Household composition change detection device and household composition change detection method

Abstract

To provide a household constitution change detection device and a household constitution change detection method capable of detecting a change in household constitution more easily.SOLUTION: A household constitution change detection device includes a control unit. The control unit acquires a usage electric power amount of each household from a power meter having a communication function via a communication network. The control unit detects, on the basis of a change tendency in usage electric power amounts common to changes in household constitution to be understood, a change in the household constitution to be understood. Changes in the household constitution to be understood are, for instance, childbirth. The control unit detects, on the basis of a year-on-year rate of daytime electric power usage for at least six consecutive months, a childbirth household, which is a household whose household constitution is changed due to the childbirth.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present invention relates to a household composition change detection device and a household composition change detection method.

Background Art

[0002] In recent years, effective utilization of the power consumption of each household in a region has been studied. For example, the household size assumption device of Patent Document 1 estimates the household size of each household in a region based on the power consumption by time zone measured by a power meter during the cooling operation in summer or the heating operation in winter. The estimated household size of each household is used, for example, to assume the evacuation situation in the region when a disaster occurs in the region.

[0003] Further, the user attribute estimation device of Patent Document 2 estimates an attribute that is a living behavior characteristic of household members based on the power consumption of each household that is a user. The user attribute estimation device estimates the attribute of household members in a characteristic time zone, which is a time zone in which the power consumption has a characteristic according to the weather, based on the power consumption in the characteristic time zone and information related to the weather in the region to which the household belongs. This estimated attribute is used, for example, for support related to life.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] For example, in marketing activities or sales activities in a company, it may be required to grasp changes in household composition. It is desired to more easily detect changes in household composition.

Means for Solving the Problems

[0006] The household composition change detection device that can solve the above problems has a control device that acquires the power consumption of each household from a power meter having a communication function via a communication network. The control device detects the change in the household composition based on the tendency of the change in power consumption common to the change in the household composition to be captured.

[0007] Based on the correlation between the change in the household composition to be captured and the change in power consumption, it is possible to detect the change in the household composition to be captured based on the characteristic change in power consumption. According to the above configuration, it is only necessary to focus on the change in power consumption, so it is possible to more easily detect the household whose household composition has changed.

[0008] In the above household composition change detection device, the change in the household composition may be childbirth. In this case, the control device may detect a childbirth household, which is a household whose household composition has changed due to childbirth, based on the change in the daytime power consumption over at least a continuous six months.

[0009] In a childbirth household, there is a tendency for the power consumption in the daytime time zone to increase for about six months from the period before and after childbirth. Therefore, it is possible to detect a childbirth household based on the continuous increasing tendency of the power consumption in the daytime time zone.

[0010] In the above household composition change detection device, the control device calculates, as a childbirth score indicating the probability of being a childbirth household, the maximum value over the year of the sum of the squares of the upward deviations from the median of the same month of the previous year of the daytime power consumption over at least a continuous six months, and may detect a childbirth household based on the calculated childbirth score.

[0011] According to this configuration, by focusing on the increment of the electricity consumption in the current year compared to the same month of the previous year, seasonality can be removed, that is, the influence caused by seasonal fluctuations in electricity consumption can be suppressed. Also, since the times such as maternity leave, childcare leave, or returning to one's hometown for childbirth vary from person to person, the time when changes occur in electricity consumption also varies from person to person. In this regard, by using the maximum value over the years of the sum of squares as the birth score, it is possible to focus on the period when the change in electricity consumption is most prominent for each household. Therefore, the group extracted based on the birth score has a significantly improved detection accuracy of the birth household compared to the randomly extracted group.

[0012] In the above household composition change detection device, the control device may reduce the value of the birth score of a household having an attribute with a low probability of giving birth based on the attribute information of each household.

[0013] According to this configuration, a household having an attribute with a low probability of giving birth becomes difficult to be detected. Therefore, the birth household can be detected more effectively. In the above household composition change detection device, the change in the household composition may be an increase or decrease in the number of household members. In this case, the control device may detect a household in which the number of household members has increased or decreased based on the change in the electricity consumption during the morning of a holiday.

[0014] In a household in which the number of household members has increased or decreased, regardless of the age of the person leaving the household, the electricity consumption during the morning of a holiday tends to change. Therefore, it is possible to detect a household in which the number of household members has increased or decreased based on the change tendency of the electricity consumption during the morning of a holiday.

[0015] In the above household composition change detection device, the control device may detect a household in which the number of household members has increased or decreased based on the comparison of the electricity consumption during the morning of a holiday with the same month of the previous year. As described above, by paying attention to the year-on-year change in power consumption, seasonality can be removed, that is, the influence of seasonal fluctuations in power consumption can be suppressed. Therefore, it is possible to effectively detect households whose household size has increased or decreased.

[0016] In the above household composition change detection device, the change in the household composition may be a decrease in the household size. In this case, the control device may detect a household whose household size has decreased due to the departure of household members of a specific age group based on the change in power consumption in a time period in which a characteristic change in power consumption is observed according to the age of the household members leaving the household.

[0017] There is a time period in which a characteristic change in power consumption is observed according to the age of the household members leaving the household. Therefore, it is possible to detect a household whose household size has decreased due to the departure of household members of a specific age group based on the change trend of power consumption in a time period specific to the age of the household members leaving the household.

[0018] In the above household composition change detection device, the control device may detect a household whose household size has decreased due to the departure of the younger generation from their 20s to 30s based on the change in power consumption on weekdays at night.

[0019] In a household whose household size has decreased due to the departure of the younger generation from their 20s to 30s, the power consumption on weekdays at night tends to decrease. Therefore, it is possible to detect a household whose household size has decreased due to the departure of the younger generation from their 20s to 30s based on the change trend of power consumption on weekdays at night.

[0020] In the above household composition change detection device, the control device may detect a household whose household size has decreased due to the departure of the elderly generation aged 60 or above based on the change in power consumption on weekdays in the morning.

[0021] In households where the household size has decreased due to the departure of the elderly generation aged 60 and above, the electricity consumption tends to decrease on weekday evenings. Therefore, based on the change trend of electricity consumption in the morning on weekdays, it is possible to detect households where the household size has decreased due to the departure of the elderly generation aged 60 and above.

[0022] In the above household composition change detection device, the control device may detect a household where the household size has decreased due to the departure of household members of a specific age group based on the year-on-year comparison of electricity consumption in the time period when characteristic changes in electricity consumption are observed according to the age of the household members leaving the household.

[0023] As described above, by focusing on the year-on-year comparison of electricity consumption, seasonality can be removed, that is, the influence of seasonal fluctuations in electricity consumption can be suppressed. Therefore, it is possible to effectively detect households where the household size has decreased due to the departure of household members of a specific age group.

