JP2006011715A - Estimation method for resource consumption, and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily estimate power consumption in each customer without installing a costly measurement device such as an interval meter or a real-time meter to each customer. <P>SOLUTION: In this estimation method, the power consumption in each hour of each the customer is estimated. The plurality of customers are divided into groups each having a similar power consumption pattern, and a standard load curve expressing a consumption pattern of a resource is generated in each the divided group. The group to which the customer that is an estimation object belongs is specified, the standard load curve of the specified group is corrected by a power consumption scale of the customer that is the estimation target, and resource consumption in each unit time within an arbitrary period to the customer that is the estimation object is estimated. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a resource consumption estimation method and apparatus for estimating the consumption of resources such as electricity, gas, and water.

  An electric power company supplies electric power to consumers of buildings such as homes, factories, shops, and other buildings via a network of transmission lines. In order to calculate the electricity rate, a consumer usually has a power meter.

  By the way, there is a case where it is desired to measure the time change of the power consumption for each customer for the purpose of power prediction and calculation of the same amount of deviation (see, for example, Patent Document 1). In such a case, a measuring instrument (interval meter) capable of measuring and recording electric energy every unit time and a measuring instrument (real-time meter) capable of measuring and transmitting electric energy in real time are individually demanded. Install it at home.

  However, the interval meter and the real-time meter are very expensive as compared with the normal integrating wattmeter, and it is very difficult to install the interval meter or the real-time meter for all consumers.

JP 2003-180032 A

  The present invention accurately and easily estimates the consumption of resources such as electricity, gas, and water for each consumer without installing expensive measuring instruments such as interval meters and real-time meters in each consumer. It is an object of the present invention to provide a resource consumption estimation method and apparatus capable of performing the above.

  A resource consumption estimation method according to the present invention is a resource consumption estimation method for estimating a consumption amount of a resource supplied from a supplier to a consumer via a network. Divide into groups with similar consumption patterns, generate a standard load curve that represents the consumption patterns of the resources of each group, identify the group to which the target consumer belongs, Using the standard load curve of the identified group, the resource consumption per unit time within an arbitrary period for the estimation target consumer is estimated.

  A resource consumption estimation device according to the present invention is a resource consumption estimation device that estimates a consumption amount of a resource supplied from a supplier to a consumer via a network. The storage unit that stores the standard load curve representing the resource consumption pattern for each group, the group to which the estimation target consumer belongs is identified, and the standard load curve of the identified group is determined. And an estimation processing means for estimating a resource consumption per unit time within an arbitrary period for an estimation target consumer based on the acquired standard load curve.

  In the present invention, “resources” are consumed by consumers such as households such as electricity, gas, and water, factories, and shops, and supply sources (power plants or substations, gas tanks, water storage facilities) ) To the consumer through a network (power transmission line, gas pipe, water pipe) physically connected.

  In the resource consumption estimation method and apparatus according to the present invention described above, for each consumer group, a standard load curve representing the resource consumption of the group is generated, and the group to which the estimation target consumer belongs is specified. Then, the consumption resource amount of the consumer is estimated using the standard load curve of the identified group.

  Thus, in the resource consumption estimation method and apparatus according to the present invention, without installing expensive measuring instruments such as interval meters and real-time meters in each consumer, for example, electricity, gas, water, etc. for each consumer Resource consumption can be estimated accurately and easily.

  Hereinafter, a load curve estimation system to which the present invention is applied will be described.

1 Overview of load curve estimation system (Overall configuration)
A load curve estimation system 10 to which the present invention is applied is a system that estimates a power consumption (load curve) per unit time of a customer 2 existing under a power system as shown in FIG.

  As shown in FIG. 1, the power system includes at least a power plant or a substation (hereinafter referred to as a power plant) 1 serving as a power supply source and one or more consumers 2 (customers 2) that consume power. -1, 2-2, 2-3,...) And a power transmission network 3 that transmits power from the power plant 1 to each consumer 2. The consumer 2 is a unit of an engine that consumes electric power such as homes, factories, shops, other buildings, power facilities such as street lights, and the like.

  The consumer's load curve (power consumption for each time) represents the power consumption per unit time such as every 30 minutes or every hour over a certain period such as one day or one week. Therefore, for example, as shown in FIG. 2, the load curve can be expressed by a graph with time (unit time increments) on the horizontal axis and power consumption (kWh) on the vertical axis.

  In the load curve estimation system 10, the load curve of an arbitrary customer 2 existing in the power system as shown in FIG. 1 is not directly measured but is calculated using an estimation process. The calculated load curve may be a past actual value of a target consumer or a future predicted value.

(Processing procedure)
FIG. 3 shows a load curve calculation procedure by the load curve estimation system.

  In the load curve estimation system, as shown in FIG. 3, load pattern classification design processing (S1), demand sampling survey design processing (S2), demand sampling result aggregation processing (S3), standard load curve creation processing (S4), individual The load curve estimation process (S5) is performed in order. In addition, the process from step S1 to step S4 is performed for every load pattern division (detailed later), and the process of step S5 is performed for each consumer as an estimation target.

--Load pattern classification design processing (S1)-
The load pattern classification is a classification for grouping consumers to be estimated for a load curve. For example, it is a group with the same or similar load curve shape among consumers belonging to the category such as region (location of customer), load factor (ratio of load of peak month to annual average load, etc.), business type, etc. . In the load curve estimation system, a common load curve (referred to as a standard load curve) is created for each load pattern category, and a load curve for each customer is created using the standard load curve. In the load pattern category design process (S1), this load pattern category is set.

--Demand sampling survey design process (S2)-
The standard load curve is created based on, for example, a load curve of a customer (hereinafter referred to as a sample customer) extracted in advance as a sample. In this case, in the demand sampling survey design process (S2), a specific sample for extracting sample consumers such as a survey for extracting sample customers, setting of the number of required sample customers, and stratification processing. Perform support processing. This demand sampling survey design process (S2) is executed for each load pattern category set in the previous step S1.

--Demand sampling result totaling process (S3)-
In the demand sampling result totaling process (S3), the power consumption at each time of the sample consumer obtained in the demand sampling survey design process (S2) is totaled and stored in a server or the like. There are two types of tabulation methods: real-time tabulation and online tabulation (summarization method that collects all at a later date).

