JP2012048620A - Shared data generation method, generation device and generation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate virtual demand data having both a high degree of secrecy in privacy such as living situation and smaller difference between a result of appliance performance evaluation from that of using original demand data.SOLUTION: A generation method includes: a step (S2) of separating original demand data measured with an extent of an actual usage of an object facility used as a demand value, into demand data per usage; steps (S3, S4) of consolidating plural usages, as a usage group, in which a demand value in the demand data per usage mutually relatively varies and summing up individual pieces of demand data per usage included in the usage group to acquire a usage group demand value; and a step (S5) of summing up a usage group demand value randomly extracted from the usage group demand values for each usage group to generate virtual demand data.

Description

  The present invention relates to a shared data generation method, a generation apparatus, and a generation program. More specifically, the present invention relates to a technology for generating sharable virtual data using data that cannot be shared as it is due to privacy restrictions, for example.

  In the present invention, a unit (specifically, for example, an individual, a household, a business operator, or a facility) that uses various facilities / equipment or uses / consumes energy is called a consumer. In addition, the degree of use of various facilities / equipment or the degree of use / consumption of energy is regarded as demand. Moreover, the actual data of the actually measured demand is referred to as original demand data. A change with time in the value of demand per unit measurement time is called a demand curve. The demand data virtualized so that the original demand data can be shared by a plurality of persons is also called virtual demand data. In addition, the premise (in other words, conditions) of the relationship between the original demand data and the virtual demand data in the present invention is the private life / business activity of the data providing consumer that can be read from the virtual demand data and the private life / business that can be read from the original demand data. While the actual activity is different (specifically, the demand curve is different), the performance evaluation results of equipment / equipment using virtual demand data and the performance of equipment / equipment using original demand data The difference from the evaluation result is small.

  In order to promote energy conservation in the consumer (ie home and business) sector, it is important to improve the efficiency of energy use in the consumer sector, and it is necessary to grasp the device performance and manage the energy demand according to the actual energy demand. Is needed. For this reason, it is important to understand the actual energy demand in the private sector, and there is a sense of resistance to recording the actual energy demand reflecting the consumer's private life and business activities rooted in the consciousness of protecting personal information and corporate information. Because of concerns about leakage of measurement data, a contract is generally made between the measurement implementation body such as a research institution / company and a consumer such as a home / business to make the measurement data private. As a result, various analyzes and evaluations based on actual demand data must be based on a limited number of privately measured data. For this reason, even if it is claimed that the performance evaluation result of equipment / equipment was good using non-public data, there is a possibility that it may be due to the specificity of the target consumer (that is, the data providing consumer), and sufficient objectivity is achieved. Hard to hold.

  Therefore, in order to make a more accurate assessment of energy equipment and energy management systems in line with various demands such as household composition and residential areas, we collect data on energy demand facts measured by a wide variety of customers. Database (hereinafter referred to as energy demand database) is required. Furthermore, in order to increase the objectivity and the possibility of mutual verification of the evaluation results, the database needs to be open to the public or at least shareable among measurement implementation bodies that cooperate with the construction of the database. .

  And for the construction of the energy demand database, at least the private life and business activities of data providing customers cannot be read from the demand data, and the customers can cooperate with the measurement with peace of mind. It is required to be useful in various evaluations of energy equipment and energy management systems.

  As a conventional shared data generation method that converts original demand data into virtual demand data that can be shared in consideration of consumer privacy, a heat pump type water heater is taken up as the target equipment / equipment of demand data, Virtual demand data is generated by focusing on feature values such as demand values, differential values, and integrated values at representative measurement times of demand data that are important for performance evaluation, and has almost the same feature values as the original demand data. As a result, a demand curve having a shape similar to the actual demand curve on the original demand data measurement date is newly generated from the original demand data by piecewise polynomial function approximation (Non-Patent Document 1).

Sano and Shinohara, "Development of privacy-protected demand data collection and sharing method (Part 2)-sharable demand data generation method-", Chuo Research Institute R08006, Central Research Institute of Electric Power, 2009

  However, in the shared data generation method of Non-Patent Document 1, when applied to measurement data with a short interval of about 1 to 10 minutes, the demand curve of the virtual demand data obtained is the demand for the original demand data. Since a curve that does not differ from the curve is generated, the difference between the virtual demand data at short time intervals and the total demand data at 30 minutes every 30 minutes is different from the original demand data converted at the same time intervals. Since it is almost lost, it cannot be said that the privacy of the data providing customer is sufficient. For this reason, there is a problem that sharing is difficult and there is a concern that a customer will not cooperate in measurement.

  Therefore, the present invention has high privacy secrecy such as living conditions even when the demand data has a short measurement interval, and the result of the performance evaluation of the target equipment / equipment of the demand data uses the original demand data. It is an object of the present invention to provide a shared data generation method, a generation device, and a generation program that can generate virtual demand data having a small difference from the above result.

  In order to achieve this object, the shared data generation method according to claim 1 is characterized in that the original demand data measured with a certain period as a demand value is the degree of actual use per unit measurement time of equipment / equipment or energy. The process of decomposing into demand data for multiple uses based on the magnitude of the demand value per unit measurement time for each fixed period and the demand values in the demand data for multiple uses for a fixed period are interrelated. A process for calculating a usage group demand value for a certain period by summarizing a plurality of fluctuating usages as usage groups and adding the demand data for each usage included in the usage group for each unit measurement time, and for each usage group And a process for generating virtual demand data for a certain period by adding randomly extracted usage group demand values for each period It is.

  Further, the shared data generating apparatus according to claim 2 is a memory in which original demand data measured at intervals of a certain period with the actual usage level per unit measurement time of equipment / equipment or energy as a demand value is stored. A device connected to the device, which reads the raw demand data from the storage device, and converts the raw demand data into demand data for a plurality of uses based on the magnitude of the demand value per unit measurement time for each fixed period. A means for disassembling and assembling a plurality of uses whose demand values in a plurality of use-specific demand data for a certain period of time change in relation to each other, and assembling a plurality of use-specific demand data included in the use group A means for calculating the usage group demand value for each fixed period by adding up each unit measurement time and the usage group demand value for each fixed period for each usage group So that a means for generating a virtual demand data for a period of time by adding the those extracted randomly.

  Further, the shared data generation program according to claim 3 stores original demand data measured with a certain period as a demand value by using the actual use degree per unit measurement time of equipment / equipment or energy as a demand value. A means for reading raw demand data from a storage device into a computer that can access the apparatus, and a means for decomposing the raw demand data into demand data for a plurality of uses based on the magnitude of the demand value per unit measurement time at regular intervals. , Collect multiple usages whose demand values in demand data for multiple usages for a certain period of time vary in relation to each other, and collect demand data for each usage included in the usage group per unit measurement time A means of calculating usage group demand values for a certain period in addition to the usage group demand values for a certain period for each usage group It is to function as means for generating a virtual demand data for a period of time by adding the ones selected at random from.