[0024] In the above household composition change detection device, the control device may detect a household that has started working from home based on the change trend of electricity consumption during the day on weekdays. In households that have started working from home, the electricity consumption tends to increase during the day on weekdays. Therefore, based on the change trend of electricity consumption during the day on weekdays, it is possible to detect households that have started working from home.

[0025] In the above household composition change detection device, the control device may detect a household that has started working from home based on the year-on-year comparison of electricity consumption during the day on weekdays. As described above, by focusing on the year-on-year comparison of electricity consumption, seasonality can be removed, that is, the influence of seasonal fluctuations in electricity consumption can be suppressed. Therefore, it is possible to effectively detect households that have started working from home.

[0026] In the above household composition change detection device, the control device may detect a household in which a household member who has entered elementary school exists based on the change tendency of the power consumption during the period from before noon to after noon in August.

[0027] In a household in which a household member who has entered elementary school exists, the power consumption during the period from before noon to after noon with reference to noon in August tends to increase. For this reason, it is possible to detect a household in which a household member who has entered elementary school exists based on the change tendency of the power consumption during the period from before noon to after noon with reference to noon in August.

[0028] In the above household composition change detection device, the control device may detect a household in which a household member who has entered elementary school exists based on the year-on-year comparison of the power consumption during the period from before noon to after noon in August.

[0029] As in this configuration, by paying attention to the year-on-year comparison of the power consumption, seasonality can be removed, that is, the influence due to seasonal fluctuations in power consumption can be suppressed. For this reason, it is possible to effectively detect a household in which a household member who has entered elementary school exists.

[0030] A household composition change detection method capable of solving the above problems acquires the power consumption of each household from a power meter having a communication function via a communication network, and detects the change in the household composition based on the change tendency of the power consumption common to the change in the household composition to be grasped.

[0031] According to this method, if a correlation between the change in the household composition to be grasped and the change in the power consumption can be found, it is possible to detect the change in the household composition to be grasped based on the characteristic change in the power consumption. Since it is only necessary to focus on the change in the power consumption, it is possible to more easily detect a household in which the household composition has changed.

Advantages of the Invention

[0032] According to the household composition change detection device and the household composition change detection method of the present invention, it is possible to more easily detect changes in the household composition.

Brief Description of the Drawings

[0033]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Mode for Carrying Out the Invention

[0034] <First Embodiment> Hereinafter, a first embodiment in which a household composition change detection device and a household composition change detection method are embodied will be described. The household composition detection device detects a change in the household composition based on the power consumption of the household.

[0035] As shown in FIG. 1, the detection device 10 is connected to a communication network 20 such as the Internet. The detection device 10 is provided in a general power transmission and distribution business operator such as an electric power company. The detection device 10 can communicate with the smart meter 30 via the communication network 20. The smart meter 30 is an electric power meter having both an electric power measurement function and a communication function, and is provided in each household in the area. The smart meter 30 measures the power consumption every predetermined period (for example, 30 minutes), and transmits power data, which is data indicating the measured power consumption, to the detection device 10 via the communication network 20. The power data also includes the location information of the smart meter 30. The detection device 10 detects a change in the household composition based on the power data of each household acquired via the communication network 20.

[0036] As shown in FIG. 2, the smart meter 30 has a measurement device 30A, a storage device 30B, a communication device 30C, and a control device 30D. These measurement device 30A, storage device 30B, communication device 30C, and control device 30D are interconnected via a bus 30E which is a signal line.

[0037] The measurement device 30A measures the power consumption of a household which is a consumer every predetermined period (for example, 30 minutes). The measurement device 30A detects the voltage of the power system and the current supplied from the power system to the load. The measurement device 30A measures the power consumption of the household based on the detected voltage and current.

[0038] The storage device 30B stores the power consumption measured by the measurement device 30A according to an instruction from the control device 30D. Also, the storage device 30B stores identification information unique to the smart meter 30 and the location information of the smart meter 30. This location information is information indicating the location where the smart meter 30 is installed, for example, information indicating the location such as the house where the smart meter 30 is installed.

[0039] The communication device 30C is an interface between the control device 30D and the communication network 20. The communication device 30C transmits and receives information via the communication network 20. The communication device 30C exchanges information with an external device connected to the communication network 20 according to an instruction from the control device 30D. The external device includes the detection device 10.

[0040] The control device 30D comprehensively controls the entire smart meter 30. The control device 30D generates power data. The power data includes the power consumption measured by the measurement device 30A and the location information of the smart meter 30. The control device 30D transmits the power data to the detection device 10 via the communication device 30C.

[0041] As shown in FIG. 3, the detection device 10 includes a communication device 10A, a storage device 10B, and a control device 10C. These communication device 10A, storage device 10B, and control device 10C are interconnected via a bus 10D which is a signal line.

[0042] The communication device 10A is an interface between the control device 10C and the communication network 20. The communication device 10A transmits and receives information via the communication network 20. The communication device 10A exchanges information with an external device connected to the communication network 20 according to an instruction from the control device 10C. The external device includes a smart meter 30.

[0043] The storage device 10B has, for example, a main storage unit and an auxiliary storage unit. The main storage unit stores a computer program executed by the control device 10C, data processed by the control device 10C, and the like. The main storage unit has a RAM (Random Access Memory) and a (Read Only Memory). The auxiliary storage unit is a non-volatile storage device that can read and write various programs including an OS (Operating System) and various data. The auxiliary storage unit is a flash memory, a hard disk drive (Hard Disk Drive), or an SSD (Solid State Drive), etc. The OS includes a communication interface program for transferring data between the external device connected via the communication device 10A. Information received via the communication device 10A is stored in the auxiliary storage device. This information includes power data sent from the smart meter 30 via the communication network 20.

[0044] The control device 10C executes various processes according to a program stored in the storage device 10B. The control device 10C has a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). The processor comprehensively controls the entire detection device 10 through the execution of a computer program developed in the work area of the storage device 10B. The control device 10C receives power data sent from the smart meter 30 via the communication network 20 through the communication device 10A, and stores the received power data in the storage device 10B in a state associated with the identification information unique to the smart meter 30.

[0045] The control device 10C detects changes in household composition based on the trend of changes in power consumption during a specific period (for example, the number of months or time zones). If the correlation between the changes in household composition to be captured and the changes in power consumption can be discovered, it is possible to detect the changes in household composition to be captured based on the discovered characteristic changes in power consumption. The changes in household composition to be captured are, for example, life events such as childbirth, children leaving home, separation, or death. Here, the case of detecting a childbirth household will be described as an example. A childbirth household refers to a household whose household composition has changed due to childbirth.