--Standard load curve creation process (S4)-
In the standard load curve creation process (S4), a standard load curve is created. The standard load curve is a standard load pattern for each target day for which a load curve is desired. The standard load curve is created for each load pattern category determined in the load pattern category design process S1. In addition, in order to create a standard load curve, a standard load curve model is prepared in this processing (S4), and the demand influence factors such as weather conditions, day of the week, and seasonal information are input to this model. Calculate the curve. The standard load curve is a normalized value (for example, the magnitude of the contract power is set to 1) without considering the scale of each individual customer (the amount of power consumption). For this reason, in the next individual load curve estimation process (S5), the correction process with respect to the scale of each consumer is implemented.

--Individual load curve estimation process (S5)-
In the individual load curve estimation process (S5), the standard load curve created previously is corrected according to the demand scale of each consumer, and a load curve for each consumer is created. For the correction of the demand scale, an increase / decrease process using contract electric power with the electric power company of the target consumer, monthly meter reading electric energy, etc. is used.

2 Device Configuration of Load Curve Estimation System Next, a specific device configuration example of the load curve estimation system to which the present invention is applied will be described.

  FIG. 4 shows a system configuration diagram of a load curve estimation system 10 to which the present invention is applied.

  As shown in FIG. 4, the load curve estimation system 10 includes a load curve creation main computer 11 that performs the calculation processing of S1 to S5, a server 12 that stores various data and information, and the Internet, a telephone line, and the like. A data receiving device 15 that receives the measurement value of the power meter 14 installed in the consumer via the external network 13, and a recording medium in which the measurement value of the power meter 14 installed in the consumer is recorded ( An optical / magnetic recording medium reading device 16 for reading information from a flexible disk or an optical disk) and a terminal device 17 serving as an operation terminal are provided. The load curve creating main computer 11, server 12, data receiving device 15, optical / magnetic recording medium reading device 16, and terminal device 17 exchange data with each other via an intra network 18.

  The load curve creating main computer 11 internally includes a standard load curve temporary storage unit 20 for temporarily storing the calculated standard load curve, a load pattern category storage unit 21 for storing the load pattern category, and a sample design result. A sample design result storage unit 22 for storing data and a standard load curve model storage unit 23 for storing standard load curve models are provided. The server 12 includes a demand survey data storage unit 24 for storing demand survey data, a weather information storage unit 25 for storing weather information, a customer information storage unit 26 for storing customer information, and a system. A system demand storage unit 27 that stores demand, a meter reading record storage unit 28 that stores meter reading records, and a load curve estimation result storage unit 29 that stores load curve estimation results are provided.

  Next, the procedure of processing performed by the load curve creation main computer 11 of the load curve estimation system 10 will be described in association with the load curve calculation procedure shown in FIG.

  The load curve calculation procedure shown in FIG. 3 includes the load curve estimation preparation process (see FIG. 5) which is a preparation for load curve estimation from step S1 to step S4, and the load curve of individual target consumers. This will be described separately in two processes, namely, an individual load curve estimation process (see FIG. 6).

(Load curve estimation preparation process)
The load curve estimation preparation process (S1 to S4) shown in FIG. 5 is a series of steps for creating a standard load curve model that is necessary for creating a standard load curve that is a basis for load curve estimation of individual consumers. It is processing.

  First, in the load pattern category design process S1, load pattern categories are set for grouping customers (customers) that are load curve estimation targets. The load pattern classification can be executed by the load pattern classification design process S1 in addition to being directly designated by the user. In this processing calculation, data and information stored in the storage units 24 to 28 in the server 12 are used, and the result is stored in the load pattern storage unit 21 in the load curve creating main computer 11. Details of this processing will be described later.

  Subsequently, in the demand sampling survey design process S2, the design for the load curve sampling survey for each load pattern category determined in the load pattern category design process S1 (the sampling survey method such as the number of consumers to be sampled, stratification, etc.) Perform detailed calculation) support calculation. In these support calculations, data and information stored in the storage units 24 to 28 in the server 12 are used, and the results are stored in the sample design result storage unit 22 in the load curve creating main computer 11. Details of this process will be described later.

  Subsequently, in the demand sampling result totaling process S3, the power consumption (sample load curve) at each time of the sample consumer obtained in the demand sampling survey design process S2 is totaled. The sample load curve is measured by the watt-hour meter 14 installed for each sample consumer. Each measured sample load curve is aggregated from a portable recording medium such as an external network 13 or a flexible disk using the data receiving device 15 and the optical / magnetic medium reading device 16, and the demand survey data storage unit 24 of the server 12. Stored in

  Subsequently, in the standard load curve model creation process S4, a standard load curve model is set for each load pattern category set in the load pattern category design process S1 based on the collected sample load curve. The standard load curve model is a load curve estimation formula (or logic) using various factors such as weather, day of the week, and system demand for each day as parameters. Various calculations for supporting the creation work are included in the standard load curve model creation process S4. The standard load curve model created by this processing S4 is stored in the standard load curve model storage unit 23. The standard load curve model is prepared by a plurality of algorithms, and details thereof will be described later.

(Load curve estimation preparation process)
In the individual load curve estimation process (S5) shown in FIG. 6, load curves of individual consumers are created.

  In the individual load curve estimation process (S5), first, the target customer (target consumer) for which the load curve is estimated and the load curve estimation target period (designating a period such as one week or one month in the past, present or future) Is set (S15). Next, a load pattern classification is assigned to each target consumer (S16). Next, a standard load curve is calculated for the estimation target period set based on the standard load curve model for the assigned load pattern category (S17). A correction process according to the demand scale of each consumer is performed on the last calculated standard load curve, and a load curve is calculated (S18).

3 Load pattern classification design process (S1)
Next, detailed processing contents of the load pattern classification design processing S1 will be described.

  The load curve estimation system 10 is a method that utilizes the fact that the shape of the load curve of each customer is common within the group such as the type of business and the contract type and can be regarded as the same. In other words, the load curve estimation system 10 classifies the customers into several groups, creates a common load curve for each group (standard load curve), and calculates the load curve of each individual customer based on the standard load curve. calculate. In the load pattern category design processing S1, calculation processing for designing a grouping index (load pattern category) for setting the consumer group is performed.

(Overall processing contents)
FIG. 7 shows the overall flow of the load pattern classification design process S1. The load pattern segment design process S1 is executed by the load curve creation main computer 11.

  First, in the load pattern classification design process S1, a range of customers to be grouped is designated (S40). In the load pattern category design process S1, a category for further grouping the designated customer range is designed.

  FIG. 8 shows an example of specification of the customer range. As shown in FIG. 8, the customer range is specified by specific names (Kanto, High Voltage) for various indexes representing customer attributes such as location, contract type of electricity charges, type of industry, and load factor. The target demand range is set by specifying the numerical value range and the like. The designation is performed by the user through the input device of the main computer 11 for creating the load curve.

