JP7817872B2 - Sensor placement management system and sensor placement management method - Google Patents

Description

The present invention relates to power analysis, and more particularly to techniques for managing sensor placement in power data collection.

Power analysis services are available for buildings such as commercial stores, factories, and homes, which obtain information on the current and power consumed and suggest improvements to power consumption within the building.

In other words, by utilizing time-series data on the power, current, and voltage generated or consumed by electrical equipment operating within a building (hereafter simply referred to as electrical signals), it is possible to understand the actual operation of that electrical equipment, which can be used to plan measures for saving energy and improving operations. For example, by collecting and analyzing the power consumption of lighting equipment in a home, it is possible to detect whether the lights have been left on. Similarly, by collecting and analyzing the power consumption of cooking equipment in a retail store, it is possible to estimate the time when products were cooked, providing information that is useful for inventory management.

Power analysis services need to be provided cheaply and quickly, and the challenge to achieve this is to collect power data from the electrical equipment operating within each building.

One way to realize such a service would be to first install a sensor on each piece of electrical equipment operating within a building and collect electrical signals. However, if a sensor is installed on each piece of electrical equipment to be analyzed, the number of sensors required increases depending on the number of electrical equipment, which creates the problem of increased costs for sensor installation work and the collection and storage of electrical signals.

The other method uses NILM sensors to measure the power of the main circuit at the entrance to the power building, and a probabilistic generation model to estimate the operating status of the electrical equipment within the building. This method using NILM sensors is inexpensive, as only one or two power sensors are required per store, but in order to develop a probabilistic generation model, it is necessary to install an electrical sensor for each piece of electrical equipment and collect data.

Therefore, it is desirable to provide power analysis services to a small number of buildings by applying electrical sensors to each electrical device, collecting data, and then using that data to create a probability generation model with reasonable accuracy, which can then be deployed to a large number of buildings using NILM sensors.

The following patent documents are considered background art in this technical field. Patent Document 1 (JP 2013-213825 A) describes a method for monitoring an electrical device having a factor corresponding to the first electrical device, the factor having three or more states, and includes the steps of acquiring data representing the sum of electrical signals from two or more electrical devices, including a first electrical device, processing the data using a probabilistic generation model to generate an estimate of the operating state of the first electrical device, and outputting the estimate of the electrical signal from the first electrical device, the probabilistic generation model being a factor corresponding to the first electrical device.

Furthermore, Patent Document 2 (JP 2013-218715 A) describes an electrical equipment estimation device that includes a data acquisition means for acquiring time-series data on the total current consumption of multiple electrical equipment, and a parameter estimation means for determining model parameters when the operating states of the multiple electrical equipment are modeled using a probabilistic model based on the acquired time-series data.

Non-Patent Document 1 discloses a method for managing the relationship between electrical device models and probability generation models as meta-information.

Japanese Patent Application Laid-Open No. 2013-213825 JP 2013-218715 A

J. Kelly and W. Knottenbelt, "Metadata for Energy Disaggregation," in 2014 IEEE 38th International Computer Software and Applications Conference Workshops, Vasteras, Sweden, Jul. 2014, pp. 578-583.

In order to estimate parameters using the techniques disclosed in Patent Documents 1 and 2, it is necessary to install sensors in advance on individual electrical devices, collect electrical signals as training data, and then use this training data to create in advance a probabilistic generation model appropriate for the type of electrical device.

Depending on the building in which the distribution board is installed, even if the same model of electrical equipment is used, the shape of the electrical signal may vary significantly if the installation environment or usage pattern of the electrical equipment is significantly different. In such cases, a probabilistic generation model created in another building with a significantly different installation environment or usage pattern cannot be used to estimate the operating status of the electrical equipment in that building. Furthermore, while it is possible to accurately estimate the operating status of electrical equipment if the probabilistic generation model is created and used in the same building, this requires installing electrical sensors on the electrical equipment in all buildings to collect learning data, which poses the issue of increasing the number of sensors required.

Here, we will continue our explanation by referring to buildings such as commercial stores, factories, and homes as a broader concept, and using the term "power consumption object." A building that installs sensors for each electrical device used inside it and collects its electrical signals to collect learning data for creating a probabilistic generation model will be called an experimental power consumption object, while a building that basically only collects electrical signals from the main circuit in order to use the probabilistic generation model will be called a general power consumption object. A general power consumption object basically only collects electrical signals from the main circuit, but it may also collect electrical signals from other circuits.

By dividing the installation environments and usage patterns of electrical equipment in general power consumption objects into several types and using a probabilistic generation model created with experimental power consumption objects of the same type, it is possible to accurately estimate operating conditions. Furthermore, if many types can be represented with a small number of experimental power consumption objects, the number of sensors required to collect training data can be reduced.

The present invention aims to realize a process for extracting general power consumption targets that have different electrical equipment installation environments and usage patterns from existing experimental power consumption targets as candidates to be added to a group of experimental power consumption targets, without installing sensors on the electrical equipment that are general power consumption targets.

A representative example of the invention disclosed in the present application is as follows: That is, a sensor placement management system includes a computing device that performs computational processing and a storage device accessible by the computing device, wherein the computing device has a candidate power consumption target calculation unit that uses, as attribute information of the experimental power consumption targets, business information including at least one of the addresses, locations, and business hours of the power consumption targets, the business information correlating with electrical signals measured from experimental power consumption targets that are power consumption targets measuring electrical signals from electrical sensors installed in internal electrical equipment, to cluster the experimental power consumption targets and general power consumption targets, and calculates a rank of the general power consumption targets as candidates for new experimental power consumption targets within the cluster to which the general power consumption targets belong ; and an output unit that outputs the candidate experimental power consumption targets with the lowest rank.

According to one aspect of the present invention, the training data required to create a probabilistic generation model that estimates the operating state of an electrical device from the electrical signal of the main circuit can be collected using a small number of sensors. Issues, configurations, and advantages other than those described above will become clearer in the following description of the embodiments.

1 is a diagram illustrating an example of a network configuration including a sensor placement support system according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of the hardware configuration of a sensor placement assistance server according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of the configuration of business information according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of the configuration of an electrical signal according to an embodiment of the present invention. FIG. 10 is a diagram illustrating an example of the configuration of correlation attribute information according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of the configuration of store attribute information according to an embodiment of the present invention. FIG. 4 is a diagram showing an example of the configuration of candidate store information according to an embodiment of the present invention. FIG. 10 is a diagram illustrating an example of the configuration of similarity information according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of the configuration of recommendation information according to an embodiment of the present invention. FIG. 10 is a diagram illustrating an example of a flowchart of a sensor placement assistance server according to an embodiment of the present invention. FIG. 10 is a diagram showing an example of the configuration of a recommendation screen according to an embodiment of the present invention. FIG. 10 is a diagram showing an example of the configuration of a recommendation screen according to an embodiment of the present invention.