[0046] As shown in FIG. 4, in a childbirth household, the power consumption during the daytime time zone tends to increase for about half a year from the period before and after childbirth. Based on the continuous increasing trend of the electricity usage fee during this daytime time zone, households with a higher probability of being a childbirth household can be extracted. Incidentally, FIG. 4 shows the change in power consumption before and after childbirth in a household where the household size has increased from 3 to 4 due to childbirth. One horizontal square in FIG. 4 represents one day, and one vertical square represents a 30 - minute increment. The darker the color of the grid, the higher the power consumption.

[0047] From the change in power consumption shown in FIG. 4, the living situation of a household expecting childbirth can be understood. The time on the first day of the month before and the month before the month of household composition change (here, the month expecting childbirth) can be divided into, for example, four time zones TZ0, TZ1, TZ2, and TZ3.

[0048] · Time zone TZ0: Time zone for morning preparation · Time zone TZ1: Time zone for two people, the wife and the child · Time zone TZ2: Time zone on days when the husband returns home earlier than usual · Time zone TZ3: Time zone after the husband returns home and all family members are together Next, the time on the day of the month of household composition change and the first ten days of the following month can be divided into, for example, three time zones TZ4, TZ5, and TZ6.

[0049] · Time zone TZ4: Time zone for morning preparation · Time zone TZ5: Time zone during the day · Time zone TZ6: Time zone after the husband returns home In the month of household composition change and the first ten days of the following month, compared with the month before and the month before the month of household composition change, the power consumption in the earlier time zone of the morning time zone TZ4, for example, from 7:00 am to 8:00 am, is increasing. From this, it is assumed that there has been a change in morning preparation. Also, in the month of household composition change and the first ten days of the following month, compared with the month before and the month before the month of household composition change, the power consumption in the daytime time zone TZ5 has decreased rapidly. From this, it is assumed that there is no one at home during the day. Also, from the change in power consumption at night, it is assumed that the husband's return time is later than usual. From the power consumption in these time zones TZ4, TZ5, and TZ6, it is assumed that the wife and child have returned to their parents' home.

[0050] After the middle ten days of the month following the month of household composition change, compared with the month of household composition change and the first ten days of the following month, the overall power consumption per day has increased. In particular, among the time zone TZ7 of one day, the power consumption in the daytime time zone has increased significantly. From this, for example, it is assumed that the mother and child have returned home, and that the air conditioner is running 24 hours a day to maintain the baby's physical condition.

[0051] <Detection algorithm> The control device 30D detects a childbirth household according to a detection algorithm. This detection algorithm is based on the perspective of comparing the median value of the power consumption during the daytime for six consecutive months with the median value of the power consumption in the same month of the previous year. The control device 30D calculates a childbirth score indicating the probability of being a childbirth household, and detects a childbirth household based on the calculated childbirth score. The control device 30D calculates, as the childbirth score, the maximum value over the year of the sum of the squares of the upward fluctuations from the median value of the same month of the previous year of the power consumption during the daytime for six consecutive months.

[0052] Incidentally, an upward fluctuation means that the median value of the power consumption during the daytime for six consecutive months exceeds the median value of the power consumption in the same month of the previous year. Also, the amount of upward fluctuation means the difference between the upward fluctuated median value of the power consumption during the daytime for six consecutive months and the median value of the power consumption in the same month of the previous year.

[0053] The detection algorithm is based on the following perspectives (A1) to (A4). (A1) Since the power consumption varies greatly depending on the season, there is a risk that the change in power consumption before and after childbirth may be masked by the seasonal difference. In this regard, seasonality can be removed by paying attention to the year-on-year comparison of the power consumption in the same month of the previous year.

[0054] (A2) Except for the time zone where the power consumption varies greatly depending on the at-home situation, there is a risk that the change in power consumption before and after childbirth may be small and be masked by other factors. In this regard, during the daytime time zone, the power consumption is more likely to vary greatly depending on the at-home situation. Therefore, by paying attention only to the daytime time zone, it is possible to make the difference in power consumption before and after childbirth more prominent.

[0055] (A3) The change in power consumption associated with childbirth continues for about half a year. If only a specific one month is focused on, it is difficult to distinguish from households with high power consumption by chance. In this regard, by paying attention to the power consumption for six consecutive months, it is possible to suitably capture the change in power consumption associated with childbirth.

[0056] (A4) The periods of maternity leave, childcare leave, or returning to one's hometown for childbirth vary from person to person. Therefore, the time when changes occur in the electricity consumption varies from person to person. For this reason, it is preferable to shift the six-month period of interest within a year by one month at a time and focus on the period when the change in electricity consumption is most prominent.

[0057] An example of the calculation procedure for the birth score is as follows. As shown in the graph of Fig. 5, for example, within one year from November 2019 to October 2020, there are seven ways to select six months as follows: (B1) to (B7).

[0058] (B1) November 2019 - April 2020 (B2) December 2019 - May 2020 (B3) January 2020 - June 2020 (B4) February 2020 - July 2020 (B5) March 2020 - August 2020 (B6) April 2020 - September 2020 (B7) May 2020 - October 2020 The control device 30D shifts the starting month of the six-month period of interest within a year by one month at a time and calculates the sum of the squares of the upward deviations from the median of the same month of the previous year in the current consumption for each period of (B1) to (B7). The electricity consumption is the electricity consumption during the daytime, for example, from 10:00 to 13:00.

[0059] The control device 30D calculates the sum of squares Σ using the following formula (1). Σ = S1 2 + S2 2 + S3 2 + S4 2 + S5 2 + S6 2 …(1) However, "S1 to S6" are the values of the upward deviations from the median of the same month of the previous year in the electricity consumption during the daytime for six consecutive months.

[0060] Note that when calculating the sum of squares Σ, the downward deviation from the median of the same month of the previous year in the daytime time zone for six consecutive months is not added by the control device 30D. As shown in the graph of Fig. 5, for example, in the six months from March 2020 to August 2020, only in April 2020 is there a downward deviation from the median. In this case, as expressed by the following formula (2), the value of the downward deviation from the median in April 2020 is not used in the calculation of the sum of squares Σ.

[0061] Σ = S1 2 + S3 2 + S4 2 + S5 2 + S6 2 …(2) However, "S1" is the value of the upward deviation from the median in March 2020, and "S3 to S6" are the values of the upward deviations from the median from May 2020 to August 2020.

[0062] Next, as shown in the following formula (3), the control device 30D takes the sum of squares Σ with the largest value among the sums of squares Σ of each period (B1) to (B7) max as the production score SCR birth and calculates it. This is to focus on the period in which the difference from the median of the same month of the previous year in power consumption is the most prominent.