  Subsequently, an index (segment design index) used for segment design is selected (S41). The division design index is an index for grouping customers, and stores these indexes and attribute information for each customer in the customer information storage unit 26. From these indexes, one or a plurality of indexes that the user uses as a measure of the division design is designated.

  FIG. 9 shows an example of the segmented design index. The category design index includes categorical variables such as location, contract type, contract power, and industry, as well as numerical variables such as load factor.

  Subsequently, the user designates a segment design algorithm (S42). In the load pattern classification design process S1, two types of algorithms, an aggregation algorithm and a division algorithm, are prepared. When designated by the user, one of the algorithms is selected (S43, S44).

  Subsequently, the design of the load pattern classification is executed by the selected algorithm (S45, S46). The design process procedure for the load pattern classification in each algorithm will be described later.

  Subsequently, when the design of the load pattern classification is completed, the designed load pattern classification is temporarily stored in association with the index (S47).

  Subsequently, the user determines whether or not to further subdivide each load pattern category created by the above processing using the selected category index (S48).

  Subsequently, when subdividing the load pattern classification, the division target population is designated (S49), and the process returns to step S41, where the user subdivides the current load pattern classification. Is selected (S41), and the processing up to step S48 is performed. The designation of the population and the index may be automatically performed.

  Finally, when finishing the segmentation of the load pattern classification, the load pattern classification finally obtained through all the above processes (see, for example, FIG. 10) is stored in the load pattern classification storage unit 21 (S50). The process is terminated.

(Explanation of load pattern classification design algorithm)
Hereinafter, a design algorithm for load pattern classification will be described. First, an algorithm when only one index is selected as the segmented design index in the previous process (for example, when only the location is selected in the example of FIG. 8) will be described.

  The flow of the aggregation algorithm process (S45) is shown in the flowchart of FIG.

  In the aggregation algorithm, first, the selected segmented design index is converted into a categorical variable (S45-1). The aggregation algorithm is based on the premise that the segment design index selected in step S41 is a categorical variable (such as a business type or contract type). For example, when a numerical variable such as the load factor in FIG. 9 is selected as an index, categorical variables are implemented as shown in FIG.

  Subsequently, in the aggregation algorithm, the load curve of the sample customer and customer information of the sample customer that are aggregated in advance in the demand survey data storage unit 24 of the server 12 are acquired (S45-2). The customer information includes attribute information of each customer with respect to the segmented design index illustrated in FIG.

  Subsequently, in the aggregation algorithm, an initial cluster is generated (S45-3). The sample consumer acquired in step S45-2 is classified using the segment design index selected in step S41 based on the attribute information. In this case, each category indicated by the segmented design index is a class, and the sample consumer is assigned to one of the classes. A sample customer group assigned to a certain class is referred to as a cluster, and a cluster generated first in step S45-3 is referred to as an initial cluster.

  Subsequently, in the aggregation algorithm, reclustering is performed (S45-4).

Reclustering is a process of fusing any two clusters into a new cluster. The reclustering is performed so that clusters having similar load curve shapes are fused.
Specifically, re-clustering is performed by repeating the following processes (I) to (V).

  (I) Cluster (For each sample, the similarity of load curves of sample customers belonging to the cluster is judged based on the evaluation value Eg of the following equation (1). Eg of the following equation (1) is the sample customer belonging to the cluster. This is a value representing similarity with the sum of squared deviations of the load curve.

  (II) The cluster similarity evaluation value Eg obtained by the equation (1) is calculated for all clusters, and the consumer population for each cluster as shown in the following equation (2) is calculated for the calculated evaluation value Eg. The weight is calculated taking into account the number and the number of sample customers, and a measure representing the similarity of the entire cluster.

  (III) Of all the clusters, only two arbitrary clusters are reclustered (fused), and a scale representing similarity of the whole clusters obtained by the following formula (2) is calculated again. The difference (similarity change value) before and after the re-clustering of the scale obtained by the same calculation is obtained. The combination of these arbitrary two clusters is changed again to obtain the change value of similarity one after another, and the combination of clusters having the smallest change value is extracted.

  (IV) For the combination of clusters with the smallest change value extracted in the process of (III), the clusters are fused. This fusion process reduces the number of clusters by one.

  (V) The above processes (III) and (IV) are repeated until the number of clusters becomes one.

  The above processes (I) to (V) are reclustering processes (S45-4).

  Subsequently, in the aggregation algorithm, classification is determined (S45-5).

  Here, the relationship between the cluster status at each stage of each reclustering and the evaluation value of the entire cluster of the above formula (2) is presented to the user. For example, a diagram showing the fusion process with respect to the number of clusters as shown in FIG. 13 and a diagram showing the relationship between the number of clusters and the evaluation value as shown in FIG. 14 are presented to the user. Based on this presentation, the user determines the fineness (number of clusters) of the clusters to be used as the load pattern classification and inputs them to the computer 11.

  In the aggregation algorithm, the cluster finally determined by the user as described above becomes the load pattern classification.

  Next, the flow of the division algorithm process (S46) will be described with reference to the flowchart of FIG.

  In the division algorithm, first, a load curve of a sample customer and customer information of the sample customer that are aggregated in advance in the demand survey data storage unit 24 of the server 12 are acquired (S46-1). The customer information includes attribute information of each customer with respect to the segmented design index illustrated in FIG. Note that the segmented design index used in this method targets only numerical data such as load factor and contract power or order data.

  Subsequently, in the division algorithm, the similarity of the entire acquired sample is evaluated based on the evaluation value Eo of the following equation (3) based on the acquired load curve of the sample consumer (S46-2).

  Subsequently, in the division algorithm, the sample consumer is divided into two groups (S46-3). Specifically, the division is performed by performing the following processes (I) to (IV).

  (I) Select an arbitrary value for the selected segmented design index, and divide the sample customers into two groups using the selected value as a dividing point.

  (II) For each divided group, the similarity of demand is evaluated by the above formula (3). Thereafter, the similarity Eo ′ of the entire group after the division is obtained by the following equation (4).

  (III) The similarity Eo of the whole group before division calculated in step S46-2 is compared with the similarity Eo ′ of the whole group after division calculated by the above processing as shown in the following formula (5). The degree of similarity improvement Δo by group division is obtained.

  (IV) The processing from (I) to (III) is carried out instead of the dividing point, the dividing point having the greatest improvement degree Δo obtained by the above equation (5) is obtained, and this point is optimally divided at the time of dividing. As a point, it is divided into two groups.