Specific embodiments of the present invention are described below with reference to the drawings.

In the sensor placement management system of an embodiment of the present invention, a store (building) such as a convenience store will be described as an example of a power consumption target that can be supported. A variety of electrical equipment is used in a store, such as lighting equipment, heating and cooling equipment, and cooking equipment. As mentioned above, power consumption targets may include various facilities that are units of power consumption, such as factories, homes (detached houses, apartment buildings), and office buildings, in addition to the commercial stores exemplified in the embodiment.

<System Configuration>
A system configuration according to an embodiment of the present invention will be described with reference to an example.
<Example>

An example of the present invention will be described.

An embodiment of the present invention may be a sensor placement management system that includes a computing device 403 that executes calculations and a storage device accessible to the computing device 403, a store attribute calculation unit 110 that calculates the attributes of experimental stores that have a high correlation with the electrical signals through correlation analysis between the experimental store's business information 211 and the electrical signals 212 measured from electrical sensors installed on electrical equipment in the experimental store, a candidate store calculation unit 120 that clusters the experimental stores and general stores using the attribute information and calculates a ranking of the general stores as candidates for new experimental power consumption targets, and an output unit that outputs candidate experimental stores with low rankings. Note that a high correlation may mean that the correlation is relatively high.

Figure 1 is a diagram showing an example configuration of a sensor placement management system according to one embodiment of the present invention. The sensor placement management support system shown in Figure 1 has a sensor placement support server 100, a store information management server 200, and a display unit 300 as an output unit, as well as a computing device 403 that executes calculations (not shown), and a storage device 401 that can be accessed by the computing device 403. The sensor placement support server 100 is connected to be able to communicate with the store information management server 200 and the display unit 300 used by users of the sensor placement support system.

The store information management server 200 stores business information 211 and electrical signals 212 for each experimental store and general store, which will be described later. The sensor placement support server 100 has a store attribute calculation unit 110, a candidate store calculation unit 120, an electrical signal acquisition unit 130, and an experimental store recommendation unit 140. An example of the hardware configuration of the sensor placement support server 100 is shown in Figure 2 and is described in the hardware configuration example described later. Examples of the business information 211 and electrical signals 212 are shown in Figures 3 and 4 and are described in the data structure example described later.

In order to collect learning data for creating a probabilistic generation model, a store building where sensors are installed for each electrical device used within the store and the electrical signals are collected is called an experimental store. Furthermore, a store building where only electrical signals from the main circuit are collected in order to use the probabilistic generation model is called a general store.

The store attribute calculation unit 110 acquires business information 211 for each store and electrical signals 212 measured for each electrical device in the store from the store information management server 200. The correlation analysis unit 111 calculates correlation attributes between the business information 211 and electrical signals 212 acquired by the store attribute calculation unit 110 and stores the calculated correlation attributes in the correlation attribute information 112. The correlation analysis unit 111 calculates correlation coefficients between the business information and the electrical signals of the electrical devices using well-known methods such as Pearson's correlation coefficient or point biserial correlation coefficient, and records a predetermined number of pieces of business information for each electrical device in the correlation attribute information 112. An example of the correlation attribute information 112 is shown in FIG. 5 and is described in the data structure example below. The business information 211 and electrical signals 212 for which the correlation analysis unit 111 calculates correlation attributes may be for all business information 211 and electrical signals 212 acquired by the store attribute calculation unit 110. Alternatively, the correlation analysis unit 111 may calculate correlation attributes for only the range of business information 211 and electrical signals 212 required for the correlation attribute information 112. Furthermore, the predetermined number of pieces of business information recorded in the correlation attribute information 112 may be recorded for each electrical device in descending order of correlation coefficient.

The candidate store calculation unit 120 selects general stores that are candidates for new experimental stores, i.e., candidate stores. The candidate store calculation unit 120 acquires business information 211 of general stores and experimental stores related to the attributes listed in the correlation attribute information 112 using the attribute information acquisition unit 121, and records this information in store attribute information 122. Here, the business information 211 of general stores and experimental stores acquired by the attribute information acquisition unit 121 may be business information 211 of all general stores and experimental stores, or may be information required for the clustering performed by the clustering unit 123. The clustering unit 123 performs clustering based on the attributes listed in the store attribute information 122 and the distance between stores, and records information on the cluster to which each store belongs, nearby experimental stores, and the distance from the experimental store in candidate store information 124. An example of store attribute information 122 is shown in Figure 6 and will be explained in the data structure example below.

The clustering calculation method can be any clustering method, such as K-Means clustering or a Gaussian mixture model. Furthermore, the distance can be defined in any way, such as Euclidean distance or Mahalanobis distance.

As a result of clustering performed by the clustering unit 123 described above based on the attributes described in the store attribute information 122 and the distance between stores, stores are divided into several clusters, for example, depending on the correlation between store attributes and the correlation between the distance between stores.

Various methods can be used to select general stores as new candidate experimental stores. For example, in candidate calculation example 1, the sum of the distances between each general store and other general stores within each cluster can be calculated, and a predetermined number of general stores can be recorded in candidate store information 124 as new experimental store candidates in descending order of sum of distances. The candidates recorded in candidate store information 124 are sent to display unit 300, which serves as the output unit, and displayed and output on display unit 300 for use by the sensor placement manager. By selecting a new experimental store as a candidate in this way, it becomes easier to create a probability generation model for a distinctive store that is relatively dissimilar to other general stores, making it easier to predict power consumption using NILM sensors. In this example, a distinctive store that is relatively dissimilar to other general stores is selected as a candidate. Therefore, if this candidate is selected as an experimental store, it is expected that a probability generation model that is more suited to the candidate can be used compared to when other general stores are selected as experimental stores, making it a useful candidate for a new experimental store.

As a second example of candidate calculation, the sum of the distances between each general store and other general stores within each cluster may be calculated, and a predetermined number of general stores may be recorded in the candidate store information 124 as new experimental store candidates in ascending order of sum of distances. The candidates recorded in the candidate store information 124 are sent to the display unit 300, which serves as the output unit, and are displayed and output on the display unit 300 for use by the sensor placement manager. In this way, by setting a new experimental store as a candidate, it becomes easier to create a probability generation model for a distinctive store that is relatively similar to other general stores, making it easier to predict power consumption using NILM sensors. In this example, a distinctive store that is relatively similar to other general stores is selected as a candidate. Therefore, if this candidate is set as an experimental store, it is expected that the probability generation model can also be used for many general stores similar to the candidate, resulting in high utilization efficiency as a candidate.