[0063] SCR birth = Σ max …(3) The control device 10C determines that it is a production household in descending order of the value of the production score SCR birth . The control device 10C sorts each household in descending order of the value of the production score SCR birth and extracts a predetermined number of households in order from the top of the sorted list. The higher the value of the production score SCR birth , the higher the probability that the household is a production household. By extracting production households based on the production score SCR birth , the detection accuracy of production households can be ensured.

[0064] <Estimation accuracy of production> The production estimation accuracy is represented by, for example, the relationship between the household selection rate and the recall rate, or the relationship between the household selection rate and the precision rate. The household selection rate refers to the proportion of households selected from the population. The recall rate refers to the proportion of households determined to have given birth among the households that have given birth. The precision rate refers to the proportion of households that have given birth among the selected households.

[0065] The production estimation accuracy is as follows. Here, for example, consider the case of capturing 18 households that gave birth from November 2019 to October 2020 out of 197 households.

[0066] As shown in the top row of the list in Fig. 6, for example, when selecting the top 20 people with high values of the birth score SCR birth from 197 people, the household selection rate is 10%. When the household selection rate is 10%, the recall rate is 33%. That is, about 6 out of the 18 households that gave birth can be identified. Also, when the household selection rate is 10%, the precision rate is 30%. That is, among the 20 selected people, the actual number of households that gave birth is 6.

[0067] As shown in the graph of Fig. 7, for example, when randomly selecting 20 people from 197 people, the household selection rate is 10% and the recall rate is 10%. From this, it can be seen that the recall rate when selecting 20 people based on the birth score SCR birth is 3.3 times that when randomly selecting 20 people. That is, the certainty (degree of certainty) that the 20 people selected based on the birth score SCR birth are households that gave birth is 3.3 times that when randomly selecting 20 people. Thus, by using the birth score SCR birth based on the change in power consumption, the detection accuracy of households that gave birth can be improved regardless of the proportion of households selected.

[0068] <Effects of the First Embodiment> Therefore, according to the first embodiment, the following effects can be obtained. (1-1) Based on the changing trend of the electricity consumption common to production households, it is possible to effectively detect production households based on the change in electricity consumption. Specifically, production households are detected based on the change in daytime electricity consumption over at least six consecutive months. In production households, the electricity consumption during the daytime tends to increase for about six months from the period before and after childbirth. Therefore, production households can be detected based on the continuous increasing trend of daytime electricity consumption. Also, since only the change in electricity consumption needs to be focused on, production households can be detected more simply and efficiently. Incidentally, the specific period is at least six consecutive months, and it may be a longer period.

[0069] (1-2) Calculate the maximum value over the year of the sum of the squares of the upward deviations from the median of the same month of the previous year of the daytime electricity consumption over at least six consecutive months as SCR birth and detect production households based on the calculated production score SCR birth . The production score SCR birth is a value indicating the probability of being a production household. The higher the value of the production score SCR birth , the higher the probability that the household is a production household. In this way, by focusing on the increment of the current year's electricity consumption relative to the electricity consumption of the same month of the previous year, seasonality can be removed, that is, the influence of seasonal fluctuations in electricity consumption can be suppressed. Also, since the periods such as maternity leave, childcare leave, or giving birth at home vary from person to person, the time when changes appear in electricity consumption also varies from person to person. In this regard, by using the maximum value over the year of the sum of the squares as the production score SCR birth , it is possible to focus on the period when the change in electricity consumption is most prominent for each household. Therefore, the group extracted based on the production score SCR birth has a much higher detection accuracy of production households compared to a randomly extracted group.

[0070] (1-3) Changes in household composition can detect childbearing households. This detection result can be utilized in marketing or sales activities in a company. For example, due to changes in household composition associated with childbirth, etc., there may be a mismatch between the property being lived in and the household composition. Therefore, changes in household composition associated with childbirth, etc., can be regarded as an indication of property renovation or relocation.

[0071] <Second Embodiment> Next, a second embodiment in which the household composition change detection device is embodied will be described. This embodiment basically has the same configuration as the first embodiment shown in FIGS. 1 to 3 above. Therefore, the same members and configurations as those in the first embodiment are denoted by the same reference numerals, and detailed descriptions thereof are omitted.

[0072] The storage device 10B stores the attribute information of each household. The attribute information is associated with the identification information unique to the smart meter 30 of each household and the position information of the smart meter 30. The attribute information is obtained, for example, through a questionnaire survey for each household. An attribute is a characteristic or property that each household has, and includes information such as the number of household members, gender, and age group.

[0073] Note that the questionnaire survey may be conducted via the communication network 20. Also, the method for obtaining the attribute information is not limited to the questionnaire survey. The attribute information may be obtained using various methods regardless of whether the communication network 20 is used.

[0074] The control device 10C corrects the value of the childbirth score SCR birth according to the attributes of each household. The control device 10C reduces the value of the childbirth score SCR birth of households having attributes with a low probability of childbirth, that is, households other than those having attributes with a high probability of childbirth, based on the attribute information of each household. The attributes of households with a high probability of childbirth include the following two attributes (C1) and (C2).

[0075] (C1) A household with two people, both women in their 20s to 30s. (C2) A household with three or more people, all women in their 20s to 40s. Incidentally, the attributes (C1) and (C2) are selected based on the following three viewpoints (D1) to (D3).

[0076] (D1) The probability of giving birth in a household where the couple lives alone or in a non-child-rearing household is almost zero. (D2) It is common for the first child to be born by the age of 30. (D3) Cases of giving birth at the age of 50 or older are extremely rare.

[0077] The control device 10C corrects the birth score SCR of households other than those with attributes having a high probability of giving birth using the following formula (4). birth To correct. SCR birth_c = SCR birth ·WT = SCR birth ·(1 / 10000) …(4) However, "SCR birth_c " is the corrected birth score. "SCR birth " is the birth score before correction calculated based on the previous formula (3). "WT" is a weight value, which is set to a fixed value of "1 / 10000", for example.

[0078] By using formula (4), the birth score SCR of households other than those with attributes having a high probability of giving birth birth is corrected to the value of "1 / 10000". The birth prediction accuracy is as follows. Here too, for example, consider the case of capturing 18 households that gave birth from November 2019 to October 2020 out of 197 households.

[0079] As shown at the top of the list in Figure 8, for example, from 197 people, the birth score SCR birthWhen selecting the top 20 people with high values, the household selection rate is 10%. When the household selection rate is 10%, the recall rate is 44%. That is, 8 out of the 18 households that gave birth can be predicted. Also, when the household selection rate is 10%, the precision rate is 40%. That is, out of the 20 people selected, 8 are the actual households that gave birth.

