(observed values) for the one or more dates and times from the power system data 3006, and extracts factor sample data including factor values ​​for one or more factors (e.g., all or some factors other than the predetermined factor (e.g., temperature)) for the one or more dates and times from the factor data 3007. The first component model estimating unit 3106 generates a load model (an example of a first component estimation model) that is a model for estimating load active power, based on the measurement sample data and the factor sample data.

[0081] Next, the first estimate value calculation unit 3107 calculates an estimate of the PV output active power at a predetermined date and time in the past based on the load model identified by the first component model estimation unit 3106, the power system data 3006, and the factor data 3007 (S102). Specifically, the first estimate value calculation unit 3107 estimates a first estimate of the load active power for the estimation target period by inputting factor values ​​of the one or more factors in the factor data 3007 for the estimation target period (e.g., one or more estimation target dates and times) into the load model, and calculates a first estimate of the PV output active power for the estimation target period from the first estimate of the load active power for the estimation target period and the power demand (so-called P value) for the estimation target period in the power system data 3006 (e.g., by calculating the difference therebetween).

[0082] Next, the PV output estimation unit 3108 generates a PV output model that estimates the PV output active power at a predetermined date and time (S103) based on the first estimated value of the PV output calculated by the first estimate value calculation unit 3107 and the factor data 3007. Specifically, the PV output estimation unit 3108 generates a PV output model that is a model that estimates the PV output active power from factor value data for the estimation target period for the predetermined factor (e.g., solar radiation) among the factor data 3007 and the first estimated value of the PV output active power for the estimation target period.

[0083] Next, the second estimate calculation unit 3109 inputs data for a predetermined date and time in the factor data 3007 into the PV output model based on the PV output model generated by the PV output estimation unit 3108, the power system data 3006, and the factor data 3007, calculates an estimate of the PV output active power for the predetermined date and time, calculates an estimate of the load active power for the predetermined date and time from the calculated estimate of the PV output active power and the observed value of the power demand in the power system data 3006, and generates power estimation data 3112 that stores the estimate of the PV output active power and the estimated value of the load active power (S104). Specifically, the second estimate calculation unit 3109 inputs factor value data for a predetermined factor (e.g., solar radiation) in the factor data 3007 for the calculation period into the PV output model, thereby calculating a second estimate of the PV output active power for the calculation period. Furthermore, the second estimate value calculation unit 3109 calculates a second estimate value for the calculation target period of the load active power by subtracting the second estimate value for the calculation target period of the PV output active power (the calculated second estimate value) from the power demand for the calculation target period in the power system data 3006. Note that the "calculation target period" may be a period that coincides with or overlaps with the estimation target period, or may be a different period that neither coincides with nor overlaps with the estimation target period.

[0084] By performing the above steps S101 to S104, each component contained in the measured power demand can be estimated with high accuracy.

[0085] Next, the prediction processing unit 3110 generates a PV output prediction model that predicts PV output and a load prediction model that predicts actual load based on the power estimation data 3112 and the factor data 3007, and uses these prediction models to calculate predicted values ​​of PV output active power and load active power for a specified date and time, and calculates a predicted value of power demand for the specified date and time based on the predicted values ​​of PV output active power and load active power (S105).

[0086] With the above processing, the power demand estimation processing in this embodiment is completed. (4) Effects of this Embodiment

[0087] FIG. 9 is a conceptual diagram illustrating the effect of this embodiment.

[0088] Figure 9 shows the observed data for a certain period (top of Figure 9), a conceptual diagram of the model estimated by each method (middle of Figure 9), and the estimated results of the PV output active power included in past power demand using each method (bottom of Figure 9).

[0089] In the upper part of Fig. 9, graph 900a is a graph showing observed values ​​of active power of power demand during a predetermined period, graph 900b is a graph showing observed values ​​of reactive power of power demand during the same period, and graph 900c is a graph showing observed values ​​of solar radiation during the same period. Below, the PV output active power included in the active power of power demand for the estimation target period "4 / 4" shown in the upper part of Fig. 9 is used as the estimation target, and comparison results of estimation accuracy are shown for cases where the method of Comparative Example 1, the method of Comparative Example 2, and the method of this embodiment are used.