Furthermore, as candidate selection example 3, instead of calculating the sum of the distances between each general store and other general stores in candidate selection example 1, the average distance between each general store and other general stores within each cluster can be calculated, and general stores whose average distance is greater than a predetermined distance from the general store to its nearest experimental store can be recorded in candidate store information 124 as candidates for new experimental stores, and displayed and output by display unit 300, which is the output unit. In this configuration, general stores whose average distance between them is relatively farther than the nearest existing experimental store can be set as candidates for new experimental stores. Therefore, it is expected that a probability generation model that is more suited to the candidates can be used for other general stores around the new experimental store, making them useful as candidates for new experimental stores.

The new method for calculating experimental store candidates described above may calculate candidate stores based on distance from the results of clustering. In addition to the method described above, various other methods can be used to calculate new experimental store candidates. For example, while the method described here selects candidates from general stores within the same cluster, it is also possible to select candidates across multiple clusters.

The results stored in the candidate store information 124 can also be sent to the display unit 300, which is the output unit, and displayed, i.e., output.

With the above-described configuration, the learning data required to create a probability generation model for estimating the operating state of an electrical device from the electrical signal of the main circuit can be collected with a small number of sensors.
<Hardware configuration example>

Next, we will explain an example of the hardware configuration of the sensor placement assistance server 100.

Figure 2 is a configuration diagram showing an example of the hardware configuration of the sensor placement assistance server 100. As shown in Figure 2, the sensor placement assistance server 100 has a storage device 401, a computing device 403, a memory 404, and a communication device 405, and each unit is interconnected via a bus.

The storage device 401 may be composed of a non-volatile storage element such as an SSD (Solid State Drive) or a hard disk drive. The storage device 401 may store a program 402 that defines the operation of the computing device 403, and various information used or generated by the computing device 403, such as correlation attribute information 112, store attribute information 122, candidate store information 124, similarity information 142, and recommendation information 144. The memory 404 may be composed of a volatile storage element such as RAM (Random Access Memory).

The arithmetic unit 403 may be configured with a processor such as a CPU (Central Processing Unit). The arithmetic unit 403 executes the program 402 stored in the storage device 401 by, for example, reading it into the memory 404. This enables the functions of the correlation analysis unit 111, attribute information acquisition unit 121, clustering unit 123, electrical signal acquisition unit 131, similarity calculation unit 141, and recommendation unit 143 shown in FIG. 1 to be realized. The communication unit 405 can communicate with an external device such as the store information management server 200 shown in FIG. 1 via the network 406.

<Data structure example>
Next, an example of a data structure used in the embodiment will be described. In the following description of the data structure, the data is expressed in a tabular format. In this tabular format, each column of the table is called a field. The reference number of the field and the reference number representing the name of the data written in that field are the same to the extent that confusion does not occur.
FIG. 3 is a diagram showing an example configuration of business information 211. The business information 211 shown in FIG. 3 may have fields 211a to 211e. Field 211a may store a store name, which is identification information for identifying a store. This store name may include not only experimental stores collecting electrical signals 212 from sensors installed on electrical equipment, but also other general stores. Field 211b may store an address indicating the location of the store. Field 211c may store location, which is information indicating the environment surrounding the store. Field 211d may store equipment configuration, which is a list of identification information for electrical equipment operating in the store. Field 211e may store business hours, which is information indicating the hours the store is open. Each field in FIG. 3 is an example of business information in the present disclosure and is not limited to this example and may include any information that can correlate to the electrical signals generated by electrical equipment. Furthermore, the relationship between the information stored in FIG. 3 and the fields is not limited to this example; any information can be stored in any field.

Figure 4 is a diagram showing an example configuration of the electrical signal 212. The electrical signal 212 shown in Figure 4 may have fields 212a to 212d. Field 212a may store the store name, which is identification information for identifying the store. The store name may include not only the experimental store collecting the electrical signal 212 from sensors installed on electrical equipment, but also other general stores. Field 212b may store the device name, which is identification information for identifying the electrical equipment in the store. The device name may also include the main circuit, which is the sum of the electrical signals of the electrical equipment in the store. Furthermore, if the store has multiple distribution boards, the multiple main circuits may be distinguished and stored in a format such as main circuit 1, main circuit 2, etc. Field 212c may store the time when the electrical signal of the device was measured. Field 212d stores the measured value of the electrical signal of the device at that time. Specifically, this electrical signal may be, for example, a measured value of a signal related to an electrical circuit, such as voltage, current, or power. The unit of the measurement value may be, for example, V for voltage, A for current, or W for power. This electrical signal may be of two or more types, such as voltage and current. This electrical signal can be measured every minute for a measurement period of one day, or every hour for a measurement period of one month, and any measurement interval and measurement period can be chosen depending on the application.

Figure 5 is a diagram showing an example configuration of the correlation attribute information 112. The correlation attribute information 112 shown in Figure 5 may have fields 112a to 112b. In this example, field 112a may store the equipment name, which is identification information for identifying the electrical equipment. In this example, this equipment name does not include the main circuit, and only information identifying the electrical equipment is stored. Field 112b may store an attribute correlated with the equipment as an "attribute name." This attribute is composed of information contained in the column names of each field of the business information 211. For example, the second row of the correlation attribute information 112 shown in table format in Figure 5 has "Air Conditioning" in the equipment name column and "{Location, Business Hours}" in the attribute name column. This configuration may be used because the correlation analysis unit 111 of the store attribute calculation unit 110 performed a correlation analysis with electrical signals for each electrical equipment, and found that the equipment name "Air Conditioning" has a relatively high correlation with "Location" and "Business Hours." For example, in the third line of Figure 5, the equipment name column has "freezer" and the attribute name column has "{location, address}". This may also mean that the equipment name "freezer" has a relatively high correlation with "location" and "address". Here, the number of attributes stored in the attribute name field 112b can be any number as needed.