  The processes from (I) to (IV) described above are group dividing processes (S46-3).

  Subsequently, in the division algorithm, the processing from step S46-2 to step S46-3 is performed on each divided group to perform re-division, and the number of divisions is increased (step S46-4).

  Subsequently, in the division algorithm, the classification is determined (S46-5).

  Here, the relationship between the group status at each stage of each subdivision and the evaluation value of the entire group of the above equation (4) is presented to the user. For example, a diagram showing the division process with respect to the number of groups as shown in FIG. 16 and a diagram showing the relationship between the number of groups and the evaluation value as shown in FIG. 17 are presented to the user. Based on this presentation, the user determines the fineness (number of divisions) of the demand group used as the load pattern classification and inputs it to the computer 11.

  In the division algorithm, the group finally determined by the user as described above becomes the load pattern classification.

  Next, in the process when the division algorithm is selected, when a plurality of segment design indices are selected (for example, when the location and contract type are selected in the example of FIG. 8), the load pattern segment design process Will be described. The processing procedure is the same as that in the flowchart of FIG.

  In the division algorithm, first, a load curve of a sample customer and customer information of the sample customer that are aggregated in advance in the demand survey data storage unit 24 of the server 12 are acquired (S46-1). The customer information includes attribute information of each customer with respect to the segmented design index illustrated in FIG.

  Subsequently, in the division algorithm, based on the acquired load curve of the sample customer, the similarity of the entire acquired sample is evaluated based on the evaluation value Eo of the above formula (3) (S46-2).

  Subsequently, in the division algorithm, the sample consumer is divided into two groups (S46-3). Specifically, the division is performed by performing the following processes (I) to (IV).

  (I) Select an arbitrary index from the selected segmented design index, select an arbitrary value for the arbitrary index, and divide the sample customers into two groups using the selected value as a dividing point.

(II) The optimum dividing point is obtained by the above equations (4) and (5). That is, the point where the difference between the similarity Eo before division and the similarity Eo ′ after division is the largest is set as a division point.

  (III) An index different from the segment design index selected in (I) is selected, the processes of (I) and (II) are performed to find the optimum dividing point, and the division is performed.

  (IV) Find the optimal division points for all the division design indexes, compare the formula (5), find the degree of similarity improvement Δo by group division, and find two groups at the division point with the highest improvement degree Δo. Divide into

  The processes from (I) to (IV) described above are group dividing processes (S46-3).

  Subsequently, in the division algorithm, the processing from step S46-2 to step S46-3 is further performed on each divided group to perform re-division, and the number of divisions is increased (step S46-4).

  Subsequently, in the division algorithm, the classification is determined (S46-5).

  Here, the relationship between the group status at each stage of each subdivision and the evaluation value of the entire group of the above equation (4) is presented to the user. For example, a diagram showing the division process for the number of groups as shown in FIG. 18 and a diagram showing the relationship between the number of groups and the evaluation value shown in FIG. 17 are presented to the user. Based on this presentation, the user determines the fineness (number of divisions) of the demand group used as the load pattern classification and inputs it to the computer 11.

  In the division algorithm, the group finally determined by the user as described above becomes the load pattern classification.

4 Demand sampling survey design process (S2)
Next, detailed processing contents of the demand sampling survey design process S2 will be described. The demand sampling survey design process S2 is executed by the load curve creating main computer 11.

  The standard load curve model is created based on a sample load curve collected separately from a sample customer. When collecting sample load curves, it is more economical (ie, using as few sample customers as possible) and less biased (only samples with certain characteristic load curve shapes will be sampled. Therefore, it is necessary to select a sample customer so that the characteristics of the average load curve of the customer class defined as the load curve category can be grasped).

  The demand sampling survey design process S2 supports sample design for sampling of a load curve (sample survey such as the number of consumers acquired as a sample and what type of customer is selected).

  FIG. 19 shows the overall flow of the demand sampling survey design process S2.

  In the demand sampling survey design process S2, first, a sample extraction frame is set (S51). A sample extraction frame is a consumer class for the purpose of sampling a load curve. Normally, each load pattern category obtained in the load pattern category design process (S1) is set as a sample extraction frame. FIG. 20 shows an example of setting the sample extraction frame. As shown in FIG. 20, one or a plurality of sections of each load pattern section stored in the load pattern section storage unit 21 is selected and designated as a sample extraction frame.

  Subsequently, in the demand sampling survey design process S2, a target accuracy is designated (S52). The target accuracy is an index for evaluating the difference of the load curve obtained by sampling with respect to the load curve of all customers in the sample frame. In this process, the difference of the sample load curve from the average value of the load curve of all customers in the sample frame (when multiple sample customers are selected, the average value of all sample load curves is used as the sample load curve). Is evaluated probabilistically. Specifically, α, β, and γ are designated as target accuracy for the computer 11 in the form of a language rule such as the following equation.

Specification of precision =
An error occurring at a probability level of α % is set to β × 100% (or γ kWh) or less.

  Subsequently, in the demand sampling survey design process S2, a sample extraction method is designated (S53).

  Examples of sample extraction methods include random extraction, systematic extraction, and village extraction. Random sampling is a method of randomly selecting a sample customer from all customers in the sample extraction frame. System extraction is a method of deterministically extracting with a certain rule based on customer numbers of all customers in the sample extraction frame. Village extraction is a method in which all customers in the sample extraction frame are collectively extracted in units such as power system transformers.

  In addition, here, the presence / absence of stratification is also specified. The stratification will be described later. As for sample customers extracted by the above extraction method, as shown in FIG. 21, a flag is attached to the customer information in the customer information storage unit 26.

  Subsequently, in the demand sampling survey design process S2, when stratification is used together in the sample extraction method, an auxiliary variable is selected (S54). One or more auxiliary variables are selected from the customer information of each customer stored in the customer information storage unit. The auxiliary variable candidates are, for example, “monthly electric energy”, “load factor (calculated from monthly electric energy)”, “industry”, and the like of each consumer.

  Subsequently, in the demand sampling survey design process S2, sample design is performed (S55).

  In the sample design (S55), a calculation process for obtaining the number of consumers to be sampled is performed. The number of sample customer extractions is automatically calculated so as to satisfy the target accuracy specified in S52.

  Finally, these sample design results are stored in the sample design result storage unit 22, and the present process ends.

  Next, the calculation process for obtaining the number of customers to be sampled in step S55 will be further described.

  First, a case where sample design is performed without using stratification will be described as an example.

  If random sampling is specified, the probability is α as the target accuracy, the error is β, tα is the value of the probability point α of the t distribution with N−1 degrees of freedom, and N is the number of all consumers in the sample extraction frame. The number n of customers to be sampled is calculated by the following equation (6).