  Therefore, according to these shared data generation method, generation apparatus, and generation program, the application group demand value is calculated by grouping a plurality of applications that satisfy the conditions and disassembling the original demand data for each application every certain period, On top of that, since virtual demand data for a certain period is generated by adding together randomly extracted usage group demand values for a certain period, decomposition and partial integration of the original demand data As a general rule, synthesis of different periods is performed (see FIG. 13).

  The equipment / equipment in the present invention specifically corresponds to, for example, a hot water heater or a stove of a general household. The energy specifically corresponds to, for example, electricity, water or gas. Specifically, for example, in the case of a water heater, facial washing, showering, hot water filling, and the like are assumed, and in the case of a stove, boiling water, stir-frying, and stewing are assumed. In addition, specific uses for energy include, for example, the use of various indoor and outdoor lighting, the use of various air conditioners and the use of various OA equipment, etc. for electricity, and the use of face washing, dishwashing and hot water filling for water. As for gas, various uses of a water heater, various uses of a stove, etc. are assumed. However, specific examples of facilities / equipment and energy or specific examples of applications are not limited to those described above. In the following, equipment / equipment or energy for which virtual demand data is generated according to the present invention is referred to as target equipment.

  In addition, the unit measurement time in the present invention is a time interval in which the demands of a plurality of uses do not overlap (in other words, the demands of a plurality of uses do not coexist), differ depending on the target equipment, and can be cited as an example. For example, an interval of about 1 minute to several minutes is assumed. In addition, the certain period in the present invention is a delimitation of virtual demand data generated by the present invention. For example, a period of one day, one week, or one month is assumed.

  According to the shared data generation method, generation apparatus, and generation program of the present invention, the decomposition and partial integration of raw demand data and the synthesis of different periods are performed in principle. It can be prevented from being read from the demand data, and it becomes possible to improve the secrecy of privacy such as living conditions. And it is expected that it will be easy to share by avoiding the privacy problems associated with demand data, and it will be easier for consumers to cooperate in measurement.

It is a flowchart explaining an example of embodiment of the shared data generation method and shared data generation program of this invention. It is a functional block diagram of a shared data generation device when the shared data generation method of the embodiment is implemented using a program. It is a figure showing the hot water supply demand amount (hot-water supply flow volume) as original demand data of an Example, and is displaying 5 days simultaneously. It is a figure showing the water heater utilization frequency per hour of each use in an Example. It is a figure which shows distribution of the total demand amount of the day of original demand data and virtual demand data when not grouping the use in an Example. It is a figure which shows distribution of the total demand amount of the day of original demand data and virtual demand data at the time of performing the grouping of the use in an Example. It is a figure which shows the demand curve of each demand data of an Example. (A) is a figure which shows the demand curve of original demand data. (B) is a figure which shows the demand curve of the virtual demand data produced | generated by the conventional shared data production | generation method. (C) is a figure which shows the demand curve of the virtual demand data produced | generated by the shared data production | generation method of this invention. It is a figure explaining the data characteristic in each time of original demand data and virtual demand data. (A) is a figure which shows the average of the hot water supply demand amount in each time of original demand data. (B) is a figure which shows the average of the hot water supply demand amount in each time of virtual demand data. (C) is a figure which shows the difference of the average of the hot water supply demand amount in each time of both data. It is a figure explaining the data characteristic in each time of original demand data and virtual demand data. (A) is a figure which shows dispersion | distribution of the hot water supply demand amount in each time of original demand data. (B) is a figure which shows dispersion | distribution of the hot water supply demand amount in each time of virtual demand data. (C) is a figure which shows the difference of the dispersion | distribution of the hot water supply demand amount in each time of both data. (A) is a figure which shows the average of the hot water supply demand amount in each time of original demand data. (B) is a figure which shows the standard deviation of the hot water supply demand amount in each time of original demand data and virtual demand data. It is a figure which shows comparative dispersion | distribution of the total demand amount of the day of the raw demand data and virtual demand data in the same day. It is a figure which shows the difference of the annual hot water supply operation cost when the case where original demand data is used, and the case where virtual demand data (10 sets) are used. It is a figure explaining the flow of the production | generation of the virtual demand data in the shared data production | generation method of this invention.

  Hereinafter, the configuration of the present invention will be described in detail based on an example of an embodiment shown in the drawings.

  1 and 2 show an example of embodiments of a shared data generation method, a generation apparatus, and a generation program according to the present invention.

  In the present invention, as the original demand data for generating virtual demand data, the degree of use of any facility / equipment or the degree of use / consumption of energy is regarded as a demand amount, for example, a predetermined value of about 1 to several minutes. Demand actual data that is a demand amount of the total (in other words, per unit measurement time) during the predetermined interval measured using the interval of 1 as the unit measurement time is used.

  First, terms and symbols used in the description of the present invention are summarized below. In the present embodiment, a certain period, which is a time delimiter of the virtual demand data to be generated and a delimitation of measurement performed over time and continuously, is defined as one day (0 to 24:00), and the actual demand situation The number of data measurement days (excluding missing days) is N days, and the number of measurements per day is T. In other words, in other words, there are N sets of actual demand data for a certain period (one day) composed of T pieces of data measured per unit measurement time.

1) Demand value y _{d, t}
The value of the demand amount measured from the date 0:00 to the t th (where t = 1, 2,..., T) from 0:00.

2) Use demand value ^yd _{, t}
Demand amount of usage u (provided that u = 1, 2,..., U) measured t-th from 0:00 on date d, where U is the total number of usage usages of the target equipment of demand data in the measurement target consumer. Represents the value of.

Then, the demand value y _{d, t} is the sum of all applications demand value y ^u _{d, t} in date d and measurement order t, it holds Equation 1.

3) Use group g
A plurality of uses u in which demand values of a demand curve change in relation to each other are collected into one group. A set of uses u belonging to the use group g is expressed as Formula 2 (in the text, Formula 2 is expressed as {u _g1 ,..., U _ggn }; the subscript gn indicates the number of uses belonging to the use group g. To express). All uses u belong to any use group g. That is, the total number of usage groups g is G (≦ U), and G usage groups are formed by collecting U usages into G.

4) Use group demand value y ^g _{d, t}
The value of the demand amount of the usage group g measured at the t-th date is represented. The demand value y ^g _{d, t} in the date d and measurement order t of the usage group g = {u _g1 ,..., U _ggn } is the sum of the usage demand values y ^u _{d, t} of the usage u belonging to the usage group g. Yes, Equation 3 holds.

Further, the relationship of Equation 4 is established between the demand value y _{d, t} and the usage group demand value y ^g _{d, t} in the date d and the measurement order t.

5) Daily total demand y _{total, d} , usage total demand y ^u _{total, d} , usage group total demand y ^g _{total, d}
The total daily demand amount y _{total, d} represents the total demand amount measured on the date d, and Equation 5 is established. In addition, the same formula as Formula 5 is established for the total demand y ^u _{total, d of} usage u measured on date d and the total demand y ^g _{total, d} of usage group g.