[0080] As shown in the graph of Fig. 9, for example, when randomly selecting 20 people from 197 people, the household selection rate is 10% and the recall rate is 10%. From this, it can be seen that the production score SCR birth and the recall rate when selecting 20 people based on the attributes of the households are 4.4 times that when randomly selecting 20 people. That is, the production score SCR birth and the probability that the 20 people selected based on the attributes of the households are the households that gave birth are 4.4 times that when randomly selecting 20 people. In this way, by using the production score SCR birth and the attributes of the households, the detection accuracy of the households that gave birth can be improved regardless of the proportion of the households selected.

[0081] Incidentally, although the details are omitted, by appropriately narrowing down based on the attributes of each household, it is possible to narrow down to 101 households with a high probability of being the households that gave birth. According to this narrowing down, when selecting 20 people from 197 people, the recall rate is 18%. That is, 3.2 out of the 18 households that gave birth can be predicted. From this, it can be seen that the production score SCR birth and the recall rate when selecting 20 people based on the attributes of the households are 2.5 times (= 8 households / 3.2 households) that when selecting 20 people based on the attributes of each household only. That is, by combining the power data and the attributes of each household, the households that gave birth can be extracted more effectively than when extracting the households that gave birth using only the attributes of each household.

[0082] <Effects of the Second Embodiment> Therefore, according to the second embodiment, the following effects can be obtained. (2-1) Based on the attribute information of each household, reduce the value of the birth score SCR of households other than those with attributes that are considered highly likely to give birth. birth This makes it less likely to detect households with attributes that are considered to have a low probability of giving birth. Therefore, it is possible to more effectively detect birth households.

[0083] <Third Embodiment> Next, a third embodiment in which the household composition change detection device is embodied will be described. This embodiment basically has the same configuration as the first embodiment shown in FIGS. 1 to 3 above. Therefore, the same members and configurations as those in the first embodiment are denoted by the same reference numerals, and detailed descriptions thereof are omitted.

[0084] The detection device 10 detects households in which the number of household members has decreased. Situations where the number of household members decreases include, for example, the independence of children, and the separation or death of household members. As a result of exploring common features when the number of household members changes (here, decreases) in various households with different living patterns, in households where the number of household members has decreased, regardless of the age of the person leaving the household, the electricity consumption in the morning on holidays tends to decrease. This is presumably due to the fact that, for example, there are not a few people staying at home in the morning on holidays. Therefore, based on the tendency of the change in electricity consumption in the morning on holidays, it is possible to detect households with a high probability of a decrease in the number of household members.

[0085] <Detection Algorithm> The control device 10C detects households in which the number of household members has decreased according to the detection algorithm. The control device 10C determines that it is a household in which any one of the events of independence, separation, and death has occurred, in ascending order of the year-on-year ratio of electricity consumption in the morning on holidays being low. The control device 10C sorts each household in ascending order of the value of the year-on-year ratio of electricity consumption in the morning on holidays, and extracts a predetermined number of households in order from the top of the sorted list.

[0086] <Estimation Accuracy of Decrease in the Number of Household Members> The estimation accuracy (independent, separated, and deceased household estimation accuracy) of households with a decreasing household size is as follows. Here, for example, we will consider the case of capturing 18 households with a decreasing household size out of 197 households. Households with a decreasing household size include independent households, separated households, and deceased households.

[0087] As shown in the top row of the list in Figure 10, for example, when selecting the top 22 people with a smaller year-on-year value of power consumption during the morning on holidays from 197 people, when truncating the digits after the units place, the household selection rate is 10%. When the household selection rate is 10%, the recall rate is 28%. That is, 5 out of the 18 households with a decreasing household size can be identified. Also, when the household selection rate is 10%, the precision rate is 23%. That is, out of the 22 selected people, 5 households actually have a decreasing household size.

[0088] As shown in the graph of Figure 11, for example, when randomly selecting 22 people from 197 people, the household selection rate is 10% and the recall rate is 10%. From this, it can be seen that the recall rate when selecting 22 people based on power data, that is, based on the change in power consumption, is 2.8 times that when randomly selecting 22 people. Also, the probability that the household size of the 22 people selected based on the change in power consumption has decreased is 2.8 times the probability when randomly selecting 22 people (5 households / 1.8 households). Thus, by using power data, regardless of the proportion of households selected, households with a decreasing household size can be effectively and efficiently detected.

[0089] <Effect of the Third Embodiment> Therefore, according to the third embodiment, the following effects can be obtained. (3-1) Based on the changing trend of power consumption common to households with a decreasing household size, it is possible to effectively detect households with a decreasing household size based on the change in power consumption. In households with a decreasing household size, regardless of the age of the person leaving the household, the power consumption in the morning on holidays tends to decrease. Therefore, it is possible to detect households with a decreasing household size based on the changing trend of power consumption in the morning on holidays. Also, since only the change in power consumption needs to be focused on, it is possible to detect newly formed households more simply and efficiently.

[0090] (3-2) Detect households with a decreasing household size based on the year-on-year comparison of power consumption in the morning on holidays. The smaller the value of the year-on-year comparison of power consumption in the morning on holidays for a household, the higher the probability that the household has a decreasing household size. By paying attention to the year-on-year comparison of power consumption, seasonality can be removed, that is, the influence of seasonal fluctuations in power consumption can be suppressed. Therefore, it is possible to effectively detect households with a decreasing household size.

[0091] (3-3) It is possible to detect households with a decreasing household size as a change in household composition. This detection result can be utilized in marketing or sales activities in a company. For example, due to changes in household size such as independence, separation, or death, there may be a mismatch between the property being lived in and the household composition. Therefore, a decrease in household size can be regarded as an indication of property renovation or relocation.

[0092] Note that the third embodiment may be implemented with the following modifications. That is, the detection device 10 detects households with an increasing household size based on the same perspective as when detecting households with a decreasing household size. In households with an increasing household size, the power consumption in the morning on holidays tends to increase. Therefore, it is possible to detect households with an increasing household size based on the changing trend of power consumption in the morning on holidays. Since only the change in power consumption needs to be focused on, it is possible to detect newly formed households more simply and efficiently.

[0093] <Fourth Embodiment> Next, a fourth embodiment in which the household composition change detection device is embodied will be described. This embodiment basically has the same configuration as the first embodiment shown in FIGS. 1 to 3 above. For this reason, the same members and configurations as those in the first embodiment are denoted by the same reference numerals, and detailed descriptions thereof are omitted.

[0094] Similar to the third embodiment above, the detection device 10 detects a household in which the number of household members has decreased based on a change in power consumption. As described above, in a household where the number of household members has decreased, the power consumption in the morning on holidays tends to decrease regardless of the age of the person leaving the household. However, there is a time zone specific to the age group in which a characteristic change in power consumption appears only in the age of the person who has left.