[0090] The method of Comparative Example 1 extracts the active power and reactive power of power demand on days when it is considered that there is no PV output (such as days when the amount of solar radiation is below a predetermined threshold) from the active power and reactive power of power demand and the amount of solar radiation, generates a conversion formula (reactive power conversion formula) that derives load active power from the reactive power of power demand, inputs the reactive power of power demand on the target estimation date and time into the generated reactive power conversion formula to estimate the load active power, and estimates the PV output active power from the difference between the estimated load active power and the active power of power demand.

[0091] The method of Comparative Example 2 generates a regression equation from the active power of power demand and the amount of solar radiation, with the amount of solar radiation as the explanatory variable and the active power of power demand as the dependent variable, derives a PV output model that estimates the PV output active power from the amount of solar radiation from the slope of the regression equation, and inputs the amount of solar radiation at the date and time to be estimated into the PV output model to estimate the PV output active power.

[0092] The common input data used for estimation in each method are observed values ​​of active power of power demand, solar radiation, and calendar information (day of the week, holidays). Note that only the method of Comparative Example 1 also uses observed values ​​of reactive power of power demand as input data.

[0093] Graph 901a shown in the middle of Figure 9 is a conceptual diagram of a reactive power conversion formula estimated by the method of Comparative Example 1. Graph 901a shows an actual correlation 901a1 between the reactive power of power demand and the load active power, and a reactive power conversion formula 901a2 estimated from the measured reactive power and active power of power demand. The measured reactive power of power demand includes reactive power originating from a reactive power compensator, etc., and therefore varies depending on the settings of the reactive power compensator equipment. Therefore, the reactive power originating from the reactive power compensator becomes noise, causing a deviation in the estimated result of the reactive power conversion formula from the actual correlation between the reactive power of power demand and the load active power to be estimated.

[0094] Graph 902a shown in the lower part of Fig. 9 shows the estimation results for the estimation target period using the method of Comparative Example 1. Graph 902a1 shows the actual PV output active power, and graph 902a2 shows the estimated value of PV output active power using the method of Comparative Example 1. Because the estimation accuracy of the reactive power conversion formula is low, an error occurs in the estimated value of the load active power for the estimation target day, which in turn causes an error in the estimated value of the PV output active power.

[0095] Graph 901b shown in the middle of Figure 9 is a conceptual diagram of a PV output model estimated using the method of Comparative Example 2. Graph 901b shows an actual correlation 901b1 between PV output active power and solar radiation, and a PV output model 901b2 estimated from the correlation between the measured active power of power demand and solar radiation. Because the measured active power of power demand also includes a load component that correlates with temperature, the value of the measured active power of power demand varies depending on changes in the load temperature even for the same amount of solar radiation. Therefore, the correlation with temperature included in the active power of power demand becomes noise in the correlation between the solar radiation and the PV output active power that is to be estimated, reducing the estimation accuracy of the PV output model.

[0096] 9 shows the estimation results using the method of Comparative Example 2. Graph 902a2 shows the estimated value of the PV output active power using the method of Comparative Example 2. Since the estimation accuracy of the PV output model is low, an error occurs in the estimated value of the PV output active power.

[0097] Graph 901c shown in the middle of Fig. 9 is a conceptual diagram of the PV output model estimated by second component model estimation unit 3108 in the method of this embodiment. As shown in graph 901c, first estimate value calculation unit 3107 calculates an estimated value of PV output by removing the load component that shows a correlation with temperature from the active power of the power demand, and estimating PV output model 901c1 from the correlation between the estimated value of PV output and the amount of solar radiation. This reduces noise due to the correlation with temperature, making it possible to estimate the PV output model with high accuracy.

[0098] 9 shows the estimation results output by the second estimate value calculation unit 3109 in the method of this embodiment. Graph 902c1 shows the estimated value of the PV output active power by the method of this embodiment, and because no reactive power is used and an accurate PV output model is obtained based on the correlation between the estimated PV output and the amount of solar radiation, it can be confirmed that the error in the PV output active power for the estimation period is smaller than in the methods of Comparative Example 1 and Comparative Example 2, and that the estimation is more accurate.

[0099] Furthermore, since past PV output and load can be estimated with high accuracy, future PV output and load can be predicted with high accuracy.