6) Demand curve _vec y _d
This is a vector notation of the change in demand value from 0 o'clock to 24 o'clock on date d (in the text, vector y is expressed as _vec y), and _vec y _d = (y _{d, 1} ,, ..., y _{d, T} ). Also, the date d of 24 o'clock midnight, applications demand curve _vec y ^u _d = is the temporal change of the demand value applications _{^{u (y u d, 1,}} ..., y u d, T), application group g The usage group demand curve _vec y ^g _d = (y ^g _{d, 1} ,..., Y ^g _{d, T} ) _, which is a change in demand value with time, is also expressed in the same manner.

7) Multi-dimensional distribution of demand curve _vec y F ^ ( _vec y)
The multidimensional distribution F ^ ( _vec y) (original expression is Equation 6-1) is a T-dimensional distribution composed of demand curves _vec y for N days. Further, the N-days, the T dimension distribution applications demand curve _vec y ^u application u is denoted as ^{_{F ^ u (vec y u)}} ( the original notation Equation 6-2), application group demand applications group g The T-dimensional distribution of the curve _vec y ^g is expressed as F ^ ^g ( _vec y ^g ) (original expression is Expression 6-3).

  If there is no problem with the specific date d, the symbol d may be omitted from the above symbols. Further, in order to distinguish between the original demand data and the virtual demand data, “*” is added to the right shoulder of the value of the virtual demand data.

  Next, the generation principle of virtual demand data in the shared data generation method of the present invention will be described.

The target demand data in the present invention is data representing the total demand amount of the target equipment during the measurement interval measured at a predetermined interval of N days, for example, about 1 minute to several minutes. F ^ ( _vec y), which is the distribution of the collected demand curve _vec y, is composed of data for N days, and is a true multi-dimensional distribution of demand curves measured by the consumers to be measured. Different from F ( _vec y). However, since F ( _vec y) is unknown and the purpose is to generate sharable virtual demand data that substitutes for the original demand data, the demand curve _{vec of the} generated virtual demand data on the date d is used. y ^* _d is by showing that it is a specimen from the _{F ^ (vec y), vec} y * d is assumed to be a demand data of the target customers.

  For example, the amount of demand data for the target equipment measured at intervals of several tens of minutes (total during the interval) is often the total amount of demand for the target equipment for multiple uses, reflecting the relationship with the uses. It is hard to be done. However, in most cases, the demand data of the target equipment measured at short intervals of about 1 to several minutes (total during the interval) clearly show the relationship between the usage and the demand value. And it is thought that the clearness of the relationship contributes to the usefulness of the target equipment demand data in equipment / equipment performance evaluation and analysis. For this reason, in the generated virtual demand data, it is assumed that the virtual demand data is used for facility / equipment performance evaluation and analysis, and the relationship between the usage and the demand value is stored.

  Therefore, in the present invention, “one-time equipment use” is used from the measurement time when the demand value has changed from zero to a positive value until the previous measurement time when the demand value becomes zero again. It is assumed that the demand value at each time in the “use of facility” is due to the use of a certain single use. Then, virtual demand data is generated in units of “one facility use”.

Specifically, first, the collected N days worth of original demand data was broken down into different applications in the "one-time equipment use" units, applications u obtained from the demand curve _vec y ^u use u of N-days a multi-dimensional distribution ^{_{F ^ u (vec y u)}} ( where, u = 1,2, ..., U ) to.

A demand curve ( _vec y ^{1 *} , ..., _vec) for all uses, with a daily use demand curve randomly extracted from the multidimensional distribution F ^{u u} ( _vec ^yu ) of use ^u as _{vec yu} ^{u *.} y ^{U *} ) is extracted, and a virtualized demand curve _vec y ^* is generated using Equation 1 above.

Similarly, when a virtual demand curve for N days is generated, _vec ^{yu *} is a random sample from F ^ ^u ( _vec ^yu ), so N-day _vec ^{yu *} multi-dimensional distribution It is consistent with the ^{_{F ^ u (vec y u)}} . Here, the demand curve _vec y _d of the date d is expressed as a total demand amount obtained by adding the usage demand curve _vec y ^u _d (where u = 1, 2,..., U) of each usage for all usages. The

Usually multidimensional distribution F ^ _(vec y) obtained from the demand curve _vec y of the original demand data, it simply represents the sum of the multidimensional distribution F ^ ^u applications demand curve _vec y ^u _(vec y ^u) (That is, F ^ ( _vec y) ≠ F ^ ¹ ( _vec y ¹ ) + F ^ ² ( _vec y ² ) + ... + F ^ ^U ( _vec y ^U )). For this ^{_{reason, F ^ u (vec y u}} ) and _{^{F ^ u * (vec y u}} *) and will be match, F ^ _(vec y) and F ^ _(vec y ^*) and do not match. Therefore, since the virtual demand data generated in this process is different from the multidimensional distribution obtained from the demand curve of the original demand data, it cannot be a substitute for the original demand data.

F ^ to match the _(vec y) and F ^ and _(vec y ^*) is application u = 1, 2, ..., applications demand value applications demand curve _vec y ^u in U varies related to each other The relationship between them needs to be the same in virtual demand data. Here, between uses that are used continuously in time, there is a correlation between fluctuations in the daily demand value. In the present invention, the relationship between uses of the original demand data is also stored in the virtual demand data.

  In order to preserve the relationship between uses, specifically, the distribution of total daily demand needs to be the same between the original demand data and the virtual demand data. Therefore, uses whose demand values fluctuate in relation to each other are combined into one group (that is, a use group) so that the distribution of the total daily demand amount is equal between the original demand data and the virtual demand data.

  In the present invention, the distribution of the total daily demand is the same between the original demand data and the virtual demand data regardless of which measurement day the usage group demand curve for different measurement days is changed. Usage groups are configured so that demand values of usage group demand curves do not fluctuate in relation to each other between groups.

  In addition, it is preferable to verify that there is no relationship between usage groups after configuring usage groups, and to confirm that there is no relationship between usage groups. Is not a mandatory requirement). The independence test method is not limited to a specific method. For example, Pearson's independence test (for example, Okamoto, Suzuki, Sugiyama, "Basic Statistics", Jikkyo Publishing Co., Ltd., 1977) is used. It is possible.

And, by making the demand value of the demand curve of the raw demand data _vec y fluctuate independently between the usage groups, in other words, by configuring the usage group like this, the demand curve of the raw demand data The resulting multidimensional distribution F ^ ( _vec y) can be expressed as the sum of the multidimensional distributions obtained from each usage group demand curve (ie, F ^ ( _vec y) = F ^ ¹ ( _vec y ¹ ) + F ^ ² ( _vec y ² ) + ... + F ^ ^G ( _vec y ^G )).

F ^ ^g ( _vec y ^g ) and F ^ ^{g *} ( _vec y ^{g *} ) coincide, and the multidimensional distribution F ^ ^* ( _vec y ^* ) obtained from the demand curve _vec y ^* of the virtual demand data is F Matches ^ ( _vec y). Accordingly, the demand curve _vec y ^* of the generated virtual demand data can be regarded as an extraction from the same distribution as the distribution of the demand curve obtained from the original demand data. For this reason, it is assumed that _vec y ^* is a sample from the original demand data, and the virtual demand data can be used as an alternative to the original demand data.