[0095] For example, when household members in the 20s to 30s generation leave the household, the power consumption at night on weekdays tends to decrease. This tendency is considered to be due to the following. For example, it is assumed that the parents in their 20s to 30s who leave the household are older and often go to bed earlier than their children. Therefore, it is considered that the decrease in power consumption associated with the departure of household members in the 20s to 30s generation is prominent.

[0096] Also, when household members in the elderly generation aged 60 or above leave the household, the power consumption in the morning on weekdays tends to decrease. This tendency is considered to be due to the following. For example, it is assumed that the occupancy rate of people other than the elderly is low in the morning on weekdays. Therefore, it is considered that the decrease in power consumption associated with the departure of household members in the elderly generation is prominent.

[0097] Therefore, based on the change tendency of the power consumption at night on weekdays, it is possible to detect a household with a high probability that household members in the 20s to 30s generation have left. Also, based on the change tendency of the power consumption in the morning on weekdays, it is possible to detect a household with a high probability that household members in the elderly generation have left.

[0098] Note that the storage device 10B stores the attribute information of each household. The attribute information is associated with the identification information unique to the smart meter 30 of each household and the location information of the smart meter 30. The attribute information is obtained, for example, through a questionnaire survey for each household. An attribute is a characteristic or property that each household has, and includes information such as the number of household members, gender, and age group.

[0099] <Detection algorithm> The control device 10C detects a household whose household size has decreased due to the departure of household members of a specific age group according to the detection algorithm. In other words, the control device 10C selectively detects a household in which household members of a specific age group have left. The control device 10C detects a household in which people of a specific age group have left based on the change in power consumption in the following two time periods (E1) and (E2).

[0100] (E1) A time period in which a characteristic change in power consumption is observed common to all age groups. Here, it is the morning on a holiday. (E2) A time period unique to an age group in which a characteristic change in power consumption is observed according to the age group. For the sub-generation from the 20s to the 30s, it is the night on a weekday, and for the elderly generation of 60 years old and above, it is the morning on a weekday.

[0101] The control device 10C calculates the departure probability for each age group using logistic regression. Logistic regression is an algorithm for classification. The control device 10C calculates the departure probability P child of the sub-generation based on the following equation (5).

[0102] P child =F child ·inv_logit(α0 - α1·Y1 - α2·Y2)…(5) However, "F child" is the value of a flag indicating the existence of a child generation aged from 20s to 30s in a household. "logit()" is the logit function, and "inv_logit" represents the inverse function of the logit function. "α0", "α1", "α2" are weights determined through the learning of power data by logistic regression. "Y1" is the year-on-year ratio of power consumption in the morning on holidays compared to the same month of the previous year. "Y2" is the year-on-year ratio of power consumption at night on weekdays compared to the same month of the previous year.

[0103] The control device 10C calculates the departure probability P of the elderly generation senior based on the following formula (6). P senior = F senior ·inv_logit(α0 - α1·Y1 - α2·Y3)…(6) However, "F senior " is the value of a flag indicating the existence of an elderly generation aged 60 or above in a household. "logit()" is the logit function, and "inv_logit" represents the inverse function of the logit function. "α0", "α1", "α2" are weights determined through the learning of power data by logistic regression. "Y1" is the year-on-year ratio of power consumption in the morning on holidays compared to the same month of the previous year. "Y3" is the year-on-year ratio of power consumption in the morning on weekdays compared to the same month of the previous year.

[0104] The control device 10C sets the value of the flag F child based on the attribute information of each household stored in the storage device 10B. When there is a child generation aged from 20s to 30s in the household, the control device 10C sets the value of the flag F child to "1". When there is no child generation aged from 20s to 30s in the household, the control device 10C sets the value of the flag F child to "0".

[0105] The control device 10C sets the value of the flag F senior based on the attribute information of each household stored in the storage device 10B. When there is an elderly generation aged 60 or above in the household, the control device 10C sets the value of the flag F seniorSet the value of senior to "1". When there is no elderly generation aged 60 or above in the household, the control device 10C sets the value of flag F to "0".

[0106] The control device 10C calculates the weighted sum of the year-on-year comparison of electricity consumption, and obtains the logit, which is the value obtained by performing a logit transformation on the calculated weighted sum, as the result of logistic regression. The control device 10C performs an inverse transformation of the logit transformation on the logit obtained as the result of logistic regression to calculate the probability P child of the departure of the younger generation or the probability P senior of the departure of the elderly generation. The probability P child of the departure of the younger generation or the probability P senior of the departure of the elderly generation is a value between "0" and "1". Incidentally, the logit refers to a value represented by the natural logarithm of the odds of the probability p. The odds refer to the ratio of the probability of an event occurring to the probability of the event not occurring (= p / 1 - p) for an event occurring with probability p.

[0107] <Estimation accuracy of household size decrease: independent younger generation> The estimation accuracy (child independent estimation accuracy) of an independent household where the younger generation is independent is as follows. Here, for example, consider the case of capturing 12 households where the younger generation is independent out of 197 households.

[0108] As shown in the top row of the list in Figure 12, for example, when selecting the top 20 people with a high probability P child of departure of the younger generation from 197 people, the household selection rate is 10%. When the household selection rate is 10%, the recall rate is 42%. That is, 5 out of 12 households where the younger generation is independent can be guessed. Also, when the household selection rate is 10%, the precision rate is 25%. That is, out of the 20 people selected, 5 are the actual independent households.

[0109] As shown in the graph of Figure 13, for example, when randomly selecting 20 people from 197 people, the household selection rate is 10% and the recall rate is 10%. From this, the probability P of the departure of the younger generation based on electricity data and household attributeschild When 20 households are selected according to [the probability], the reproduction rate is found to be 4.2 times that when 20 households are randomly selected. That is, the probability of leaving P of the offspring generation child The probability that the 20 households selected according to [the probability] are independent households is 4.2 times that when 20 households are randomly selected. Thus, by using the probability of leaving P based on power data and household attributes, the detection accuracy of independent households can be improved regardless of the proportion of households to be selected. child

[0110] Incidentally, although details are omitted, by appropriately narrowing down based on the attributes of each household, it is possible to narrow down to 127 households with a high probability of being independent households, that is, households in which offspring from their 20s to 30s exist as household members. Assuming this narrowing down, when 20 households are selected from 197 households (substantially 127 narrowed-down households), that is, when the household selection rate is 10%, the reproduction rate is 16%. It is possible to hit 1.9 households out of 12 independent offspring households. From this, it can be seen that the reproduction rate when 20 people are selected according to the probability of leaving P of the offspring generation is 2.6 times (= 5 households / 1.9 households) that when 20 people are selected based only on the attributes of each household. That is, it is possible to effectively extract independent households by using a combination of power data and the attributes of each household rather than using only the attributes of each household. child

[0111] <Estimation accuracy of household population decrease: Death of the elderly> The estimation accuracy (estimation accuracy of the death of the elderly) of households where the elderly generation has passed away is as follows. Here, for example, the case of capturing 4 households where the elderly generation has left among 197 households is considered.