[0100] Furthermore, by accurately predicting the PV output and load, it is possible to accurately calculate the predicted value of the power demand at each point, and it is possible to accurately control the system equipment 41 over a wide area of ​​the power system 4 based on the predicted values ​​of the power demand at many points, thereby enabling efficient operation of the system equipment 41, such as reducing transmission loss in the power system 4. (5) Other Embodiments (5-1) Additional Example of Threshold Calculation Unit

[0101] In the above-described embodiment, the first component model estimation unit 3106 performs data extraction processing in the sample extraction unit 31061 based on the threshold represented by the threshold data 3011 that has been determined in advance by input from the information input / output terminal 20, etc. However, this is not limited to this, and a threshold calculation unit 31603 shown in FIG. 10 may be added to the first component model estimation unit 3106, and the threshold calculation unit 31603 may calculate the threshold data 3011.

[0102] The threshold calculation unit 31603 determines an appropriate threshold to be compared with the factor value of a predetermined factor indicated by the factor data 3007 in order to identify a date and time at which the power demand indicated by the power system data 3006 can be considered substantially as load active power, based on a parameter calculation method based on a predetermined performance index value, and outputs threshold data 3011 indicating the determined threshold. Measurement sample data including measurement values ​​for one or more dates and times at which the factor value of the predetermined factor is equal to or less than the determined threshold is extracted from the power system data 3006, factor sample data including factor values ​​for one or more factors at the one or more dates and times is extracted from the factor data 3007, and a load model is identified based on the measurement sample data and the factor sample data.

[0103] The "predetermined factor" is a factor that shows a strong correlation with PV output, such as solar radiation (at least a factor that has a stronger correlation with PV output than with load active power). If there are multiple predetermined factors, a threshold indicated by the threshold data 3111 may exist for each predetermined factor.

[0104] Furthermore, the "parameter calculation method" may be a known method such as a grid search method in which the performance index value results when several pre-prepared thresholds are adopted and the threshold value that gives the best performance index value is calculated, or Bayesian optimization in which the threshold value that is likely to give the best performance index value is sequentially and probabilistically searched for and determined.

[0105] Furthermore, the "performance index value" is a value indicating the estimation accuracy of the load model identified using the measurement sample data at the date and time that can be considered as the extracted load active power. For example, for each threshold, the performance index value may be a value such as the average value of the absolute values ​​of the estimation error. For each threshold, the "estimation error" is the difference between the estimated load active power calculated by the load model identified using the measurement sample data extracted at that threshold and the observed power demand represented by the data at the date and time that can be considered as the extracted load active power. Specifically, an example of calculating the performance index value when the specified factor is "solar radiation" and the threshold is "50" will be described. First, assume that three data items are extracted as data at dates and times that can be considered as load active power when the solar radiation amount is "50" or less: data A "solar radiation amount: 20, power demand: 1.5", data B "solar radiation amount: 40, power demand: 1.4", and data C "solar radiation amount: 10, power demand 1.6". For the sake of simplicity, a load model is generated that outputs the average value of the power demand for each data item, "1.5," as the load. The estimated value of the load model is "1.5" for each data item, and the absolute values ​​of the estimation errors for each data item are "Data A: 0.1, Data B: 0.0, Data C: 0.1." The average value of the absolute values ​​of the estimation errors, which are performance index values, is "0.066."

[0106] This allows the load model to be identified accurately, thereby improving the estimation accuracy. (5-2) Modification of the first component model estimating unit 3106

[0107] In the first component model estimator 3106 in the above-described embodiment, the sample extractor 31061 extracts measurement sample data from the power system data 3006 and factor sample data from the factor data 3007. However, both the measurement sample data and the factor sample data may be data composed of data having the same attribute value that affects the similarity of the fluctuation patterns of the load active power. The "attribute value that affects the similarity of the fluctuation patterns of the load active power" is a value that is predefined as a value that affects the similarity of the fluctuation patterns of the load active power, such as a value indicating a holiday or a weekday generated from calendar day data included in the factor data 3007. For example, in the power system data 3006 illustrated in FIG. 5 , such an attribute value may be associated with each power demand, and in the factor data 3007 illustrated in FIG. 6 , such an attribute value may be associated with each factor value for each factor. The measurement sample data includes one or more power demands (measured values) associated with the same attribute value, and the factor sample data includes one or more factor values ​​associated with the same attribute value for each of one or more factors. Specifically, an example of extracting only weekday data from each data will be described. When there are three pieces of data, namely, data A (attribute value: weekday, power demand: 1.0), data B (attribute value: weekday, power demand: 1.2), and data C (attribute value: holiday, power demand: 0.9), that can be considered as load active power, only data A and data B are extracted as measurement sample data.