In this operation, by replacing the application group demand applications Group demand curve curves different measurement date _vec y g ^* in a subject date to generate a demand curve virtualized, demand values of _vec y ^g of _vec y ^g multidimensional It can be considered that the demand value of _vec y ^{g *} is shifted in the time direction according to the distribution F ^ ^g ( _vec y ^g ). Furthermore, and if the demand is not the _vec y ^g by replacing the _vec y g ^* is newly generated, or when the demand was in _vec y ^g is lost in _vec y g ^* is, F ^ ^It can be assumed that demand is generated or lost according to ^g ( _veci ^g ).

  For virtual demand data generated by combining each usage group demand curve on different days, depending on the target equipment of the demand data, for example, the external factors that change the demand of the target equipment are different, for example, the season is completely different When usage group demand curves are combined, it is unlikely that they will actually be measured by the target consumer. Therefore, the characteristics of the raw demand data are examined and combined with demand curves measured on different days of the same use group in a period when changes in external factors that change the demand for the target equipment (such as changes in climate and lifestyle patterns) are small. . For customers whose daily demand varies depending on the day of the week, such as when they are absent on weekdays and only on Saturdays and Sundays, it is also effective to limit them to combinations of usage group demand curves between weekdays or between weekends. It is.

  Then, in order to generate virtual demand data based on the above-described principle, the shared data generation method of the present invention is measured at intervals of a certain period with the actual usage level per unit measurement time of the target equipment as a demand value. In the process (S2) for decomposing the original demand data into a plurality of use-specific demand data based on the magnitude of the demand value per unit measurement time for each fixed period, and for a plurality of use-specific demand data for each fixed period Combine multiple usages whose demand values fluctuate in relation to each other as a usage group, and add the demand data for each usage included in the usage group for each unit measurement time. Virtual demand for a certain period of time by adding together the processing (S3, S4) to be calculated and the random value extracted from the usage group demand value for each period of use group It is to have a process (S5) of generating over data. In the present embodiment, the predetermined period is one day.

  The shared data generation method can be realized as the shared data generation apparatus of the present invention. This shared data generation device is a device connected to a storage device in which original demand data measured with a certain period as a demand value is a degree of actual use per unit measurement time of the target equipment. Means for reading the raw demand data from the storage device, means for decomposing the raw demand data into demand data for a plurality of uses based on the magnitude of the demand value per unit measurement time every fixed period, and for each fixed period Combining multiple usages in which demand values in multiple demand data for each usage fluctuate in relation to each other, and add the demand data for each usage included in the usage group for each unit measurement time Add the means to calculate the usage group demand value for a certain period and the randomly extracted from the usage group demand value for each period for each usage group. And means for generating a virtual demand data for a period of time. In the present embodiment, the predetermined period is one day.

  The shared data generation method and shared data generation apparatus described above can also be realized by executing the shared data generation program of the present invention on a computer. In the present embodiment, a case where a shared data generation program is executed on a computer will be described as an example.

  FIG. 2 shows the overall configuration of the computer 10 (that is, the shared data generating apparatus 10) for executing the shared data generating program 17. The shared data generation apparatus 10 includes a control unit 11, a storage unit 12, an input unit 13, a display unit 14, and a memory 15, and is connected to each other by a signal line such as a bus. A data server 16 is connected to the shared data generation apparatus 10 through a signal line such as a bus, and signals such as data and control commands are transmitted / received (that is, input / output) to / from each other via the signal line.

In the present embodiment, for example, the demand value y _{d, t} (that is, the original demand data) measured at the t-th date from 0:00 of the date d in the consumer to be measured over a period of about one year to several years is The original demand database 18 is stored in the data server 16.

  The control unit 11 performs operations related to control of the entire shared data generation device 10 and generation of virtual demand data as shared data by the shared data generation program 17 stored in the storage unit 12. Central processing unit).

  The storage unit 12 is a device that can store at least data and programs, and is, for example, a hard disk.

  The memory 15 serves as a memory space that is a work area when the control unit 11 executes various controls and calculations, and is, for example, a RAM (abbreviation of random access memory).

  The input unit 13 is an interface for giving at least an operator's command to the control unit 11, and is, for example, a keyboard.

  The display unit 14 performs drawing / display of characters, graphics, and the like under the control of the control unit 11, and is a display, for example.

  Then, by executing the shared data generation program 17, as a storage device in which the original demand data measured with a certain period as a delimiter with the actual usage level per unit measurement time of the target equipment as a demand value is stored. The control unit 11 of the shared data generating apparatus 10 which is a computer accessible to the data server 16 includes an original demand data reading unit 11a as means for reading the original demand data from the data server 16, and the original demand data is a unit for each fixed period. The application decomposing unit 11b as means for decomposing the demand data for each application based on the magnitude of the demand value per measurement time, and the demand values in the plurality of application-specific demand data for each fixed period are related to each other. Combines multiple fluctuating usages into usage groups, and simply stores demand data for each usage included in the usage group. Means (virtualized demand curve generation unit 11c, distribution comparison unit 11d, usage group configuration unit 11e) for calculating the usage group demand value for each fixed period by adding each measurement time, usage group demand for a fixed period for each usage group A virtual demand data generation unit 11f is configured as a means for generating virtual demand data for a certain period by adding together randomly extracted values. In the present embodiment, the predetermined period is one day.

  In the execution of the shared data generation method of the present embodiment, first, the original demand data reading unit 11a of the control unit 11 receives the original demand data for N days measured and collected by the measurement target consumer from the data server 16. Read (S1).

Specifically, in the present embodiment, the original demand data reading unit 11 a reads the demand value yd _{, t} stored as the original demand database 18 from the data server 16. Then, the raw demand data reading unit 11a stores the read demand value yd _{, t} in the memory 15.

Next, the application decomposing unit 11b of the control unit 11 decomposes the original demand data for each application (S2). In addition, when there are a plurality of target customers for generating virtual demand data and the original demand data (demand values y _{d, t} ) of a plurality of customers are read together in the processing of S1, the processing after S2 is demand. Do it for each house.

  In the present embodiment, the application decomposing unit 11b decomposes the original demand data by performing clustering for each application that is assumed to be a facility that is a collection target of the original demand data and is a generation target of the virtual demand data.

  The clustering performed to decompose the original demand data according to usage may use any method as long as the data can be classified into classes. Specifically, for example, it is conceivable to use the X-means method, which is one of methods for automatically determining the number of classes and the average value of data belonging to each class from the data.

  In the X-means method, data is classified into classes by the K-means method, the average value of data belonging to each class is obtained, and the number of classes is automatically calculated based on a Bayesian information criterion (also called BIC). To decide. Since the X-means method, the K-means method, and the Bayesian information criterion itself are well-known techniques, details are omitted here (for example, Dan Pelleg and Andrew Moore, “Accelerating exact k-means algorithms with geometric”). reasoning ”, International Conference on Knowledge Discovery and Data Mining, pp.277-281, 1999; CM Bishop, translated by Motoda, Kurita, Higuchi, Matsumoto, Murata,“ Pattern Recognition and Opportunity Learning, ”Springer Japan, 2008; see Konishi and Kitagawa, “Science information criteria for prediction and discovery”, Asakura Shoten, 2004).