[0112] As shown at the top of the list in Fig. 14, for example, from 197 people, the probability of leaving P of the elderly generation senior ​​When selecting the top 20 people with high [[ID=]], the household selection rate is 10%. When the household selection rate is 10%, the recall rate is 75%. That is, three out of the four households where the elderly generation has left can be identified. Also, when the household selection rate is 10%, the precision rate is 15%. That is, out of the 20 people selected, three households are those where the elderly generation has truly left.

[0113] As shown in the graph of Fig. 15, for example, when randomly selecting 20 people from 197 people, the household selection rate is 10% and the recall rate is 10%. From this, the probability P of the departure of the elderly generation based on power data and household attributes senior It can be seen that the recall rate when selecting 20 people according to is 7.5 times that when randomly selecting 20 people. That is, the probability P of the departure of the elderly generation senior The probability that the 20 people selected according to are independent households is 7.5 times that when randomly selecting 20 people. Thus, by using the probability P of the departure of the elderly generation based on power data and household attributes senior the detection accuracy of households where the elderly generation has left can be improved regardless of the proportion of households selected.

[0114] Incidentally, although the details are omitted, by appropriately narrowing down based on the attributes of each household, it is possible to narrow down to 52 households, which are households with a high probability of being households where the elderly generation has left, that is, households with household members in the age group of 60 and above. Assuming this narrowing down, when selecting 20 households from 197 households (substantially the 52 narrowed-down households), that is, when the household selection rate is 10%, the recall rate is 38%. 1.5 out of the four households where the elderly generation has left can be identified. From this, it can be seen that the recall rate when selecting 20 people according to the probability P of the departure of the elderly generation senior is 2.0 times (= 3 households / 1.5 households) that when selecting 20 people based only on the attributes of each household. That is, using a combination of power data and the attributes of each household can more effectively extract households where the elderly generation has left than using only the attributes of each household.

[0115] <Effect of the Fourth Embodiment> (4-1) By finding the tendency of change in the amount of power used that is common for each age group of household members who have left the household, it is possible to effectively and efficiently detect the household in which household members of a specific age group have left based on the change in the amount of power used. For example, there may be a time period in which characteristic changes in power consumption are observed depending on the age of the household members leaving the household. In this case, based on the tendency of change in the amount of power used in the time period in which characteristic changes in power consumption are observed only for a specific age group, it is possible to detect the household in which the number of household members has decreased due to the departure of household members of a specific age group. Also, since it is only necessary to focus on the change in the amount of power used, it is possible to more easily and efficiently detect the household in which the number of household members has decreased due to the departure of household members of a specific age group.

[0116] (4-2) In a household in which the number of household members has decreased due to the departure of the child generation from their 20s to 30s, the amount of power used on weekday nights tends to decrease. Therefore, based on the tendency of change in the amount of power used on weekday nights, it is possible to detect the household in which the number of household members has decreased due to the departure of the child generation from their 20s to 30s.

[0117] (4-3) In a household in which the number of household members has decreased due to the departure of the elderly generation aged 60 or above, the amount of power used on weekday nights tends to decrease. Therefore, based on the tendency of change in the amount of power used in the early morning on weekdays, it is possible to detect the household in which the number of household members has decreased due to the departure of the elderly generation aged 60 or above.

[0118] (4-4) Based on the year-on-year comparison of power consumption on ordinary weeknights, households with a decreasing household size due to the departure of the younger generation aged from their 20s to 30s are detected. The lower the value of the year-on-year comparison of power consumption on ordinary weeknights, the higher the probability that the household is one where the younger generation aged from their 20s to 30s has left. Also, based on the year-on-year comparison of power consumption in the morning on weekdays, households with a decreasing household size due to the departure of the elderly generation aged 60 or above are detected. The lower the value of the year-on-year comparison of power consumption in the morning on weekdays, the higher the probability that the household is one where the elderly generation aged 60 or above has left. In this way, by focusing on the year-on-year comparison of power consumption, seasonality can be removed, that is, the influence caused by seasonal fluctuations in power consumption can be suppressed. Therefore, households with a decreasing household size can be effectively detected.

[0119] (4-5) By appropriately narrowing down based on the attribute information of each household before the change in household composition, households with household members of a specific generation can be narrowed down. That is, based on the attribute information of each household, households where household members other than the generation of interest have left can be easily excluded. Therefore, households where household members of a specific generation have left can be detected more effectively and efficiently. For example, by narrowing down households with the younger generation aged from their 20s to 30s, it is possible to efficiently detect households with a high probability of being those where the younger generation has left. Also, by narrowing down households with the elderly generation aged 60 or above, it is possible to efficiently detect households with a high probability of being those where the elderly generation has left.

[0120] (4-6) Households where household members of a specific generation have left can be detected as a change in household composition. This detection result can be utilized in marketing or sales activities in enterprises. For example, when household members of a specific generation leave the household, there may be a mismatch between the property they live in and the household composition. Therefore, the departure of household members of a specific generation can be regarded as a sign of property renovation or relocation.

[0121] <Fifth Embodiment> Next, a fifth embodiment in which the household composition change detection device is embodied will be described. This embodiment basically has the same configuration as the first embodiment shown in FIGS. 1 to 3 above. Therefore, the same members and configurations as those in the first embodiment are denoted by the same reference numerals, and detailed descriptions thereof are omitted.

[0122] The detection device 10 detects a household whose work pattern has changed, here, a household that has started working from home. As a result of exploring the common characteristics among households that have started working from home in various households with different living patterns, in households that have started working from home, the electricity consumption during weekdays tends to increase during the day. This is considered to be due to the use of lighting fixtures, air conditioners, and electronic devices in the home along with the start of working from home. In particular, characteristic changes often appear in the electricity consumption from April to May. This is thought to be because many companies set a one-year accounting period from April to the following March, and work pattern changes are often made with the beginning of the accounting period as a target. Therefore, it is possible to detect households with a high probability of starting to work from home based on the change trend of electricity consumption from April to May and during weekdays. Incidentally, "during the day" refers to a period from, for example, 9:00 to 17:00.

[0123] <Detection Algorithm> The control device 10C detects a household that has started working from home according to the detection algorithm. The control device 10C determines that it is a household that has started working from home in descending order of the value of the year-on-year comparison of electricity consumption from April to May and during weekdays from 9:00 to 17:00. The control device 10C rearranges each household in descending order of the value of the year-on-year comparison of electricity consumption from April to May and during weekdays from 9:00 to 17:00, and extracts a predetermined number of households in order from the top of the rearrangement.