Figure 6 is a diagram showing an example configuration of store attribute information 122. The store attribute information 122 shown in Figure 6 may have fields 122a to 122d. Field 122a may store the store name, which is identification information for identifying the store. This store name may include not only experimental stores that collect electrical signals 212 from sensors installed on electrical equipment, but also other general stores. Field 122b may store the equipment name, which is identification information for identifying equipment in the store. Field 122c may store an attribute value indicating the attributes of the store. Field 122d may store a value indicating the purpose of the electrical signal. This attribute value may store the value of the store for the attribute name in field 112b of the correlation attribute information 112. This is obtained by extracting information corresponding to the column names listed in the correlation attribute information 112 and the store names and other items that contain that information from the business information 211 of the store information management server 200. For example, the second line of the store attribute information 122 shown in the table format in FIG. 6 has "A" in the store name column, "air conditioning" in the equipment name column, "{station building, midnight to midnight}" in the attribute value column, and "{power saving}" in the purpose column. This corresponds to the "air conditioning" in the second line of the correlation attribute information 112 shown in the table format in FIG. 5 , where "air conditioning" is stored in the "equipment name" column in the second line of the store attribute information 122. From the business information 211 corresponding to this "air conditioning," "A" is stored as the "store name," and "{station building, midnight to midnight}" is stored in the "attribute value" column of the store attribute information 122 corresponding to the attribute name "{location, business hours}." The "purpose" column of the store attribute information 122 may be provided as needed, and may contain a transcription and storage of the "purpose" targeted by the store attribute information 122. Information regarding the "purpose" may be included in the business information 211. This information regarding the use may be information indicating the use of the electrical signal, such as considering energy-saving measures based on the electrical signal of an air conditioner, or estimating cooking times based on the electrical signal of a microwave oven to help plan future cooking.

Figure 7 is a diagram showing an example configuration of candidate store information 124. The candidate store information 124 shown in Figure 7 may have fields 124a to 124e. Field 124a may store information for identifying the general store as the "general store name." Field 124b may store the name of the cluster to which the general store belongs as the "cluster number." Field 124c may store the name of the experimental store that belongs to the same cluster as the general store and is the shortest distance from the general store as the "nearby experimental store name." Field 124d may store the distance between the general store and the experimental store with the nearby experimental store name as the "distance from experimental store." Field 124e may store information indicating the order in which the general store should be added as an experimental store as the "candidate rank." For example, in the second row of the candidate store information 124 shown in table format in Figure 7, the general store name column stores "A," the cluster number column stores "1," which is the number of the cluster to which store A belongs, the nearby experimental store name column stores "X," which is the name of the experimental store located closest to store A, the distance from experimental store column stores "12," which is the distance between store A and the experimental store in the nearby experimental store name, and the candidate ranking column stores "1," which is the candidate ranking.

Figure 8 is a diagram showing an example configuration of similarity information 142. The similarity information 142 shown in Figure 8 may have fields 142a to 142c. In this example, field 142a may store information for identifying a candidate store as "candidate store name." Field 142b may store information for identifying an experimental store as "experimental store name." Field 142c may store the similarity of the electrical signals between the candidate store and the experimental store as "similarity." This similarity may be stored for all combinations of candidate stores and experimental stores. For example, the second row of candidate store information 124 shown in table format in Figure 8 stores "A" in the candidate store name column, "X" in the experimental store name column, and "80%" in the similarity column. This is the result of the similarity calculation unit 141 receiving the electrical signal selected by the electrical signal acquisition unit 131 and calculating the similarity of the electrical signals between the candidate store and the experimental store, and storing the result. This similarity calculation may be performed between the candidate store and all experimental stores, or between as many experimental stores as necessary.

Figure 9 is a diagram showing an example configuration of recommendation information 144. The recommendation information 144 shown in Figure 9 may have fields 144a to 144c. Field 144a may store information for identifying a candidate store as a "candidate store name." Field 144b may store an electrical device whose electrical signal will be measured at the candidate store as a "target device." Field 144c may store an order in which the candidate store is recommended as an experimental store for measuring the electrical signal of the target device as a "recommendation order." For example, the second row of recommendation information 144 shown in table format in Figure 9 stores "C" in the candidate store name column, "{air conditioner, rice cooker}" in the target device column, and "1" in the recommendation order column.

<Flowchart example>
FIG. 10 is a flowchart illustrating an example of the operation of the sensor placement assistance server 100.

First, the correlation analysis unit 111 of the sensor placement assistance server 100 may acquire the business information 211 of the experimental store and the electrical signals 212 of each electrical device in the experimental store from the store information management server 200 (step S101). Next, the correlation coefficient between the electrical signals 212 and the business information 211 for each electrical device may be calculated and stored in the correlation attribute information 112 (step S102). This correlation coefficient may be calculated using a known method such as Pearson's correlation coefficient or point biserial correlation coefficient to determine the correlation coefficient between the business information and the electrical signals of the electrical device, and a predetermined number of pieces of business information are recorded in the correlation attribute information 112 for each electrical device. When recording, the business information can be recorded in descending order of correlation coefficient. Furthermore, the number of pieces of business information to be recorded may be an appropriately set predetermined number.

The attribute information acquisition unit 121 can extract information corresponding to the column names listed in the correlation attribute information 112, as well as the store names containing that information, from the business information 211 on the store information management server 200, and store the extracted information as store attribute information 122 (step S103).

The clustering unit 123 performs clustering on general stores and experimental stores based on the attribute values described in the store attribute information 122, and calculates the cluster number to which the general store belongs, the name of the nearest experimental store that belongs to the same cluster as the general store, and the distance between the general store and the nearest experimental store (step S104). Here, clustering can be performed on all general stores and experimental stores. Next, the candidate ranking can be set in descending order of distance to the nearest experimental store, and the result can be stored in the candidate store information 124 together with the calculation result of step S104 (step S105).

As one example, the process can jump from step S105 to step S110, and the results stored in the candidate store information 124 can be sent to the display unit 300, which is the output unit, and displayed, i.e., output.

The electrical signal acquisition unit 131 can select a predetermined number of general stores as candidate stores from each cluster listed in the candidate store information 124 in descending order of candidate rank (step S106). This predetermined number can be set as appropriate. Next, electrical signals from the main circuits of the candidate stores and the experimental store can be acquired (step S107). At this time, if electrical signals from electrical equipment are also recorded at the candidate stores, the electrical signals from the electrical equipment may be acquired instead of or in addition to the electrical signals from the main circuits.

The similarity calculation unit 141 receives the electrical signals acquired by the electrical signal acquisition unit 131, calculates the similarity of the electrical signals between the candidate store and the experimental store, and stores the results in the similarity information 142 (step S108). This similarity calculation can be performed between the candidate store and all experimental stores. Alternatively, for example, experimental stores that are known to have a high degree of similarity can be excluded from the calculation. In this case, if the electrical signals of electrical equipment are selected instead of the electrical signals of the main circuit, the similarity of the electrical signals may be calculated only for electrical equipment with the same equipment name.