  By applying the X-means method to the raw demand data, the total usage number U of the raw demand data measured by the measurement target consumer is determined, and the usage decomposition of “one facility use” is performed.

Specifically, it is presumed that the use of equipment having almost the same amount of demand is the use of equipment for the same purpose. However, depending on the type of the target equipment, the demand amount for use with a long duration may be more stable, and may be more appropriate as a representative value (that is, an average value) of the demand amount for each application. For this reason, the appropriate total usage number U and the representative demand μ _k for each usage are estimated using the K-means method and the Bayes information criterion, while considering the duration.

In the present invention, the amount of demand is not zero (ie, demand is generated) in the original demand data (ie, demand is generated) and immediately before it becomes zero again (that is, demand is terminated). Is used to estimate the total number of uses U and the average demand μ _u for each use u. Here, in the “facility use” analysis, the equipment use for the duration s or more during the measurement period is targeted. The number of occurrences of such equipment use is expressed as N _s times.

In the K-means method, it is assumed that the demand for usage u follows a normal distribution with mean μ _u and variance σ ² (same regardless of usage). At this time, if the usage to which each equipment usage i belongs is u (i) and the demand for each equipment usage is y _i , the maximum likelihood estimate μ ^ _u of the average demand for the usage _u (original expression is a mathematical expression) 7-1) is an average of the demands of _nu pieces of equipment belonging to the use u, and the maximum likelihood estimate σ ^ ^{2 of} variance (original expression is Expression 7-2) is the demand amount y _i of equipment use _i. Is the root mean square of the difference from the average demand μ ^ _u of the usage u (i) belonging to (Equation 8).

In the K-means method, the usage to which each facility use belongs is determined so that the maximum likelihood estimation value σ ^ ² of the variance is minimized.

Note that the optimum total number of uses U for use classification is determined so that the Bayes information criterion (BIC) shown in Equation 9 is minimized.

However, when the duration time s is changed, the optimum number of uses U based on the Bayes information criterion changes. As the duration time s increases, the target facility usage number N _s decreases, and the representativeness of the demand per unit measurement time of the application increases. Therefore, the duration s is successively reduced while not greatly changing the applications obtained with the large duration s, and the optimum number of uses U and its expected value μ ^ _u are determined. Specifically, the following procedure is performed.

<Step 1>
The continuation time s is increased from s _start by Δs, and the time point when the optimum number of uses is 1 or the continuation time s = s _end is defined as the start continuation time s = s _max . The optimal number of uses obtained at this time is U _smax , and the maximum likelihood estimate of the average demand for each use is μ ^ _u . Note that s _start , Δs, and s _end are not limited to specific values, and are set to appropriate values in consideration of the type / characteristics of the target equipment and the actual usage in the original demand data. Specifically, for example, s _start may be about 5 minutes, Δs may be about 1 minute, and s _end may be about 10 minutes.

<Step 2>
The duration time is s = s−Δs. At this time, the maximum likelihood estimate μ ^ _u of U _{s + 1} applications already obtained is fixed, and the BIC improves when a new application (U _{s + 1} +1) having an unknown average μ is added. Investigate whether or not to do so. In the case of improvement, new usages are further added to obtain the optimum usage number U _s for the duration s.

<Step 3>
If the optimal number of uses does not increase (that is, U _s = U _{s + 1} ), the process ends.

<Step 4>
When the optimal number of uses U _s increases, the optimal μ ^ _u for all U _s uses is reestimated by the K-means method, and the process returns to Step 2.

In the present embodiment, the application decomposing unit 11b reads the demand value y _{d, t} of the original demand data (of a certain consumer) stored in the memory 15 in the process of S1 from the memory 15, and “uses one facility use”. Is extracted. Specifically, date d, measurement order t (which also has information as unit measurement time and measurement time), an identifier (for example, a serial number from 1) for each “one-time equipment use”, and the demand value is zero A set of combination data of demand values yd _{, t} from when the value becomes positive value until it becomes zero after that is stored in the memory 15 as "one-time equipment use" data. In addition, it is "one use of equipment" until the demand value changes from zero to a positive value and then becomes zero, and the same identifier is given.

Further, as described above, the use decomposing unit 11b determines that the use is the same if the demand value is the same for each “one-time use of equipment”, so that the use value y _{d, t} is used for clustering. To separate the raw demand data by application.

Then, the usage decomposing unit 11b adds the usage u to each of the demand values yd _{, t} of the original demand data and stores them in the memory 15.

  Next, as a usage group review (S3), the virtualized demand curve generation unit 11c of the control unit 11 generates a virtualized demand curve (S3-1).

On the condition that the difference in distribution of _total daily demand (ie, y _total and y ^* _total ) obtained from each of N days of original demand data and virtualized demand data becomes small, Uses whose demand values fluctuate in relation to each other are grouped together.

Specifically, a combination of all possible uses is grouped into a combination of temporary use groups. Then, a usage group demand value y ^g _{d, t} is calculated for each combination of the temporary usage groups using Formula 3, and a usage group demand curve _vec y ^g _d for N days is generated.

Then, N-days of application groups demand curve _{^{vec y g d = (y g}} d, 1, ..., y g d, T) of each provisional application group from one day to randomly use groups demand curve _vec y ^{g *} is extracted, and a virtualized demand curve _vec y ^* for one day is generated using Equation 4. Similarly, X sets of virtualized demand curve _vec y ^* for N days for each combination of temporary usage groups with various random numbers when extracting usage group demand curve _vec y ^{g *} for one day Generate. In extracting the usage group demand curve in this processing, a specific usage group demand curve may be extracted a plurality of times, while there may be usage group demand curves that are not extracted at all.

  According to the above-described procedure, X sets of virtualized demand curves for N days are generated for all combinations of provisional usage groups configured as combinations of all possible usages. Note that the value of the demand curve set number X is not limited to a specific value, and is set to an appropriate value based on the subsequent processing in S3-2 and S4. Specifically, for example, it is conceivable to set a range of several tens to several hundreds. Further, the value of the set number X of the demand curve may be specified in advance in the shared data generation program 17, or may be designated / changed via the input unit 13 in the process of S3-1. May be.

In the present embodiment, the virtualized demand curve generation unit 11c reads the demand value ^yud _{, t} of the original demand data stored in the memory 15 with the use u added in the process of S2, and the above-described procedure. Thus, a usage group demand curve _vec y ^g _d for N days is generated for each combination of temporary usage groups, and a virtualized demand curve _vec y ^* is generated for each X set.

Then, the virtualized demand curve generation unit 11c generates N sets of usage group demand curves _vec y ^g _d (that is, usage group demand values y ^g _{d, t} ) and X sets for each temporary usage group combination. The demand value y ^* _{d, t} of the virtualized demand curve _vec y ^* for N days is stored in the memory 15.