[0108] By extracting only data of specific attribute values ​​such as weekdays and holidays and identifying the respective load models, it is possible to improve the estimation accuracy of the load model. (5-3) Modification of the second component model estimation unit 3108

[0109] In the above-described embodiment, the second component model estimator 3108 identifies a regression model (an example of a PV output model) based on the first estimated value of the PV output active power for the estimation period calculated by the first estimated value calculator 3107 and the amount of solar radiation for the estimation period included in the factor data 3007. Alternatively, the second component model estimator 3108 may identify a regression model based on the first estimated value of the PV output active power for the estimation period and the corrected amount of solar radiation (amount of solar radiation subjected to a predetermined correction process) at the estimation time included in the factor data 3007. The "predetermined correction process" may be, for example, a process of subtracting a predetermined correction value from the amount of solar radiation for the estimation time included in the factor data 3007. The "predetermined correction value" may be the threshold value indicated by the threshold data 3011 itself or a value based on the threshold value (e.g., a statistical value (e.g., an average value) of solar radiation for a time range in which the amount of solar radiation stored in the factor data 3007 is equal to or less than the threshold value). At least a portion of this "time range" may be included in the estimation period. Specifically, an example will be described in which correction processing is performed when the amount of solar radiation at a certain time included in the factor data 3007 is "100." If the threshold value indicated by the threshold data 3011 is "40" and the predetermined correction value is assumed to be "40," the amount of solar radiation after correction at the certain time is calculated as "100-40=60."

[0110] This makes it possible to correct the influence of the PV output component derived from a slight amount of solar radiation contained in the date and time data that can be considered as the load active power extracted by the first component model, thereby improving the estimation accuracy.

[0111] In the above-described embodiment, the second estimate value calculation unit 3109 stores the output of the PV output model in the power estimation data 3112 as the second estimate value of the PV output active power, but the second estimate value of the PV output active power may be the result of applying a predetermined correction process that utilizes the frequency characteristics of the power demand to the output of the PV output model (in other words, the value resulting from applying a predetermined correction process to the estimated value of the PV output active power calculated using the PV output model).

[0112] The "predetermined correction process utilizing the frequency characteristics of power demand" refers to a correction process based on frequency characteristics, such as the fact that PV output fluctuates significantly over short periods compared to load. Specifically, for example, the "predetermined correction process utilizing the frequency characteristics of power demand" refers to a process of generating frequency features by Fourier transforming the observed values ​​of power demand in the power system data 3006, generating high-frequency components of power demand by inverse Fourier transforming components of the frequency features that are equal to or higher than a predetermined frequency, and performing an adjustment process on the output of the PV output model so that the high-frequency components and the output of the PV output model are similar. The "adjustment process" refers to, for example, a process of adding the difference between the output of the PV output model and the high-frequency components of power demand to the output of the PV output model.

[0113] This makes it possible to estimate minute changes in PV output even when the solar radiation data included in the factor data 3007 does not have sufficient time granularity, thereby improving the estimation accuracy. (5-5) Modification of the prediction processing unit 3110

[0114] In the above-described embodiment, the prediction processing unit 3110 outputs a predicted value of power demand calculated from a predicted value of load active power and a predicted value of PV output active power at a predetermined prediction date and time. However, this is not limited to this. The predicted value may also be a single value estimated at a certain time or an interval estimate estimated over an interval over a range of possible values ​​at a certain time. Furthermore, the system control computing device 21 may generate an optimal voltage setting value at each point in the power system at a predetermined future time based on the predicted value of PV output active power, and control instructions for each system equipment 41 to achieve the optimal voltage setting value, and transmit control data including the control instructions for each system equipment 41 to the system equipment management device 22. Below, a specific example of processing is described assuming control in a situation where the fluctuation range of PV output active power due to solar radiation (the range between the upper and lower limits of the interval estimate) is large, such as on a cloudy day, but is larger than the fluctuation range of the load due to temperature. The system control computing device 21 references the maximum value of the range of the estimated value of the PV output active power interval, and generates a voltage setting value and control instructions so that the voltage does not exceed the reference range even when the PV output active power reaches its maximum value. A "control instruction" is, for example, an instruction to lower the output voltage of a transformer by changing the transformer tap from No. 1 to No. 2. The fluctuation range of the predicted value of power demand is a combination of fluctuations in PV output due to solar radiation and fluctuations in load due to temperature, and therefore is larger than the fluctuation range of the predicted value of PV output.