  However, since it is impossible to obtain the variation (standard deviation) and the average value, the values from the load curve obtained by pre-sampling or the like in advance as in the following formulas (7-1) and (7-2), or Set based on user's subjectivity.

  Note that sample demand may have a large gap in the power consumption of sample consumers. In that case, normalization is performed based on the maximum value of the load curve of each sample consumer as follows as necessary.

  Next, a case where stratification is performed together will be described.

  Stratification is a method for achieving higher sample extraction accuracy with the same number of samples by further subdividing the load pattern classification selected as the sample extraction frame. A case where random sampling is selected as the sample extraction method will be described as an example. In this case, first, in order to obtain the accuracy of the sample design when stratification is performed, the variance is calculated based on the following equation (9).

  Next, a sample error is calculated by the following equation (10).

  Since it is impossible to obtain variations and average values of auxiliary variables in the partial extraction frame, the auxiliary variables obtained in advance by pre-sampling or the like in the same manner as equations (7-1) and (7-2). It is set based on the value from or the subjectivity of the user.

  In the stratification process, the sample extraction frame is divided such that the value of the above equation (10) and the target error E satisfy the condition of the following equation (11).

  Further, as a method of subdividing the extraction frame, (A) a method in which the user arbitrarily divides, and (B) a method in which the load pattern classification design process is used. (C) A method using the Dalenius-Hodges method. Three types of methods can be used.

  In the method (B), the customer stored in the load pattern category storage unit 21 is set as the target population in the sample extraction frame. After that, the auxiliary variable for which the selection of the segment design index is specified first is specified. Thereafter, the sample extraction frame is divided along the processing flow.

  In addition, about the number of sample consumers of the partial extraction frame after a division | segmentation, Neyman distribution etc. of following formula (12) can be utilized.

5 Standard load curve creation processing (S4)
Next, detailed processing contents of the standard load curve creation processing S4 will be described. The standard load curve creation process S4 is executed by the load curve creation main computer 11.

  The load curve estimation system 10 creates a load curve for each individual customer using a standard load curve model. The standard load curve model is a load curve estimation formula or load curve estimation logic prepared in the computer 11 for each load pattern category.

  The load curve output from the standard load curve model is called a standard load curve. The standard load curve is a load curve that has been normalized (for example, the maximum demand for one year is set to 1, for example) so as to facilitate the correction of the difference in the size (the amount of power consumption) of each consumer. .

(Standard load curve model types)
The load curve of electric power is affected by various factors such as season, day of the week, time of day, and weather conditions of each day, in addition to the industry of the customer. The standard load curve model is a load curve estimation formula for creating a load curve for any target day (or a fixed period) after taking into account the demand influence factors for each day for each load pattern category. Load curve estimation logic.

  As specific examples of the standard load curve model, types such as “common shape model”, “immediate sample model”, “driving pattern model”, “similar day model”, and “demand modeling model” are prepared. The user selects one of these or selects a plurality of them (in this case, by averaging a plurality of standard load curve models to be output and combining them into one model). Note that these models are examples, and other models may be used. In addition, in the process of creating a load curve for each individual customer (S5), which will be described later, in addition to arbitrarily selecting one model from these, it is also possible to calculate all the computable models.

  Hereinafter, the “common shape model”, “immediate sample model”, “driving pattern model”, “similar day model”, and “demand modeling model” will be specifically described.

--Common shape model--
The common shape model includes the entire power system (corresponding to the total power generation output pattern of the power plant) and a part of it (corresponding to the hourly measured value of the secondary power consumption of the transformer in the substation). It is a standard load curve model for creating a standard load curve from an actual measurement pattern of electric energy. This standard load curve model can be used when a rough category such as a region unit is applied to the load pattern category. In addition, when there is a consumer who can measure the amount of power per unit time such as an interval meter in the system, processing such as subtracting the demand is performed by the user's selection.

  FIG. 22 shows a specific example of a standard load curve creation method using a common shape model.

  In the common shape model, first, the total demand (curve 1) of the power system on the target date (x year y month z day) at the power plant 1 (or substation) is measured, and the target date (x year y) The actual measured value (curve 2) of the customer 2-1 in which the interval meter of (month z day) is set is directly measured.

  Subsequently, the actual measured value of the customer 4-1 load in which the interval meter is set is subtracted from the overall demand of the power system. As a result, the total load curve of the target day (x year y month z day) of the target customer group (for example, the customers 2-2 to 2-N in the district A) is calculated.

  Subsequently, the load pattern is normalized to obtain a standard load curve.

  As a normalization method, for example, a method using a method using contract power as shown in the following formula (13) can be applied. In addition, you may normalize using not only this method but another method.

--Immediate sample model--
In the common shape model, the measured value of the load curve in the system unit was used, but in the immediate sample model, one or more sample customers are set for each load pattern category, and the set sample customers are targeted. The amount of power demand per unit time of day (x year y month z day) is measured and a standard load curve is created based on the measured value.

  FIG. 23 shows a flow of a standard load curve creation process using an immediate sample model.

  In the immediate sample model, first, the amount of power per unit time on the target day (x year y month z day) is measured by a real time meter or interval meter installed in a sample consumer (S101). The sample customer is a customer set by the demand sampling survey design process (S2).

  Subsequently, in the immediate sample model, the measured electric energy of each sample consumer is totaled for each unit time, and a total load curve is created (S102). It should be noted that the weighted average may be taken in consideration of the demand scale (contract power) of the sample consumer as required.

  Subsequently, in the immediate sample model, the obtained load curve of the sample customer is normalized based on the contract power of the sample customer, and this is used as the standard load curve on the target date (x year y month z day). (S103).

  In the immediate sample model, a standard load curve for each day is generated by performing the above-described processing every day.

--Driving pattern model--
The driving pattern model is a model in which a load pattern is modeled in advance for a consumer whose operation pattern such as a streetlight is certain to some extent, and is used as it is as a standard load curve.

  Consumers that can be used as driving pattern models include, for example, street lamps and traffic lights. Street lights turn on at sunset and turn off at sunrise. For this reason, the load pattern of the street lamp is constant at a predetermined value after sunset, and is almost constant at 0 after sunrise. Therefore, as shown in FIG. 24A, it is possible to model as an operation pattern using the sunrise and sunset times for each region as variables. The traffic light operates throughout the day. Accordingly, as shown in FIG. 24 (b), the power consumption is always almost constant. That is, it can be modeled as an operation pattern in consideration of the rated capacity and load factor of the device.