  Next, the distribution comparison unit 11d of the control unit 11 compares the distribution of the total demand amount between the original demand data and the virtualized demand data (S3-2).

  The total demand amount in the process of S3-2 is the total demand amount for a certain period that is a unit of measurement (in other words, a delimiter) that is continuously performed over time. In the present embodiment, since the fixed period is one day (from 0:00 to 24:00), the distribution of the total daily demand is compared.

The total daily demand y _{total, d} based on the original demand data is calculated using the demand value y _{d, t} of the original demand data and Equation 5. Further, the total daily demand y ^* _{total, d} based on the virtualized demand data is the demand value y ^* _{d, t of} the virtualized demand curve _vec y ^* generated in the process of S3-1 and Equation 5 And is calculated using As for the total daily demand based on the virtualized demand data after the process of S 3-1, X sets of daily total demand for N days are calculated for each provisional usage group combination.

  Then, it is verified whether or not the distribution for the N days of the total daily demand based on the original demand data is equal to the distribution for the N days of the total daily demand based on the virtualized demand data. Any method may be used to test whether the two distributions are equal as long as the method can test whether the distributions of the two data sets are equal. Specifically, for example, the Colmo Golf Smirnov test (also called KS test; for example, Okamoto, Suzuki, Sugiyama, “Basic Statistics”, Jikkyo Publishing Co., Ltd., 1977) may be used. In the present embodiment, the shared data generation program 17 incorporates processing for testing whether the distributions of the two data sets are equal.

In the present embodiment, the distribution comparison unit 11d reads the demand value y _{d, t} of the original demand data stored in the memory 15 in the processing of S1 from the memory 15 and reads the total daily demand y _{total, d} (N days Min) and the demand value y ^* _{d, t} of the virtualized demand curve stored in the memory 15 in the process of S3-1 is read from the memory 15 and the total daily demand y ^* _{total, d} ( N sets of N days are calculated for each combination of provisional usage groups.

Furthermore, the distribution comparison unit 11d uses the KS test, for example, to distribute the daily total demand amount y _{total, d} of the original demand data for N days and the total daily demand amount y ^* _{total of the} virtualized demand curve. _{, d} is tested for each X set for each combination of provisional usage groups. The distribution test is performed for each of the distribution of one original demand data and the distribution of virtualized demand data of X sets for each combination of temporary use groups.

  Then, the distribution comparison unit 11d stores in the memory 15 whether or not each of the X sets of virtualized distribution of demand data is equal to the distribution of the original demand data for each combination of temporary use groups.

  Next, the usage group configuration unit 11e of the control unit 11 configures a usage group for generating virtualized demand data (S4).

The virtualized demand data that is regarded as having no significant difference from the distribution of the total daily demand for N days obtained from the original demand data by the process of S3-2 is X _t set (hereinafter, the number of judgment sets X _t ) In order to ensure confidentiality among the combinations of temporary use groups that have become above, specify the use combination that has the largest number of use groups as the use group that generates virtualized demand data. . Note that as the number of usage groups increases, virtual demand data is generated using more demand data for each fixed period, so that the generated virtual demand data is different from the original demand data. It becomes difficult to restore the original demand data itself. Further, the judgment set number X _t is not limited to a specific value, and the distribution of data created by random extraction from the demand curve set number X and original data (original demand data in this case) is the original. It is set appropriately in consideration of the level expected to be approximately equal to the distribution of data. Specifically, for example, the determination set number _Xt may be about 90% of the set number X of the demand curve. Further, to the value of the determination number of sets X _t may be configured to be pre-defined in the shared data generation program 17, may be specified or changed via the input unit 13 at the stage of processing of the S4 .

  In addition, because the number of usage groups is the same according to the above procedure, when multiple combinations of temporary usage groups are specified, it is determined that there is no significant difference from the distribution of the original demand data among them. A combination of temporary use groups having the largest number of sets of virtualized demand data is set as a use group for generating virtualized demand data. In addition, one usage group may be constituted by only one usage.

  In the present embodiment, the usage group configuration unit 11e reads the test result of the equality of the distribution of the total daily demand stored in the memory 15 in the process of S3-2, and for each temporary usage group combination, The number of sets determined to be equal to the distribution of the original demand data among the virtualized demand data of the sets is counted.

Furthermore, application group configuration unit 11e is configured to select a combination of provisional application group number set is counted in the above is equal to or higher than the determination number of sets X _t, among the combinations of applications selected group of provisionally Identify the one with the largest number of usage groups. If a combination of multiple provisional usage groups is selected, the provisional usage with the largest number of virtualized demand data sets determined to have no significant difference from the distribution of the original demand data. Identify group combinations.

  Then, the usage group configuration unit 11e stores the correspondence between each usage u and the usage group g of the usage u as a combination of usage groups g that generate demand data by virtualizing the combination of the specified temporary usage groups. Remember me.

  Next, the virtual demand data generation unit 11f of the control unit 11 generates virtual demand data (S5).

Specifically, the usage group demand curve _vec y ^{g *} _d of date d for each usage group g in the usage group combination specified in the processing of S4 is used as the usage group within the period of M days before and after the date d. Randomly extracted from the demand curve, and using formula 4, virtual demand data y ^* for one day (that is, demand value y ^* _{d, t} ) is generated. The random demand for random extraction is changed in various ways to generate N days worth of virtual demand data y ^* . Note that the random sampling period M is not limited to a specific number of days, but is set to an appropriate number of days in consideration of a period in which a change in an external factor that changes demand is considered to be small. In addition, this period M is set when there is no external factor such as a climate or a life pattern that changes demand for the target equipment (in other words, there is no change in demand characteristics depending on, for example, the season). It is not necessary.

In the present embodiment, the virtual demand data generation unit 11f uses the N-day usage group demand curve _vec y ^g _d (that is, usage) for each temporary usage group combination stored in the memory 15 in the process of S3-1. Among the group demand values y ^g _{d, t} ), N days of usage group demand curves _vec y ^g _d (usage group demand values y ^g _{d, t} ) for the combination of usage groups specified in the process of S4 are stored in the memory 15 And use these usage group demand curves _vec y ^g _d (use group demand value y ^g _{d, t} ) to generate one day of virtual demand data y ^* (that is, demand value y ^* _{d, t} ). .

Then, the virtual demand data generation unit 11 f accumulates one day of virtual demand data y ^* (demand value y ^* _{d, t} ) in the virtual demand database 19 stored in, for example, the data server 16.

  And the control part 11 complete | finishes the production | generation process of virtual demand data (END).

  According to the shared data generation method, the generation apparatus, and the generation program of the present invention configured as described above, decomposition of the original demand data, partial integration, and synthesis of different periods are performed in principle. It is possible to prevent the private life and business activities of the house from being read from the demand data, and it is possible to improve the privacy of privacy such as living conditions.