[0115] As a result, compared to when control is performed based on the fluctuation range of the predicted value of power demand, by performing control based on the fluctuation range of the predicted value of PV output, it is possible to perform control with reduced unnecessary control margin, thereby enabling efficient operation of the grid equipment 41, such as reducing transmission loss. (5-6) Example of adding an estimation processing management unit

[0116] In the above-described embodiment, the power estimation data 3112 output by the second estimate value calculation unit 3109 is input to the prediction processing unit 3110, but this is not limited to this. As shown in FIG. 11 , an estimation processing management unit 3113 may be added after the second estimate value calculation unit 3109, and the estimation processing management unit 3113 may receive the power estimation data 3112.

[0117] The estimation process management unit 3113 may receive the power system data 3006, the factor data 3007, the power estimation data 3112, and the PV output model. The estimation process management unit 3113 may determine whether the power estimation data 3112 satisfies a predetermined condition. If the power estimation data 3112 satisfies the predetermined condition (if the estimation accuracy of the estimated value in the power estimation data 3112 is high), the estimation process management unit 3113 may output the power estimation data 3112 to the prediction processing unit 3110. If the power estimation data 3112 does not satisfy the predetermined condition (if the estimation accuracy of the estimated value in the power estimation data 3112 is low), the estimation process management unit 3113 may output the power estimation data 3112 to the first component model estimation unit 3106 and re-execute a series of processes from the first component model estimation unit 3106 to the second estimate value calculation unit 3109 (the processes of S101 to S104 in the processing flow shown in FIG. 8 ). Specifically, in this series of processes, in addition to the extracted measurement sample data and factor sample data, at least a portion of the power estimation data 3112 output to the first component model estimation unit 3106 is used to identify the load model.

[0118] The "predetermined condition" for the power estimation data 3112 is a condition based on the definition that the estimation accuracy of the value in the power estimation data 3112 is high. For example, the average value of the error between the second estimated value of the load active power in the power estimation data 3112 and the time period when there is no solar radiation in the power system data 3006 is equal to or less than a predetermined threshold, or the modeling error of the PV output model identified by the second component model estimation unit 3108 is equal to or less than a predetermined threshold. Note that the "modeling error" is a value obtained by performing statistical processing such as summation or averaging on the error between the input data used to identify the model and the estimation result by the model. To explain this using specific values ​​as an example, if the amount of solar radiation at a certain time is "100," the input data for the PV output active power at that time is "1.0," and the estimate by the model is "1.1," the error between the input data at that time and the estimation result by the model is "0.1."

[0119] When the first component model estimation unit 3106 receives input of power estimation data 3112 from the estimation processing management unit 3113, the sample extraction unit 31061 of the first component model estimation unit 3106 extracts data of the second estimated value of the load active power from the power estimation data 3112 and data for a specified time range from the factor data 3007, and outputs the extracted data to the first component model identification processing unit 31062.

[0120] This makes it possible to estimate the load model and the PV output model repeatedly until a predetermined estimation accuracy is reached, thereby enabling estimation of a highly accurate model and improving the estimation accuracy. (5-7) Modification of the estimation target

[0121] In the power estimation system 3 according to the above-described embodiment, the estimation process is performed with the power demand at each point in the power grid 4 as the estimation target. However, this is not limited to this. The estimation process may be performed with the power demand value obtained by integrating the power demands at several points as the estimation target, and the integrated power demand estimate may be used to calculate the estimated value at each point before integration using a predetermined distribution method. For example, the prediction processing unit 3110 may generate first integrated data that is data obtained by integrating second estimated values ​​of load active power at multiple points in the power grid 4, and generate second integrated data that is data obtained by integrating second estimated values ​​of PV output active power from the multiple viewpoints. The prediction processing unit 3110 may generate integrated forecast data that is data obtained by integrating predicted values ​​of power demand at the multiple points from the first integrated data and the second integrated data. In other words, the second estimated value of the load active power may be the “first integrated data,” the second estimated value of the PV output active power may be the “second integrated data,” and the predicted value of the load active power and the predicted value of the PV output active power for the prediction period may be the “integrated forecast data.” The prediction processing unit 3110 may calculate, based on the integrated prediction data, a predicted value of the power demand at each location before the integrated prediction data was generated (i.e., before integration) using a predetermined distribution method.