The recommendation unit 143 can set a recommendation order for the candidate stores in descending order of similarity listed in the similarity information 142 and store the order in the recommendation information 144 (step S109).

The display unit 300 displays the candidate stores and the electrical equipment to be measured in descending order of recommendation rank listed in the recommendation information 144, and can recommend installing a sensor on the electrical equipment to designate it as an experimental store (step S110). At this time, if the similarity calculation unit 141 calculates the similarity for the electrical signals of the main circuit, all electrical equipment in the candidate store will be displayed as the electrical equipment to be measured, but this can be displayed using the equipment configuration values listed in the business information 211. If the similarity is calculated for the electrical signals of an electrical equipment, this electrical equipment can be displayed as the measurement target. This display unit 300 may function as an output unit that outputs candidate experimental stores with high recommendation ranks (low recommendation ranks). In this embodiment, the display unit 300 is used as an output unit, but the candidate experimental stores can also be output as electronic data depending on the application.

Figures 11(a) and (b) show examples of recommendation screens 301. These recommendation screens 301 may be displayed and output on a display unit 300, for example, on a personal computer.

The recommendation screen 301 shown in FIG. 11(a) may be an example of a case where, when recommending a new experimental store as a candidate, electrical equipment on which an electrical sensor is to be installed is specified and recommended along with the store. In other words, this example may be an example of a case where electrical sensors are not installed on all electrical equipment in the new experimental store. The recommendation screen 301 shown in FIG. 11(a) may have fields 301a to 301g. Field 301a may display the order in which the candidate stores are recommended as new experimental stores as a "recommended order." Field 301c may display the name of the store recommended as a candidate store as a "candidate store name." Field 301b may display the cluster to which the store displayed in field 301c belongs as a "cluster." Field 301d may display the electrical equipment on which a sensor is to be installed in the store displayed in field 301c as a "target device." Field 301e may display the candidate stores specified in the "recommended order" along with the handling operation selected by the user of the sensor placement management system related to the present invention as an "operation." The recommendation screen 301 outputs and displays candidate experimental stores with low recommendation rankings, as shown in field 301a. In this example, candidate experimental stores with low recommendation rankings are output and displayed in ascending order of recommendation ranking. The number of candidate experimental stores with low recommendation rankings to display may be determined as needed. The user may select how to handle the recommended information displayed on the recommendation screen 301. For example, the user may select to adopt the recommended item by moving the pointer of the PC to an operation field of the information displayed in table format in FIG. 11( a) and clicking there on the PC. Before performing this adoption operation, the operation field may display "[Not Selected]," and when the user clicks to adopt, the corresponding operation field may display "[Adopt]." This click operation may cause the operation field to cycle through the following options: [Not Selected], [Adopt], and [Exclude]. For example, "Accept" can be configured to mean that the recommendation of the store or the like is adopted, "Exclude" means that the recommendation of the store or the like is not adopted, and "Not Selected" means that the recommendation of the store or the like is not adopted. Field 301f may be an icon that is clicked to confirm and confirm the selection of the store or the like recommendation after the selection is completed and the selection information is confirmed. The "Recalculate" icon in field 301g may be an icon used to recalculate using conditions different from the information currently displayed on the recommendation screen 301 and display new recommendation information. In other words, when recalculating using conditions different from the current conditions, the conditions may be entered, for example, from a computer keyboard, and the "Recalculate" icon may be clicked, causing the calculation to be recalculated according to the newly entered conditions.

The recommendation screen 301 shown in FIG. 11(b) differs from the case in FIG. 11(a) in that it may be an example in which electrical sensors are installed on all electrical equipment in a new experimental store when the new experimental store is recommended as a candidate. The recommendation screen 301 shown in FIG. 11(b) may have fields 301a-301c and 301e-301g. The explanation of each field is the same as that in FIG. 11(a), so the explanation will be omitted.

By adopting a configuration like the recommendation screen 301 shown in Figures 11(a) and (b), coupled with the processing required to display this recommendation screen 301, it is possible to collect the learning data required to create a probabilistic generation model that estimates the operating status of electrical equipment from the electrical signal of the main circuit using a small number of sensors. In particular, by adopting a configuration like the recommendation screen 301 shown in Figures 11(a) and (b), users of the sensor placement management system according to the present invention can easily grasp information about recommended stores, etc., and the system is easy to operate.

In the embodiment described above, the candidate store calculation unit 120 may further cluster all store attribute information and calculate a predetermined number of experimental store candidates from the clusters to which the candidate stores belong, in descending order of distance from the nearest experimental store to the candidate stores. Here, "distance" refers to the distance within the cluster resulting from clustering, and this applies throughout this specification and drawings unless otherwise specified.

In this case, the candidate store calculation unit 120 processes the data in the same way as described above, with the clustering unit 123 performing clustering based on the attributes described in the store attribute information 122 and the distances between stores, and recording information about the cluster to which each store belongs, nearby experimental stores, and the distances from those experimental stores in the candidate store information 124. The clustering calculation method and distance definition are also the same. As in the previous case, the clustering unit 123 performs clustering based on the attributes described in the store attribute information 122 and the distances between stores, and divides the stores into several clusters according to, for example, the correlation between store attributes and the distances between stores.

Furthermore, after clustering all store attribute information 122 in the clustering unit 123, in each cluster to which the candidate store belongs, i.e., in clusters in which general stores exist, a predetermined number of general stores are selected as candidates for a new experimental store in order of greatest distance from the nearest experimental store. Here, in the case of a cluster in which only one general store exists, the predetermined number may be 1.

Figure 7 shows an example of information stored as candidate store information 124 through this process. Figure 7 shows that a general store with the general store name "A" belongs to a cluster with cluster number "1", its nearby experimental store is "X", the distance between store A and store X is "12", and its candidate ranking is "1".

In this example, for each general store in a cluster, a predetermined number of experimental stores belonging to that cluster may be selected as candidates to be experimental stores, i.e., candidate stores, in descending order of distance from the nearest experimental store. In other words, general stores that do not have any existing experimental stores nearby become candidates for new experimental stores, and a probabilistic generation model derived using these new experimental stores is expected to reproduce the electricity consumption within the store with high accuracy. Furthermore, by using this probabilistic generation model, it is expected that the electricity consumption within other general stores near the new experimental store will also be reproduced with high accuracy.