  In addition, although the above-mentioned form is an example of the suitable form of this invention, it is not limited to this, A various deformation | transformation implementation is possible in the range which does not deviate from the summary of this invention. For example, in the present embodiment, the data server 16 is a storage device in which original demand data measured with a certain period as the demand value is the actual usage level per unit measurement time of the target equipment. The storage device is not limited to this. For example, the storage unit 12 may be used as a storage device, or various external storage media / storage devices that can input / output to / from the control unit 11. good.

  The target equipment of the demand data is a heat pump type hot water heater, and the shared data generation method, generation device, and generation program of the present invention are virtual using the actual hot water supply demand data in a customer who uses the heat pump type hot water heater as the original demand data. An embodiment applied to generation of demand data will be described with reference to FIGS.

  In this example, hot water supply demand data for one year (measured days N = 358 because there was a missing measurement) measured at one minute intervals in a heat pump type hot water heater for a certain consumer was used as original demand data. . In this embodiment, the fixed period is one day, and the unit measurement time is one minute.

  Here, the heat pump type water heater which is the target equipment of the demand data in the present embodiment will be described. In the heat pump type hot water heater control having a hot water storage tank, for example, a large amount of hot water demand for one day is stored in the early morning, and when hot water is likely to be shorted, boiling is increased. At this time, when determining the amount of hot water to be stored in the early morning, the integrated value of the demand amount such as the total daily demand amount is an important factor, while when determining the amount of hot water to be heated up, the time when hot water is likely to run out The maximum demand in Japan is an important factor. Therefore, in the equipment performance evaluation of the heat pump type hot water heater, the demand value, the differential value, and the integrated value at the measurement time of the demand data are important demand characteristics.

  Then, the original demand data was decomposed according to usage (S2).

  First, in order to provide an overview of the characteristics of hot water supply demand data, which is the original demand data of this embodiment, the hot water supply flow rate (unit: L (liter)) measured for a certain five days is overlapped for all five days. The result shown in FIG. 3 was obtained.

  From FIG. 3, it was confirmed that there are a plurality of representative values in which the water supply flow rate per unit measurement time is substantially the same despite the use of hot water at different times. Each representative value is the amount of hot water supply based on hot water filling control or additional cooking control, and was considered to be due to the degree of opening of the faucet specific to the user or application.

  The number of uses of hot water supply data was determined using the Bayes information criterion, and the average flow rate of each use was determined by the K-means method, and the use of hot water supply data as raw demand data was decomposed.

As a result of the use decomposition according to the Bayes information criterion, the number of uses of the hot water supply demand data targeted in this example was six. Table 1 shows the average flow rate for each application. The duration s was 3 minutes.

  In the present embodiment, the demand value for which the time of “one hot water heater use” is within 2 minutes is set as “short-time use”, and the demand value that does not belong to any application is set as “other use”. And the total number of uses U was 8.

  In order to give an overview of the characteristics of the hot water supply demand data of this example, and in each application, the frequency of use of the hot water heater within one hour every hour from 0:00 to 24:00 of the day, and the demand value is 0 in each time zone. The result shown in FIG. 4 was obtained by counting the number of times as a positive value that did not include.

  From FIG. 4, it was confirmed that there are applications that are frequently used only in the evening and those that are frequently used in both morning and evening.

  Next, a virtual demand curve was generated (S3-1) as a study of usage groups (S3).

That is, in order to make the distribution of the _total daily demand (ie, y _total and y ^* _total ) of the demand curves of the original demand data and the virtual demand data equal, the distribution of y _total and y ^* _total by the KS test The usage groups were constructed by combining the usages that were considered equal and the distributions considered equal.

More specifically, initially, application demand for N-days for each of the eight applications without collectively applications into groups curve ^{_{vec y u (u = 1,2,}} ..., 8) of one day from the demand curve _vec y ^{u *} Randomly extracted and generated virtual demand data using Equation 1.

  Next, the distribution of total daily demand between the original demand data and the virtual demand data was compared (S3-2).

The total daily demand y ^* _total of the generated virtual demand data was calculated using Formula 5, and the result shown in FIG. 5 was obtained as the distribution of the _total demand y ^* _total . In the example shown in FIG. 5, the distribution of the total daily demand is different between the original demand data and the virtual demand data.

  Next, a usage group for generating virtual demand data was configured (S4).

As a result of grouping the applications, the eight applications are grouped into four application groups A, B, C, and D. Table 2 shows usages belonging to each usage group.

  The results shown in FIG. 6 are obtained as a distribution of the total daily demand by grouping the usages so that the distribution of the total daily demand is the same between the original demand data and the virtual demand data. .

  From the results shown in FIG. 6, the distribution of the total daily demand of the virtual demand data is the original demand data compared to the case where the virtual demand data is generated by combining a plurality of uses into a group by combining the uses. It was confirmed that it would be close to the distribution of.

Further, the results shown in Table 3 were obtained by calculating the correlation coefficient between applications group by calculating the total demand y ^g _total daily each application group from the application group demand curve for N-days.

  From the results shown in Table 3, it was confirmed that the correlation coefficient was almost zero between the application groups.

  Furthermore, the Pearson independence test confirmed that the usage group demand curves were not related to demand value fluctuations.

  Next, virtual demand data for the number of measurement days N (= 358) days is generated (S5).

  In this embodiment, changes in external factors that change the demand for hot water supply, such as climate change, are within two weeks before or after the target date for generating virtual demand data (ie, random sampling period M = 14 days). I assumed it was small. And the virtual demand data were produced | generated using the use group demand curve extracted at random from the raw demand data within two weeks before and after the data generation target date.

  In this embodiment, in order to prevent the demand curves of the virtual demand data and the original demand data on the same day from being greatly different, the usage group demand curve of the usage group A having the largest daily demand is The original demand data and the virtual demand data are the same day.

  Virtual demand data for N (= 358) measurement days was created, and as an example, the result shown in FIG. 7C was obtained as a demand curve of virtual demand data for a certain day. FIG. 7 shows the demand curve of the original demand data on the same day as the virtual demand data (in other words, corresponding to the data generation target date) (FIG. 7A), and the shared data generation of Non-Patent Document 1. The demand curve of the virtual demand data generated by the method ((B) in the figure) is also shown.

  From the result shown in FIG. 7, the demand curve of the virtual demand data (FIG. (B)) generated by the shared data generation method of Non-Patent Document 1 is very similar to the demand curve of the original demand data (FIG. (A)). Therefore, it was confirmed that the private life of the data providing customer who can be read from the virtual demand data is the same as that which can be read from the original demand data, and the confidentiality is low.

  On the other hand, the demand curve (FIG. 7C) of the virtual demand data generated by the shared data generation method of the present invention is combined with the usage group demand curves for different days in each of the usage groups A, B, C, and D. It was confirmed that it was generated. And the private life of the data providing consumer that can be read from the virtual demand data is different from that that can be read from the original demand data (for example, in the virtual demand data (FIG. (C)), it is daytime (approximately 11:00 to 14:00). However, it was confirmed that data with high confidentiality was generated, although there was a demand for hot water supply in the area, but there was no demand for hot water supply during the same time period in the original demand data (FIG. (A)).