--Similar day model--
In the similar day model, as shown in FIG. 25, the standard load curve of each day (x year y month z day) is obtained for each load pattern category, as in the immediate sample model (S101 to S103).

  Further, in the similar day model, external factors such as measurement day of the week, temperature, and weather are created as indexes (S104), and the created indexes and standard load curves are associated with each other to create a database (S105).

  FIG. 26 shows a database example of standard load curves associated with indexes. In this database, various indexes such as date, day of week, temperature, weather, and system demand can be used as indexes, and these can be selected arbitrarily by the user. For categorical variables such as days of the week and weather, for example, fine = 0, cloudy = 1, rain = 2, snow = 3, Monday-Friday = 0, Saturdays, Sundays, and holidays = 1 after considering the influence on the power load. And so on.

  In the similar day model, a past day load curve close to the condition of the day (target day) for which a specific customer's load curve is desired is extracted from the database and used as a standard load curve.

  As shown in FIG. 27, the standard load curve is extracted from the database by calculating the similarity between the index value (S111) of each past day in the database and the conditions (S112) of the target day for which the load curve is to be obtained. The calculation is performed (S113), and the load curve on the day with high similarity is extracted as the standard load curve (S114).

  The similarity can be calculated by, for example, the following formula (14).

  Next, setting examples (Case 1 to Case 3) of standard load curve extraction conditions are shown.

  In setting case 1, the standard load curve for the past day in which the load curve estimation target day, the month, and the day of the week match is extracted from the database and used. When there are a plurality of extraction results, averaging is performed.

  In setting case 2, a standard load curve corresponding to a past day similar to the demand pattern of all systems on the target day of load curve calculation (see the description in the common shape model) is extracted from the database and used. In this case, the hourly value of the system demand index shown in FIG. 26 can be extracted by specifying each condition variable in the above equation (14).

  In setting case 3, a standard load curve for a past day similar to the weather condition on the target day for calculating the load curve is retrieved from the database and used. In this case, it is possible to extract by specifying the weather value (temperature, weather, etc.) of the past day and the weather value of the target day as the condition variables of the above equation (14). When the target date for calculating the load curve is the future (for example, the next day), the predicted weather value may be used as the weather value for that day. Further, when the weather condition is the weather or the like, as shown in FIG. 28, a value obtained by digitizing each weather condition in advance is used.

--Demand modeling model--
The demand modeling model creates a prediction formula that formulates the relationship between the values at each time of the standard load curve and other factors for each load pattern category based on statistical data of past standard load curve values and the same factors. This is a method of obtaining the standard load curve on the target day by substituting the value of the factor on the target day for calculating the load curve.

  Formulation methods used for the demand modeling model include “regression model”, “time series model”, “neural network”, and the like. Here, the “regression model” will be described as an example.

  In the regression model, first, a demand modeling model is created in which the value at each time of the standard load curve is an objective variable, and the influencing factors such as temperature are explanatory variables, as in the following equation (15).

  For example, as a relatively simple model, a regression model having temperature and day of the week flags as explanatory variables as shown in the following equation (16) can be used.

  In the estimation method (regression coefficient value, etc.) based on the demand modeling model, a load curve of sample customers collected in advance is used. In this case, in order to make the output of the created demand modeling model a standard load curve (normalized load curve), the obtained sample customer's load curve is processed with the above-mentioned immediate sample model. Similarly, after summing up the load curves of sample customers every day, normalize them based on contracted power, etc., and use this normalized load curve as past data to determine the relationship between this and explanatory variables such as weather factors. What is necessary is just to model by said Formula (15).

6 Individual load curve estimation processing (S5)
Next, detailed processing contents of the individual load curve estimation processing S5 will be described. The individual load curve estimation process S5 is executed by the load curve creation main computer 11.

  In the individual load curve estimation process (S5), as shown in FIG. 6, a target customer (target consumer) for performing load curve estimation and a load curve estimation target period are set (S15), and then each target demand is set. A load pattern category is assigned to the house (S16), then a standard load curve is calculated for the estimation target period set based on the standard load curve model for the load pattern category (S17), and finally the calculated standard load is calculated. Correction processing corresponding to the demand scale of each consumer is performed on the curve (S18). This makes it possible to create an individual load curve for each consumer.

--Standard load curve calculation process (S17)-
First, regarding the processing content of the standard load curve calculation process (S17), for each standard load curve model ("common shape model", "immediate sample model", "operation pattern model", "similar day model", "demand modeling model" )).

(Common shape model)
The flow of step S17 in the individual load curve estimation process using the common shape model will be described with reference to FIG.

  In the common shape model, first, the measurement period of the standard load curve and the load pattern classification of the consumer that is the target of load curve creation are set (S211).

  Subsequently, a system range to be referred to as a common shape model is set (S212). Here, when the load curve of a part of the systems is used as the system range, the process proceeds to the subsequent step S213, and when the load curves of all the systems are used, the process proceeds to the subsequent step S214.

  Subsequently, the actual value of the load curve of each system is collected (S215, S216). The actual value is measured by a power measuring meter installed in the power plant or substation or a power measuring meter set in the substation and stored in the system demand storage unit 27.

  Subsequently, when there is a consumer (meter installation consumer) in which a real-time meter or an interval meter is installed in each designated system, an actual measurement value of the load curve is collected. By subtracting the load curve of the meter installation consumer from the collected system load curve, the total load curve of the consumers other than the meter installation consumer is calculated (S217).

  Subsequently, the total load curve is normalized using the above formula (13) based on the contracted power of the consumer other than the meter installed consumer, the meter reading result by the integrating meter, and the standard load curve is created. (S218).

  Finally, the obtained standard load curve is stored in the standard load curve temporary storage unit 20 (S219).

(Immediate sample model)
The flow of step S17 in the individual load curve estimation process in the immediate sample model will be described with reference to FIG.

  In the immediate sample model, first, the measurement period of the standard load curve and the load pattern classification of the customer who is the load curve creation target are set (S221).

  Subsequently, the actual measurement values of the load curve during the measurement period of the sample consumer for the load pattern classification are collected and summed (S222).

  Subsequently, normalization processing is performed using the above formula (13) and the like based on the contract power of the sample customer and the like, and a standard load curve is created (S223).

  Finally, the obtained standard load curve is stored in the standard load curve temporary storage unit 20 (S224).

(Driving pattern model)
The flow of step S17 in the individual load curve estimation process using the operation pattern model will be described with reference to FIG.