In Figure 7, candidate rankings are assigned based on the distance between the general store and its nearest experimental store. Here, the ranking is not dependent on the cluster, and the candidate ranking 124e may be changed to 1, 2, 3, etc., in order of the general store corresponding to the general store name 124a with the largest distance 124d from the experimental store. Note that the "distance from experimental store" in field 124d refers to the distance between the general store and its nearest experimental store. That is, in the example shown in Figure 7, the distances 124d from the experimental store to the nearest are "12," "5," and "3," so correspondingly, the candidate ranking 124e may be set to "1" for general store A, "2" for general store C, and "3" for general store B.

When ranking is performed in this way regardless of cluster, candidate rankings 124e are assigned to all general stores in descending order of distance 124d from the experimental store. Therefore, by selecting a general store with a high candidate ranking 124e as the new experimental store, a highly accurate probability generation model that can be applied to many general stores can be efficiently created.

Alternatively, candidate rankings can also be assigned for each cluster. This will be explained using an example in which the data from the general store name 124a to the distance 124d from the experimental store in Figure 7 is the same. First, the cluster to which the general store corresponding to the general store name 124a with the longest distance 124d from the experimental store belongs may be selected. In this case, cluster 1 may be selected, which includes general store A with "A" in the general store name 124a and the longest distance 124d from the experimental store (12). Cluster 1 may then be assigned a candidate ranking 124e. In this case, the general stores belonging to the cluster with cluster number 124b of "1" are general stores A and B, based on the general store name 124a. The candidate rankings 124e for general stores A and B can be determined in descending order of distance 124d from the experimental store. In this example, the distances 124d from the experimental store corresponding to general stores A and B are "12" and "3," respectively. Therefore, the candidate rankings 124e can be set to "1" and "2," respectively. These are the general stores belonging to cluster 1. Next, the cluster with the largest distance 124d from the experimental store can be selected. That is, for the general stores belonging to cluster 1, the cluster with the second largest distance 124d from the experimental store after the largest distance 124d from the experimental store, "12," is selected. That is, the cluster with cluster number 124b of "2" can be represented as cluster 2, and cluster 2 to which general store C belongs can be selected. Then, the general stores in the general store name 124a belonging to cluster 2 can be re-assigned candidate rankings 124e in descending order of distance 124d from the experimental store. As a result of this processing, in this example, the candidate rankings 124e can be set as follows: general store A is set to "1," general store B is set to "2," and general store C is set to "1." Although the same candidate ranking may appear multiple times, they can be distinguished because the candidate rankings correspond to the cluster numbers.

A single cluster may contain many stores with similar attributes. For example, one cluster may have many stores located in station buildings, while another cluster may have many stores located along national highways. In this case, by ranking candidates within each cluster, it is possible to identify new experimental stores as candidates based on the unique conditions of being located in a station building.

In the above example, after clustering all store attribute information 122, a predetermined number of general stores in each cluster were selected as candidates for new experimental stores in descending order of distance to the nearest experimental store. However, when deriving the distance between a general store in each cluster and its nearest experimental store, the present invention also covers cases where the nearest experimental store does not belong to the same cluster as the general store. This has the advantage of being able to find candidates for new experimental stores even when a cluster contains general stores but no experimental stores. Clustering can also be performed on only some of the store attribute information 122 as needed, rather than on all store attribute information 122. For example, if there is a general store that is not a candidate for a new experimental store for some reason, that general store can be excluded from clustering. This can shorten processing time for clustering, etc.

Next, a further modification will be described. The above embodiment may further include an electrical signal acquisition unit 130 that collects electrical signals from electrical equipment used in candidate stores and experimental stores from electrical sensors installed in those stores, and an experimental store recommendation unit 140 that calculates the similarity of the collected electrical signals between the candidate store and the experimental store, and recommends electrical equipment in the candidate store with a low similarity as a sensor installation location where an electrical sensor should be added. Note that a low similarity may mean that the similarity is relatively low.

As described in the above embodiment, the clustering unit 123 performs clustering to obtain candidate store information 124. Next, the electrical signal acquisition unit 131, which is the main function of the electrical signal acquisition unit 130, can select, for each cluster listed in the candidate store information 124, a predetermined number of general stores with high candidate rankings 124e within that cluster as candidate stores, i.e., stores that are candidates for use as new experimental stores (see Figure 10, S106). Next, for these selected candidate stores and the experimental stores associated with the nearby experimental store names 124c corresponding to those candidate stores, electrical signals from electrical equipment used in those stores may be acquired and collected from electrical sensors installed in those stores (see Figure 10, S107). In Figure 10, step S107 is simply described as "acquire electrical signals from candidate stores and experimental stores."

Next, the similarity calculation unit 141 receives the electrical signals acquired and collected by the electrical signal acquisition unit 131, calculates the similarity of the electrical signals between the candidate store and all experimental stores, and stores the results in the similarity information 142 (step S108). Here, when calculating the similarity, it is possible to select only a portion of the experimental stores as appropriate, such as the experimental store corresponding to the candidate store, rather than all of the experimental stores. Next, the experimental store recommendation unit 140 may set a recommendation order for the candidate stores listed in the candidate store name 142a in the similarity information 142 in ascending order of similarity 142c, and store the order in the recommendation information 144 (step S109).

The display unit 300 may then display the candidate stores and the electrical devices whose electrical signals will be measured in descending order of the recommendation ranking of the candidate stores stored in the recommendation information 144, and recommend that an electrical sensor be installed in the electrical device to make it an experimental store (step S110). The display of the recommendation ranking of the candidate stores can be done in a manner other than descending order of recommendation ranking, depending on the application.

By configuring it in this way, candidate stores that are not only highly ranked through clustering but also have electrical signals with low similarity to those of the experimental store are recommended as new experimental stores. Therefore, if the candidate store uses unique equipment not found in other stores, it is expected that a probabilistic generation model that is more suited to the candidate can be created, making it useful as a candidate for a new experimental store. In other words, the learning data required to create a probabilistic generation model can be collected with a small number of sensors.