[0122] Note that "integration" refers to converting multiple data series into a single data series through statistical processing such as summation or averaging. One method for selecting multiple locations to be integrated is to select locations with similar trends in power demand. One method for selecting locations with similar trends in power demand is to group highly similar locations by clustering the locations using feature quantities such as the average daily demand transition curve or daily maximum power value of each location, and select locations classified into the same group as targets for integration. Any known clustering method may be used. For example, hierarchical clustering such as the Ward method or non-hierarchical clustering such as k-means may be used.

[0123] Furthermore, the "predetermined distribution method" may be a method in which the estimated value of each point is obtained by multiplying the estimated value of the integrated power demand value by the distribution ratio of each point (i.e., a method in which the value is distributed according to the distribution ratio of each point). As a specific example, in a case where the estimated value of the integrated power demand value of points A and B is "100", the distribution ratio of point A is "0.7", and the distribution ratio of point B is "0.3", the estimated value of point A is "70" and the estimated value of point B is "30". Note that, as a method for determining the distribution ratio, there is a method in which the distribution ratio is calculated based on the average value of the average daily demand amount of each point before integration, and the distribution ratio is set in advance for each value indicating the state of the power system generated based on system configuration information indicating the connection relationship of the system equipment 41 of the power system.

[0124] This reduces the number of estimation targets, thereby reducing the processing load, and also reduces the deterioration of estimation accuracy due to the law of large numbers. (5-8) Modification of the second component model estimation unit 3108

[0125] In the above-described embodiment, it is assumed that a predicted value of power demand at one location is calculated, and the second component model estimation unit 3108 performs processing based on the estimated load and estimated PV output at that location. However, this is not limited to this. When calculating predicted values ​​of power demand at multiple locations, data from other locations may be used in the identification process of a PV output model at a certain location. As a specific example, an example will be described in which the second component model estimation unit 3108 uses the solar radiation at location B in the process of estimating a PV output model from a first estimated value of the solar radiation and PV output active power at location A. Note that the solar radiation at location B is a value measured at a shorter time interval than the solar radiation at location A. The second component model estimation unit 3108 generates a new solar radiation amount (corrected solar radiation amount) based on predetermined data processing of the solar radiation amounts at locations A and B using the solar radiation amounts at locations A and B, and estimates a PV output model from the corrected solar radiation amount and the first estimated value of the PV output active power at location A. The "predetermined data processing" refers to a weighted average based on a predetermined weighting value, or data integration processing that complements the amount of solar radiation at a time not recorded in the amount of solar radiation at point A with the value of the amount of solar radiation at point B. The "predetermined weighting value" may be calculated based on, for example, the distance between point A and point B or the similarity between the amount of solar radiation at point A and the amount of solar radiation at point B.

[0126] In this way, the second component model estimation unit 3108 may apply predetermined data processing to the factor value of the amount of solar radiation (an example of a predetermined factor) for a first point in the power system 4, based on data including the factor value of the amount of solar radiation for a second point in the power system 4, thereby calculating new data including the factor value for the amount of solar radiation to which the predetermined data processing has been applied, and may generate a PV output model for the second point from the new data and the first estimated value of the PV output active power. This allows for accurate estimation of the PV output model using accurate data from other points, making it possible to improve estimation accuracy. (5-9) Other Modifications

[0127] In the power estimation system 3 in the above embodiment, a display unit is omitted for simplicity of explanation, but the calculation results of each processing unit and intermediate results of each processing unit may be calculated as appropriate through an output device such as a display or printer.

[0128] Although several embodiments of the present invention have been described above, these are merely examples for the purpose of explaining the present invention and are not intended to limit the scope of the present invention to these embodiments. The present invention can be embodied in various other forms. For example, a form in which two or more of the above-described embodiments are used in combination may be adopted.