  In the operation pattern model, first, the measurement period of the standard load curve and the type of consumer (type of power demand: electrical equipment, etc.) to be created are set (S231).

  Subsequently, an operation pattern model corresponding to the consumer is called (S232).

  Subsequently, the values of variables (sunrise, sunset time, etc.) necessary for calculating the driving pattern model for the target period for the called driving pattern model are input (S233). These input values can be input directly by the user via the input interface, or converted into data in advance in the computer together with the calendar information as a standard load curve model in advance, and this can be given automatically.

  Subsequently, an operation pattern model is calculated based on the input result, and this is used as a standard load curve (S234).

  Finally, the obtained standard load curve is stored in the standard load curve temporary storage unit 20 (S235).

(Similar day model)
The flow of step S17 in the individual load curve estimation process in the similar day model will be described with reference to FIG.

  In the similar day model, first, the measurement period of the standard load curve and the load pattern classification of the customer who is the target of load curve creation are set (S241).

  Subsequently, variables and weighting factors used for the similarity distance calculation of the above equation (14) for the load pattern classification are set (S242). The variable and the weighting factor can be set directly by the user via the input interface of the system, or values stored in advance as a standard load curve model can be used.

  Subsequently, the standard load curve for each day obtained from a past sample customer is placed on the extraction target list for the target load pattern classification (S243).

  Subsequently, the similarity (similar distance) between each past day in the extraction target list and each day in the previously specified measurement target period is calculated, and a past day having a small similar distance is extracted (S244). .

  Subsequently, the standard load curve is loaded from the database for the extracted past day (S245).

  Subsequently, when there are a plurality of past days extracted, the standard load curves are averaged to obtain one standard load curve (S246).

  Finally, the obtained standard load curve is stored in the standard load curve temporary storage unit 20 (S247).

(Demand modeling model)
The flow of step S17 in the individual load curve estimation process in the demand modeling model will be described with reference to FIG.

  In the demand modeling model, first, the measurement period of the standard load curve and the load pattern classification of the customer who is the target of load curve creation are set (S251).

  Subsequently, a demand modeling model corresponding to the corresponding load pattern classification is called (S252).

  Subsequently, the value for each time within the measurement period set in step S251 of each variable used as the explanatory variable of the demand modeling model is set (S253). In addition to being directly input by the user via the input interface, the setting can be automatically set based on variable values of the measurement period stored in the databases 24 and 25 such as related weather and demand.

  Subsequently, a demand value at each time of the measurement period is calculated based on the demand modeling model, and a standard load curve is obtained (S254).

  Finally, the obtained standard load curve is stored in the standard load curve temporary storage unit 20 (S247).

(Online processing, etc.)
In the standard load curve creation process as described above, three processes of “offline process”, “online process”, and “prediction process” can be applied.

  The off-line process is a process for creating a standard load curve for a certain period or a specific date going back in the past, and creating a load curve for each customer based on the standard load curve.

  The online process is a process of creating a standard load curve corresponding to the present or very close past (within several hours or the like) and creating a load curve of each customer based on the standard load curve.

  The prediction process is a process of creating a standard load curve for a certain period in the future or a specific day, and creating a load curve for each customer based on the standard load curve.

  Each model of “common shape model”, “immediate sample model model”, “driving pattern model”, “similar day model”, and “demand modeling model” can correspond as shown in Table 1 below.

--Calculation target setting process (S15)-
Next, the calculation target setting process (S15 in FIG. 6) of the individual load curve estimation process S5 will be described.

  FIG. 34 shows the flow of the calculation target setting process (S15).

  In the calculation target setting process, first, a customer name registered in advance or designated directly by the user via an interface or the like is input (S15-1).

  Subsequently, in the calculation target setting process, based on the customer name, from the customer information storage unit 26, the customer's contract type, contract power load rate, electricity usage, maximum power, customer industry, which are the target load curve creation targets Data necessary for setting the load pattern classification such as the area is read (S15-2).

  Subsequently, in the calculation target setting process, the load pattern classification, the load curve creation period, the unit time, etc. assigned to the load curve creation target consumer are set, and these calculation conditions are temporarily saved (15). -3).

--Load pattern classification assignment processing (S16)-
Next, the load pattern classification assignment process (S16 in FIG. 6) of the individual load curve estimation process S5 will be described.

  FIG. 35 shows the flow of the load pattern category assignment process (S16).

  In the load pattern classification assignment process, first, the calculation conditions stored in the calculation target setting process (S15) are read (S16-1).

  Subsequently, in the load pattern category assignment process, the read calculation condition and the load pattern setting condition stored in the load pattern category storage unit 21 are compared to determine the load pattern category of the target consumer ( S16-2).

  Subsequently, in the load pattern category assignment process, the confirmed load pattern category is added to the temporarily stored calculation conditions in correspondence with information such as the customer name (S16-3). The temporarily stored conditions are used as conditions for calculating the standard load curve (S17) described above.

--Correction process according to customer demand scale (S18)-
Next, the correction process (S18 in FIG. 6) according to the demand scale of the customer in the individual load curve estimation process S5 will be described.

  FIG. 36 shows the flow of the correction process (S18) according to the demand scale of the customer.

  In the correction process according to the demand scale of the consumer, first, in the correction process according to the demand scale of the consumer, the standard load curve calculated in the calculation of the standard load curve (S17) is called (S18-1).

  Subsequently, in the correction process according to the demand scale of the consumer, the standard load curve is a normalized value, so the correction according to the scale of each individual consumer is performed, and the load curve of the target consumer is targeted. Is calculated (S18-2).

  For example, when the contract power of Expression (13) is used, as shown in the following Expression (17), the final load curve is calculated by multiplying the contract power of each consumer by the standard load curve.

  Finally, in the correction process according to the demand scale of the consumer, the load curve obtained by the correction is stored in the load curve estimation result storage unit 29 (S18-3).

  In the description of the correction process (S18) according to the demand scale of the customer and the description of the standard load curve model, the method using the contract power as the normalization method has been described. It is also possible to use meter reading values (such as sample consumers and load curve target consumers) (monthly power consumption, annual power consumption, annual maximum monthly power consumption, etc. instead of contract power). However, the method using contract power makes it possible to create a load curve more easily and at a lower cost than the method of measuring the load curve of each individual customer with a watt-hour meter. It is also possible to measure the load curve corresponding to.

7. Effect of Load Curve Estimation System to Which Present Invention is Applied As described above, in the load curve estimation system, a standard load curve is generated for each load pattern category, and the load pattern category to which the estimation target consumer belongs. And the load curve of the consumer is calculated using the standard load curve of the specified load pattern classification.