  Moreover, the average difference between the original demand data for the number of measurement days N and the virtual demand data at each time and the average difference between the two data are calculated, and the result shown in FIG. 8 is obtained ((A) in FIG. The average of the data, (B) is the average of the virtual demand data, (C) is the difference between the averages of both data) The results shown are obtained (FIG. (A) is the distribution of the original demand data, (B) is the distribution of the virtual demand data, (C) is the difference in the distribution of both data). From FIG. 8 and FIG. 9, it was confirmed that the average difference and the variance difference at each time of the original demand data and the virtual demand data are distributed around 0 and are small, and no large bias is seen.

  Moreover, the result shown in FIG. 10 is obtained by expressing the difference between the demand value of the original demand data (that is, the amount of hot water supply) and the demand value of the virtual demand data (that is, the amount of hot water supply) at each measurement date and time by the standard deviation. Obtained ((B) in the figure). FIG. 10 also shows the average of demand values at each time of day in the original demand data (FIG. 10A).

  From FIG. 10, it was confirmed that the standard deviation at each time has a difference as much as the average value of the original demand data. In addition, the average and variance of demand values at each time of the original demand data and the virtual demand data are substantially equal (see FIGS. 8 and 9), whereas the demand of the original demand data at each time is shown in FIG. It was confirmed that the difference was clearly seen between the value and the demand value of the virtual demand data. For this reason, the virtual demand data is different from the original demand data, but the average and variance of the demand values at the respective times of the original demand data and the virtual demand data for N measurement days are almost equal. confirmed.

  Moreover, the scatter diagram shown in FIG. 11 was obtained by comparing the daily total demand of the original demand data and virtual demand data on the same date. From FIG. 11, it is confirmed that the total daily demand is different between the original demand data and the virtual demand data, and the virtual demand data is highly confidential in terms of the total daily demand. It was done.

  Subsequently, in this embodiment, in order to confirm that the virtual demand data generated by the shared data generation method of the present invention has utility in the device performance evaluation, a simulation tool of a heat pump type water heater is used. The performance evaluation results when using virtual demand data generated by the shared data generation method of the present invention were compared with the performance evaluation results when using original demand data.

  In this embodiment, the original demand data and virtual demand data for a heat pump type water heater having a rated output of 4.5 [KW], energy consumption efficiency (work rate) of 4.02, a tank capacity of 300 [L], and a daily heat release rate of 8%. The hot water heater operation cost and the number of hot water runs out (10 sets with 358 days as one set) were examined using a simulation tool. In this example, as a simulation tool for a water heater, an optimum operation configuration selection tool for a heat pump type water heater (Tokoro, Hashimoto, Shinohara, “Development of a tool for finding the optimum configuration of a heat pump type water heater”, Research Report R06018, 2007 (Year) can be used for demand data at intervals of 10 minutes, and an improvement is made so as to search for an optimum driving rule.

  In this embodiment, 10 sets of virtual demand data generated by the shared data generation method of the present invention are used in addition to one set of original demand data as a comparison source, as hot water supply demand data to be input to the simulation tool.

The operation rule of the heat pump type water heater in this example was as follows.
<Rule 1> By 7:00 in the morning, [c1] times the maximum value of the total demand for [h1] hours is stored from 7:00 in the nearest week.
<Rule 2> At 17:00 in the evening, [c2] times the maximum value of the total demand for [h2] hours from 17:00 in the nearest week is stored.
<Rule 3> When the amount of remaining hot water falls below [x1] kcal, the water is boiled until it reaches [x2] kcal.

  Each parameter of the above operating rules (specifically, h1, c1, h2, c2, x1, x2) does not cause hot water shortage in each of the original demand data and the virtual demand data (first set). The demand data was optimized so that the total sum of annual hot water supply operation costs was minimized.

  As a result of optimization, the annual hot water supply operation cost when using raw demand data was 15,000 yen. Then, the difference between the annual water heater operating cost when using the virtual demand data (10 sets) generated by the shared data generating method of the present invention and the annual water heater operating cost when using the original demand data is calculated. The result shown in FIG. 12 was obtained by calculation.

  From FIG. 12, since the maximum value of the difference in operating cost is 140 yen, which is within 1% of the annual water heater operating cost (= 15,000 yen) when using raw demand data, the shared data of the present invention It was confirmed that the virtual demand data generated by the generation method has no problem when used for equipment / equipment performance evaluation.

  The average difference between the virtual demand data (10 sets) and the annual water heater operating cost is 6 yen, and as shown in FIG. 12, the original demand data for each virtual demand data. The cost difference from the case of both positive and negative values, and the bias was confirmed to be small.

  In addition, since the cost difference from the case of the original demand data takes both positive and negative values, the difference in the hot water supply operation cost seen in FIG. 12 is the difference in the occurrence time of the hot water supply demand application in the target consumer. It was thought to reflect the fluctuations in annual water heater operating costs caused by such factors. From this, it was considered that a plurality of sets of virtual demand data generated by the shared data generation method of the present invention is effective as data for evaluating the influence of demand fluctuation.

10 Shared data generation device 17 Shared data generation program

Claims (3)

  Based on the magnitude of the demand value per unit measurement time for each fixed period, the raw demand data measured with a certain period as the demand value of the actual use degree per unit measurement time of equipment / equipment or energy And a plurality of uses in which demand values in the demand data for each of the plurality of uses for each predetermined period fluctuate in relation to each other as a use group. A process for calculating a use group demand value for each fixed period by adding the plurality of use-specific demand data included in the unit measurement time, and a use group demand value for the fixed period for each use group And a process for generating virtual demand data for a certain period of time by adding together randomly extracted ones from among the shared data Data generation method.   A device connected to a storage device in which original demand data measured with a certain period as a demand value is a degree of actual use per unit measurement time of equipment / equipment or energy, and the storage device Means for reading the original demand data from, means for decomposing the original demand data into demand data for a plurality of uses based on the magnitude of the demand value per unit measurement time for each fixed period, and the fixed period A plurality of usages in which demand values in the demand data for each of the plurality of uses vary in relation to each other are collected as a usage group, and the demand data for the plurality of usages included in the usage group are collected for each unit measurement time. Means for calculating the usage group demand value for each fixed period in addition to the usage group demand value for the fixed period for each usage group Shared data generating apparatus characterized by having a means for generating a virtual demand data of the certain period sum of which is extracted at random from.   A computer capable of accessing a storage device in which original demand data measured with a certain period as a demand value is defined as a demand value of a degree of actual use per unit measurement time of equipment / equipment or energy from the storage device Means for reading the original demand data, means for decomposing the original demand data into a plurality of demand data for each use based on the magnitude of the demand value per unit measurement time for each predetermined period, and the plurality for each predetermined period A plurality of usages in which demand values in the demand data for each usage vary in relation to each other as a usage group, and the demand data for each usage included in the usage group are added for each unit measurement time. Means for calculating a usage group demand value for each predetermined period, for each usage group, Shared data generation program for causing to function as means for generating a virtual demand data of the certain period sum of which is extracted at random from.