[0129] REFERENCE SIGNS LIST 1 Power system management system 3 Power estimation system 4 Power system 20 Information input / output terminal 21 Power system control calculation device 22 Power system equipment management device 30 Power estimation data management device 31 Power estimation device 40 Measuring device 41 Power system equipment 50 Factor data distribution device

Claims

1. A method for estimating the first component from the power system data, comprising: an interface device for receiving power system data and factor data; a storage device for storing the power system data and the factor data; and a processor connected to the interface device and the storage device, wherein for each of one or more points in a power system, the power system data includes a time series of measured values ​​of power demand at the point; and the factor data includes a time series of factor values ​​for a predetermined factor, the predetermined factor having a stronger correlation with the second component than with the first component out of a first component and a second component included in the power demand at the point; and the processor: (A) identifies from the factor data one or more dates and times at which a factor value of the predetermined factor related to the second component is equal to or less than a predetermined threshold, extracts from the power system data measurement sample data including the measured values ​​at the one or more dates and times, extracts from the factor data factor sample data including factor values ​​for the one or more factors at the one or more dates and times, and generates a first component estimation model that is a model for estimating the first component based on the measurement sample data and the factor sample data; (B) estimating a first estimate of the first component for the estimation period by inputting factor values ​​of the one or more factors among the factor data into the first component estimation model, and calculating a first estimate of the second component for the estimation period from the first estimate of the first component for the estimation period and power demand of the power system data for the estimation period; (C) generating a second component estimation model that is a model for estimating the second component from factor value data for the specified factor among the factor data for the estimation period and the first estimate of the second component for the estimation period; (D) calculating a second estimate of the second component for the calculation period by inputting factor value data for the specified factor among the factor data for the calculation period into the second component estimation model, and calculating the second estimate of the first component for the calculation period by subtracting the second estimate of the second component for the calculation period from the power demand of the power system data for the calculation period.

2. The power estimation system according to claim 1, characterized in that the processor calculates an estimate of the first component for the estimation period using the first component estimation model, performs a correction process to add to the calculated estimate of the first component a statistical value of the difference between an estimate of the first component for a specified time period among the estimates and the power demand for the specified time period of the power system data, and calculates the corrected estimate as a first estimate of the first component for the estimation period.

3. The power estimation system according to claim 1, characterized in that the processor sets the threshold value at which a performance index value, which is the estimation accuracy of the first component estimation model, is optimized as the threshold value for comparison with the factor value of the specified factor.

4. The power estimation system according to claim 1, wherein both the measurement sample data and the factor sample data are composed of data having the same attribute values ​​that affect the similarity of the fluctuation pattern of the first component.

5. The power estimation system according to claim 1, characterized in that the processor performs a correction process to subtract a correction value from a factor value of the specified factor during the estimation period, and generates the second component estimation model from data on factor values ​​of the specified factor during the estimation period, the data including the factor values ​​subjected to the correction process, and a first estimated value of the second component during the estimation period.

6. The power estimation system according to claim 1, characterized in that the processor detects an incompatible value included in the first estimated value for the estimation period of the second component, performs processing to address the incompatible value, and then generates the second component estimation model.

7. The power estimation system according to claim 1, characterized in that the processor performs a predetermined correction process based on frequency characteristics of power demand on the estimated value of the second component calculated using the second component estimation model, and calculates the result of the predetermined correction process as a second estimated value of the second component.

8. The power estimation system of claim 1, characterized in that, for each of one or more locations, the processor calculates a predicted value of the first component at the location during the prediction period and a predicted value of the second component at the location during the prediction period from the second estimated value of the first component at the location and the second estimated value of the second component at the location.

9. The power estimation system according to claim 1, characterized in that the processor determines whether or not power estimation data including the second estimated value of the first component and the second estimated value of the second component satisfies a predetermined condition based on a definition of high estimation accuracy, and if the predetermined condition is not satisfied, generates a first component estimation model using at least a part of the power estimation data in addition to the measurement sample data and the factor sample data, and then executes (B) to (D).

10. The power estimation system of claim 1, wherein the processor: generates first integrated data that is data obtained by integrating second estimated values ​​of the first component at multiple points in the power system; generates second integrated data that is data obtained by integrating second estimated values ​​of the second component from the multiple viewpoints; generates integrated forecast data that is data obtained by integrating forecast values ​​of power demand at the multiple points from the first integrated data and the second integrated data; and calculates forecast values ​​of power demand at each point before the integrated forecast data was generated using a predetermined distribution method based on the integrated forecast data.