Next, a further modification will be described. In the above-described embodiment, the system may include an electrical signal acquisition unit 130 that collects electrical signals from electrical devices used in the candidate stores and experimental stores from electrical sensors installed in those stores, and an experimental store recommendation unit 140 that calculates the similarity between the collected electrical signals for the candidate stores and the experimental stores and recommends electrical devices in the candidate stores with low similarity as sensor installation locations where electrical sensors should be added. In this modification, the experimental store recommendation unit 140 may also be a sensor placement management system configured to calculate the similarity between the main power electrical signals for each candidate store and the experimental store. That is, when acquiring electrical signals from electrical devices used in the candidate stores selected in step S106 and the experimental stores associated with the nearby experimental store names 124c corresponding to those candidate stores from electrical sensors installed in those stores, the electrical sensors may be installed in the main circuits of those stores, thereby acquiring and collecting electrical signals from all electrical devices connected to the main circuits (see FIG. 10, S107). Thereafter, the similarity calculation unit 141 may calculate the similarity, and the display unit 300 may recommend a new experimental store in the same manner as described above (step S110). In the description of the above embodiment, the experimental store recommendation unit 140 recommended electrical devices with low similarity as sensor installation locations where electrical sensors should be added. However, in this modified example, these electrical devices with low similarity may refer to all electrical devices connected to the main circuit. The experimental store recommendation unit 140 may recommend all electrical devices with low similarity as sensor installation locations where electrical sensors should be added, i.e., a group of electrical devices, in other words, a main circuit with low similarity. Note that "low similarity" may mean that the similarity is relatively low.

According to this variant, for a group of one or more electrical devices connected to the main circuit at each of the candidate store and the experimental store, the sum of the electrical signals from those electrical devices is obtained, the similarity of these sums is calculated, and the candidate store with a relatively low similarity is recommended as the new experimental store. This allows for the generation of a more accurate probabilistic generation model, enabling better predictions of store electricity consumption. In other words, the learning data required to create the probabilistic generation model can be collected with a smaller number of sensors. Furthermore, the probabilistic generation model can be obtained by acquiring electrical signals at a single point on the main circuit, which also contributes to reducing data acquisition costs.

Next, a further variation will be described. In the above-described embodiment, the system may have an electrical signal acquisition unit 130 that collects electrical signals from electrical devices used in candidate stores and experimental stores from electrical sensors installed in those stores, and an experimental store recommendation unit 140 that calculates the similarity between the collected electrical signals for the candidate store and the experimental store and recommends electrical devices in the candidate store with low similarity as sensor installation locations where electrical sensors should be added. In this variation, the system may further calculate a second similarity for the sum of electrical signals from some of the electrical devices corresponding to the candidate store and the experimental store, and the experimental store recommendation unit 140 may be a sensor placement management system that recommends the installation of electrical sensors for electrical devices with low similarity installed in the candidate store.

Here, the electrical devices with low similarity that are recommended may be electrical devices with a relatively low calculated second similarity. The number of electrical devices to be recommended may be determined arbitrarily as needed. Furthermore, the electrical devices recommended here may be those for which electrical sensors are to be installed only on electrical devices installed in candidate stores with a relatively low similarity.

In this variant, the similarity between the electrical signals collected for the candidate store and the experimental store, as described in the above embodiment, is calculated, and electrical equipment in the candidate store with low similarity is recommended as a sensor installation location where an electrical sensor should be added. In addition, it may be possible to recommend further sensor installation locations where an electrical sensor should be added.

In a store, external power is drawn into the store via a main circuit via a power line, branching off into several circuits, which then branch off further to distribute power to various electrical devices. An example of such a branch is a circuit breaker. By measuring electrical signals at such a branch location, the sum of the signals from one or more electrical devices connected downstream of the branch can be measured. In other words, from the perspective of the store, the sum of the electrical signals from some of the electrical devices corresponding to the store can be measured and obtained. In this variation, for example, a similarity can be calculated for the sum of the electrical signals from some of the electrical devices corresponding to the candidate store and the experimental store, obtained in this manner, and this similarity can be used as the second similarity. The experimental store recommendation unit 140 may recommend the installation of electrical sensors for electrical devices installed in the candidate store with low similarity. When recommending the installation of electrical sensors using the sum of signals from electrical devices at branch locations such as circuit breakers, the electrical device on which the recommended electrical sensor is to be installed may be, for example, the branch described above, or a breaker installed at the branch. Note that low similarity may mean that the similarity is relatively low.

In addition to the effects described in the above-mentioned embodiment, this variant is expected to recommend the installation of electrical sensors on electrical equipment that is effective when some of the electrical equipment in a candidate store is distinctive, that is, when that equipment is distinctive compared to other equipment, such as consuming a lot of power or being used during specific time periods, and to obtain a highly accurate probabilistic generation model. In other words, the learning data required to create a probabilistic generation model can be collected using a small number of sensors.

The above-mentioned similarity will be explained in more detail. In the present invention, similarity refers to the degree of similarity between the electrical signal of a candidate store and the electrical signal of an experimental store. Even if the electrical signal of a candidate store is highly similar to the electrical signal of an existing experimental store, the electrical signal of the existing experimental store can be used as a substitute, and therefore the candidate store does not necessarily need to be recommended in the recommendation information 144. When the recommendation unit 143 recommends candidate stores, etc. using similarity, variations such as (1) filtering out stores with relatively high similarity and excluding them from the candidate stores, (2) selecting the candidate store with the lowest similarity among experimental stores with relatively low similarity, or (3) recommending the candidate store with the lowest similarity among the maximum similarity values corresponding to each experimental store can be adopted. Regarding "(1)" and "(2)," the efficiency of selection can be improved. Regarding "(3)," a more effective probability generation model can be obtained when general stores with relatively similar electrical signals are selected as candidate stores.

An even more refined embodiment will now be described. In this embodiment, the candidate store calculation unit 120 in the above-described embodiment acquires the use of electrical signals described in the business information and uses this information for clustering, making it a sensor placement management system.

When the attribute information acquisition unit 121 in the candidate store calculation unit 120 creates store attribute information 122, it acquires information about the use of electrical signals at that store and adds it to the store attribute information 122. This information about use is output as store attribute information 122, as shown in FIG. 6, or added to the items in candidate store information 124, and can be displayed on the display unit 300, for example, for use.

In Figure 6, for example, this information regarding use is listed as "power saving" for Store A, whose name is A. This means, for example, that the "air conditioning" listed in the equipment name used in Store A is controlled and used with the aim of power saving. Also, in Figure 6, for example, it means that the "microwave oven" listed in the equipment name used in Store B is controlled and used with the aim of cooking plans. When considering a new experimental store using this sensor placement management system, knowing this information regarding use can serve as reference information when developing a power saving plan for air conditioning at another store, using the control and use of air conditioning at Store A as a reference. Furthermore, using the cooking plan for microwave ovens at Store B, whose name is B, as a reference, can serve as reference information when developing a cooking plan, such as considering cooking times and staff allocation, at another store.