  As a result, the load curve estimation system can accurately and easily estimate the power consumption for each consumer without installing expensive measuring instruments such as interval meters and real-time meters in each consumer.

  In addition, although the system which measures the demand amount of electric power was demonstrated as embodiment of this invention, this invention is applied not only to electric power but the system which calculates consumption of resources, such as gas and water. Can do.

  In other words, it is consumed not only by electric power but also by consumers such as gas and water, households, factories, shops, etc., and from the supply source (power plant or substation, gas tank, water storage facility) to the physical Since the demand amount can be estimated as long as it is supplied via a network (power transmission line, gas pipe, water pipe) connected to each other, the load curve estimation system described above can be applied.

  The present invention can be configured not only by hardware resources but also by software programs that can be executed by a computer.

It is the figure which showed the structure of the electric power grid | system. It is the figure which showed an example of the load curve. It is a figure which shows the process sequence of the load curve calculation method to which this invention was applied. It is a system block diagram of a load curve estimation system. It is a figure for demonstrating a load curve estimation preparation process. It is a figure for demonstrating an individual load curve estimation process. It is a flowchart of a load pattern division design process. It is a figure which shows an example of designation | designated of a customer range. It is a figure which shows an example of a division | segmentation design parameter | index. It is a figure which shows an example of a load pattern division. It is a flowchart which shows the process sequence for the division | segmentation design at the time of selecting the aggregation algorithm. It is a figure which shows the example which made the numerical variable the categorical variable. It is a figure which shows the fusion process with respect to the number of clusters. It is a figure which shows the relationship between the number of clusters and an evaluation value. It is a flowchart which shows the process sequence for the division | segmentation design at the time of selecting a division | segmentation algorithm. It is a figure which shows the division | segmentation process with respect to the number of groups. It is a figure which shows the relationship between the number of groups and an evaluation value. It is a figure which shows the division | segmentation process with respect to the number of groups at the time of using a some division | segmentation design parameter | index. It is a flowchart which shows the procedure of a demand sampling investigation design process. It is a figure which shows the example of a setting of a sample extraction frame. It is a figure which shows the customer information of the state to which the flag which identifies a sample consumer was attached. It is a figure for demonstrating the algorithm of the common shape model among standard load curve models. It is a figure for demonstrating the algorithm of the immediate sample model among standard load curve models. It is the figure which showed the operation | movement pattern of the demand which can be modeled by the driving | running pattern model of a standard load curve model. It is a figure for demonstrating the algorithm of the similar day model among standard load curve models. It is a figure which shows the example of a database of the standard load curve matched with the index. It is the figure which showed the process which extracts a standard load curve from a database. It is the figure which showed the example which digitized each condition of the weather. It is the flowchart which showed the flow of the calculation process of the standard load curve at the time of using a common shape model. It is the flowchart which showed the flow of the calculation process of the standard load curve at the time of using an immediate sample model. It is the flowchart which showed the flow of the calculation process of the standard load curve at the time of using an operation pattern model. It is the flowchart which showed the flow of the calculation process of the standard load curve at the time of using a similar day model. It is the flowchart which showed the flow of the calculation process of the standard load curve at the time of using a demand modeling model. It is the flowchart which showed the flow of the calculation object setting process. It is the flowchart which showed the flow of the allocation process of a load pattern classification. It is the flowchart which showed the flow of the correction process according to the demand scale of a consumer.

Explanation of symbols

  10 Load curve estimation system, 12 servers, 20 standard load curve temporary storage unit, 21 load pattern classification storage unit, 22 sample design result storage unit, 23 standard load curve model storage unit, 24 demand survey data storage unit, 25 weather information etc. Storage unit, 26 Customer information storage unit, 27 System demand storage unit, 28 Meter reading record storage unit, 29 Load curve estimation result storage unit

Claims (11)

In the resource consumption estimation method for estimating the consumption of resources supplied from the supplier to the customer via the network,
Divide multiple customers into groups with similar consumption patterns,
Generate a standard load curve that represents the resource consumption pattern of the group for each group.
A resource characterized by identifying a group to which an estimation target consumer belongs and estimating a resource consumption per unit time within an arbitrary period for the estimation target consumer using a standard load curve of the specified group How to estimate consumption. When generating a standard load curve for each group, select a sample customer belonging to that group, measure the consumption pattern of the selected consumer, and based on the measured load pattern of each consumer, The resource consumption estimation method according to claim 1, wherein a standard load curve is generated. The resource consumption estimation method according to claim 1, wherein the standard load curve is stored together with an index representing an attribute of a customer included in the group. Claims characterized in that the resource consumption of each network system is measured, and a pattern generated based on the measured values is used as a standard load pattern for a group of customers included under the system. Item 1. The resource consumption estimation method according to Item 1. Measure the resource consumption per unit time of one or more consumers as samples belonging to an arbitrary group, and use the pattern generated based on the measured value as the standard load curve of the group The resource consumption estimation method according to claim 1. The resource consumption estimation method according to claim 1, wherein a standard load curve is formed in accordance with the operation pattern for a consumer whose operation pattern can be estimated. The standard load curve is associated with external factors that are related to resource consumption,
The external factor of the estimation target consumer or the estimation target period is specified, and the resource consumption is estimated using a standard load curve associated with the specified external factor. The method for estimating resource consumption according to 1. The resource consumption estimation method according to claim 1, wherein the standard load curve is modeled by a prediction formula. The standard load curve of the identified group is corrected according to the consumption scale or integrated measurement value of the consumer, and the corrected value is used as a resource amount per unit time consumed by the consumer. To estimate resource consumption of food. In a resource consumption estimation device that estimates the consumption of resources supplied from a supplier to a customer via a network,
A storage unit that divides a plurality of consumers into groups with similar resource consumption patterns, and stores standard load curves representing resource consumption patterns for each group;
The group to which the estimation target consumer belongs is identified, the standard load curve of the identified group is acquired from the storage unit, and the resource per unit time within an arbitrary period for the estimation target consumer based on the acquired standard load curve A resource consumption estimation device comprising: an estimation processing means for estimating consumption. In the resource consumption estimation program, which is a computer program that estimates the consumption of resources supplied from the supplier to the customer via the network,
Divide multiple customers into groups with similar resource consumption patterns, store standard load curves representing resource consumption patterns for each group,
The group to which the estimation target consumer belongs is identified, the standard load curve of the identified group is acquired from the storage unit, and the resource per unit time within an arbitrary period for the estimation target consumer based on the acquired standard load curve A resource consumption estimation program characterized by estimating consumption.

Images