11. The power estimation system of claim 1, wherein the processor performs predetermined data processing on the factor values ​​of the predetermined factors for a second location in the power system based on data including the factor values ​​of the predetermined factors for a first location in the power system, thereby calculating new data including factor values ​​for the predetermined factors subjected to the predetermined data processing, and generating the second component estimation model for the second location from the new data and a first estimated value of the second component.

12. The power estimation system according to claim 1, characterized in that the interface device is communicatively connected to a system control and arithmetic device, the processor calculates, for the one or more locations, a predicted value of the first component in a prediction period and a predicted value of the second component in the prediction period from the second estimated value of the first component and the second estimated value of the second component, and transmits prediction result data based on the predicted values ​​calculated for the first component and the second component to the system control and arithmetic device, the system control and arithmetic device being a device configured to generate control data for system equipment based on the predicted values ​​of power demand at the one or more locations and transmit the generated control data to a system equipment management device configured to control the system equipment based on the control data, and the system control and arithmetic device generates the control data based on the prediction result data from the processor.

13. receiving power system data and factor data, for each of one or more points in a power system, the power system data including a time series of measured values ​​of power demand at the point, and the factor data including a time series of factor values ​​for a predetermined factor having a stronger correlation with the second component than with the first component among a first component and a second component included in the power demand at the point; identifying from the factor data one or more dates and times at which a factor value of the predetermined factor related to the second component is equal to or less than a predetermined threshold; extracting from the power system data measurement sample data including the measured values ​​at the one or more dates and times; extracting from the factor data factor sample data including factor values ​​for the one or more factors at the one or more dates and times; and generating a first component estimation model that is a model for estimating the first component based on the measurement sample data and the factor sample data; a first estimate of the first component for the estimation target period by inputting factor values ​​of the one or more factors among the factor data into the first component estimation model, and calculating a first estimate of the second component for the estimation target period from the first estimate of the first component for the estimation target period and power demand of the power system data for the estimation target period; generating a second component estimation model that is a model for estimating the second component from factor value data for the predetermined factor among the factor data for the estimation target period and the first estimate of the second component for the estimation target period; calculating a second estimate of the second component for the calculation target period by inputting factor value data for the predetermined factor among the factor data for the calculation target period into the second component estimation model, and calculating the second estimate of the first component for the calculation target period by subtracting the second estimate of the second component for the calculation target period from the power demand of the power system data for the calculation target period.

14. A power estimation system comprising: a system control arithmetic device connected to the power estimation system; and a system equipment management device for controlling system equipment, wherein for each of one or more points in a power system, power system data includes a time series of measured values ​​of power demand at the point, and factor data includes a time series of factor values ​​for a predetermined factor, the predetermined factor having a stronger correlation with the second component than with the first component out of a first component and a second component included in the power demand at the point, the power estimation system receives the power system data and the factor data, identifies from the factor data one or more dates and times at which a factor value of the predetermined factor related to the second component is equal to or lower than a predetermined threshold, extracts from the power system data measurement sample data including the measured values ​​at the one or more dates and times, extracts from the factor data factor sample data including factor values ​​for the one or more factors at the one or more dates and times, and generates a first component estimation model that is a model for estimating the first component based on the measurement sample data and the factor sample data, estimating a first estimate of the first component for the estimation period by inputting factor values ​​of the one or more factors among the factor data into the first component estimation model, and calculating a first estimate of the second component for the estimation period from the first estimate of the first component for the estimation period and the power demand of the power system data for the estimation period; generating a second component estimation model that is a model for estimating the second component from factor value data for the specified factor among the factor data for the estimation period and the first estimate of the second component for the estimation period; calculating a second estimate of the second component for the calculation period by inputting factor value data for the specified factor among the factor data for the calculation period into the second component estimation model, and calculating the second estimate of the first component for the calculation period by subtracting the second estimate of the second component for the calculation period from the power demand of the power system data for the calculation period; calculating a predicted value of the first component during a prediction period and a predicted value of the second component during a prediction period from the second estimated value of the first component and the second estimated value of the second component for the one or more locations;a system management system comprising: a system control device that transmits prediction result data, the prediction result data being based on prediction values ​​calculated for the first component and the second component, to the system control calculation device; the system control calculation device generates control data for system equipment based on the prediction values ​​of power demand at the one or more points and transmits the generated control data to the system equipment management device; and the system equipment management device controls the system equipment based on the control data from the system control calculation device.

Images