In the above explanation of the embodiments, commercial stores such as convenience stores have been used as an example of an application of the sensor placement management system of the present invention. Therefore, the names of the matters specifying the invention are those corresponding to commercial stores. The sensor placement management system of the present invention can be applied not only to commercial stores, but also to various facilities that are units of electricity consumption, such as factories, homes (single-family homes, apartment buildings), and office buildings. Here, the names of the matters specifying the invention used in explaining the embodiments will be replaced with names corresponding to these various facilities. The replacements are as follows: That is, the store attribute calculation unit described as store attribute calculation unit 110 can be interpreted as a power consumption target attribute calculation unit, the candidate store calculation unit described as candidate store calculation unit 120 can be interpreted as a candidate power consumption target calculation unit, store attribute information described as store attribute information 122 can be interpreted as attribute information, candidate store information described as candidate store information 124 can be interpreted as candidate power consumption target information, the experimental store recommendation unit described as experimental store recommendation unit 140 can be interpreted as an experimental power consumption target recommendation unit, the store information management server described as store information management server 200 can be interpreted as a power consumption target information management server, and the display unit described as display unit 300 can be interpreted as an output unit. Furthermore, the term "business information" used in the above description of the embodiments is not particularly interpreted differently, but it is understood to have the meaning of, for example, in-home usage information when the present invention is applied to a home. The same applies to other applications.

The above describes one embodiment of the present invention using examples. In the above description, the electrical signal acquisition unit 130 acquires electrical signals from candidate stores and the like. This electrical signal acquisition can also acquire data that has already been measured and stored in the electrical signal 212 in the store information management server. Alternatively, electrical signal measurement sensors can be installed in candidate stores and the like, and electrical signal data can be acquired through online data communication. Furthermore, the diagrams used in the explanation of the examples do not completely represent the paths of data and signals, and the necessary movements will occur even if the paths of data and signals are not explicitly shown in the diagrams.

The present invention is not limited to the above-described embodiments, and includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and are not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace part of the configuration of each embodiment with other configurations, either alone or in combination.

Furthermore, the above-mentioned configurations, functions, processing units, processing means, etc. may be realized in part or in whole in hardware, for example by designing them as integrated circuits. Furthermore, the above-mentioned configurations and functions, etc. may be realized in software by a processor interpreting and executing programs that realize each function. Information such as programs, tables, and files that realize each function can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

In addition, the control lines and information lines shown in the diagram are those considered necessary for explanation, and do not necessarily show all control lines and information lines in the product. In reality, it can be assumed that all components are interconnected as needed.

100: Sensor placement support server 110: Store attribute calculation unit 111: Correlation analysis unit 112: Correlation attribute information 120: Candidate store calculation unit 121: Attribute information acquisition unit 122: Store attribute information 123: Clustering unit 124: Candidate store information 130: Electrical signal acquisition unit 131: Electrical signal acquisition unit 140: Experimental store recommendation unit 141: Similarity calculation unit 142: Similarity information 143: Recommendation unit 144: Recommendation information 200: Store information management server 211: Business information 212: Electrical signal 300: Display unit 301: Recommendation screen

Claims (10)

A sensor placement management system,
A computing device that executes computational processing and a storage device that can be accessed by the computing device,
a candidate power consumption target calculation unit configured to use, as attribute information of the experimental power consumption target , business information including at least one of the address, location, and business hours of the power consumption target, the business information correlating with an electrical signal measured from an experimental power consumption target, which is the power consumption target that measures an electrical signal from an electrical sensor installed in an internal electrical device , to cluster the experimental power consumption target and general power consumption targets, and to calculate a ranking of the general power consumption target as a candidate for a new experimental power consumption target within the cluster to which the general power consumption target belongs ;
The computing device has an output unit that outputs the candidate experimental power consumption target with the lowest ranking. The sensor placement management system according to claim 1 ,
The sensor placement management system is characterized in that the candidate power consumption target calculation unit calculates the ranking based on a distance between the general power consumption target and the experimental power consumption target. The sensor placement management system according to claim 2,
The candidate power consumption target calculation unit calculates the ranking in descending order of distance to the nearest experimental power consumption target within a cluster to which the candidate experimental power consumption target belongs, and calculates a predetermined number of candidate experimental power consumption targets from the cluster to which the candidate experimental power consumption target belongs based on the ranking. The sensor placement management system according to claim 2,
The sensor placement management system is characterized in that, when there is no experimental power consumption object in the cluster to which the candidate experimental power consumption object belongs, the candidate power consumption object calculation unit calculates the ranking in order of the greatest distance to the nearest experimental power consumption object among the experimental power consumption objects included in a cluster other than the cluster to which the candidate experimental power consumption object belongs. 5. The sensor placement management system according to claim 1,
A sensor placement management system characterized in that the calculation device further has a power consumption object attribute calculation unit that calculates attributes of the experimental power consumption object that are highly correlated with the electrical signal by correlation analysis between the business information of the experimental power consumption object and the electrical signal. 5. The sensor placement management system according to claim 1,
the computing device includes an electrical signal acquisition unit that collects electrical signals of electrical devices from electrical sensors installed on the candidate experimental power consumption objects and the experimental power consumption objects;
The sensor placement management system is characterized in that the calculation device has an experimental power consumption object recommendation unit that calculates the similarity of the electrical signals between the candidate experimental power consumption object and the experimental power consumption object, and recommends the candidate experimental power consumption object electrical equipment with the low similarity as a sensor installation location where an electrical sensor should be added. The sensor placement management system according to claim 6,
The sensor placement management system is characterized in that the similarity is a similarity of the main power electrical signal for each power consumption object. The sensor placement management system according to claim 6,
The sensor arrangement management system is characterized in that the similarity is a similarity for a sum of electrical signals of some of the corresponding electrical devices. 5. The sensor placement management system according to claim 1,
the business information includes a use of the electrical signal;
The candidate power consumption target calculation unit acquires the use of the electrical signal and uses the use of the electrical signal for clustering. A sensor placement management method executed by a sensor placement management system,
the sensor placement management system includes a computing device that executes computational processing and a storage device that is accessible by the computing device;
The sensor placement management method includes:
a candidate power consumption target calculation step in which the computing device uses, as attribute information of the experimental power consumption target, business information including at least one of the address, location, and business hours of the power consumption target, the business information correlating with an electrical signal measured from an experimental power consumption target, which is the power consumption target that measures an electrical signal from an electrical sensor installed in an internal electrical device, to cluster the experimental power consumption target and general power consumption targets, and calculate a ranking of the general power consumption target as a candidate for a new experimental power consumption target within the cluster to which the general power consumption target belongs;
an output step in which the computing device outputs the candidate experimental power consumption target with the lowest ranking.