[0124] <Estimation Accuracy of Starting to Work from Home> The estimation accuracy (work-from-home start estimation accuracy) of households that started working from home is as follows. Here, for example, we will consider the case of capturing 29 households out of 197 households that started working from home.

[0125] As shown in the top row of the list in Fig. 16, for example, when selecting the top 20 people with a large year-on-year value of power consumption from 197 people during the period from April to May and during weekdays from 9:00 to 17:00, the household selection rate is 10%. When the household selection rate is 10%, the recall rate is 21%. That is, 6 out of the 29 households that started working from home can be identified. Also, when the household selection rate is 10%, the precision rate is 30%. That is, out of the 20 people selected, 6 households actually started working from home.

[0126] As shown in the graph of Fig. 17, for example, when randomly selecting 20 people from 197 people, the household selection rate is 10% and the recall rate is 10%. From this, it can be seen that the recall rate when selecting 20 people based on power data, that is, based on the change in power consumption, is 2.1 times that when randomly selecting 20 people. Also, the probability that the 20 people selected based on the change in power consumption are households that started working from home is 2.0 times (6 households / 2.9 households) the probability when randomly selecting 20 people. Thus, by using power data, regardless of the proportion of households selected, households that started working from home can be effectively and efficiently detected.

[0127] <Effect of the Fifth Embodiment> (5-1) By finding the characteristics or trends of the changes in the electricity consumption common to households that have started working from home, it is possible to effectively and efficiently detect households that have started working from home based on the changes in electricity consumption. In households that have started working from home, the electricity consumption tends to increase during the period from April to May and on weekdays from 9:00 to 17:00. Therefore, it is possible to detect households that have started working from home based on the change trend of the electricity consumption during the period from April to May and on weekdays from 9:00 to 17:00. Also, since only the change in the electricity consumption amount needs to be focused on, it is possible to detect households with new births more simply and efficiently.

[0128] (5-2) Detect households that have started working from home based on the year-on-year comparison of the electricity consumption in the period from April to May and on weekdays from 9:00 to 17:00. For example, the higher the year-on-year value of the electricity consumption in the period from April to May and on weekdays from 9:00 to 17:00, the higher the probability that the household has started working from home. By paying attention to the year-on-year comparison of the electricity consumption, seasonality can be removed, that is, the influence of seasonal fluctuations in electricity consumption can be suppressed. Therefore, it is possible to effectively detect households that have started working from home.

[0129] (5-3) It is possible to detect households that have started working from home as a change in household composition. This detection result can be utilized for marketing or sales activities in a company. For example, starting to work from home may cause a mismatch between the property where one lives and the household composition or lifestyle. Therefore, starting to work from home can be regarded as a sign of property renovation or relocation.

[0130] Note that the fifth embodiment may be implemented with the following modifications. That is, the detection device 10 detects households that have started working from home based on the changing trend of power consumption during weekdays in the daytime. In households that have started working from home, the power consumption from 9:00 to 17:00 on weekdays tends to increase. Therefore, it is possible to detect households that have started working from home based on the changing trend of power consumption from 9:00 to 17:00 on weekdays. Since only the change in power consumption needs to be focused on, it is possible to detect households with newborns more simply and efficiently.

[0131] <Sixth Embodiment> Next, a sixth embodiment in which the household composition change detection device is embodied will be described. This embodiment basically has the same configuration as the first embodiment shown in FIGS. 1 to 3 above. Therefore, the same members and configurations as those in the first embodiment are denoted by the same reference numerals, and detailed descriptions thereof are omitted.

[0132] The detection device 10 detects households with children entering elementary school. A household with children entering elementary school refers to a household in which there are household members who have entered elementary school. As a result of exploring the common characteristics of households with children entering elementary school, in households with children entering elementary school, the power consumption tends to increase during the period from before noon to after noon based on noon in August. This is presumably due to the fact that, for example, kindergartens basically do not have summer vacations, while elementary schools do. Therefore, based on the changing trend of power consumption during the period from before noon to after noon in August, it is possible to detect households with a high probability of having household members who have entered elementary school. Incidentally, the period from before noon to after noon refers to, for example, the period from 10:30 to 13:30.

[0133] <Detection Algorithm> The control device 10C detects primary school enrollment households according to a detection algorithm. The control device 10C determines that a household is a primary school enrollment household in descending order of the year-on-year value of the power consumption during the period from 10:30 to 13:30 in August. The control device 10C sorts each household in descending order of the year-on-year value of the power consumption during the period from 10:30 to 13:30 in August, and extracts a predetermined number of households in order from the top of the sorted list.

[0134] <Estimation accuracy of primary school enrollment> The estimation accuracy of primary school enrollment households is as follows. Here, for example, consider the case of capturing 18 households out of 63 households with infants in the previous year who entered primary school. Incidentally, households with infants in the previous year are extracted based on attribute information.

[0135] As shown in the graph of Fig. 18, for example, when selecting the top 6 households with the largest year-on-year value of power consumption during the period from 10:30 to 13:30 in August from 63 households with infants in the previous year, the household selection rate is 10% and the recall rate is 22%. That is, 4 out of 18 households that entered primary school can be identified.

[0136] On the other hand, when randomly selecting 6 households from 63 households with infants in the previous year, the household selection rate is 10% and the recall rate is 10%. From this, it can be seen that the recall rate when selecting 6 households based on power data is 2.2 times that when randomly selecting 6 households. Also, the probability that 20 households selected based on the change in power consumption are primary school enrollment households is 2.2 times the probability when randomly selecting 6 households (4 households / 1.8 households). Thus, by using power data, primary school enrollment households can be effectively and efficiently detected regardless of the proportion of households selected.

[0137] <Effect of the sixth embodiment> (6-1) By finding the characteristics or trends in the changes in the electricity consumption common to households with children entering elementary school, it is possible to effectively and efficiently detect households with children entering elementary school based on the changes in electricity consumption. In households with children entering elementary school, the electricity consumption tends to increase during the period from early morning to afternoon in August. Therefore, based on the change trend of the electricity consumption during the period from early morning to afternoon in August, it is possible to detect the households in which there are household members who have entered elementary school. Also, since only the change in the electricity consumption needs to be focused on, it is possible to detect households with children entering elementary school more simply and efficiently.

[0138] (6-2) Detect households with children entering elementary school based on the year-on-year comparison of the electricity consumption during the period from early morning to afternoon in August with the same month of the previous year. The higher the value of the year-on-year comparison of the electricity consumption during the period from early morning to afternoon in August for a household, the higher the probability that the household is a household with children entering elementary school. By paying attention to the year-on-year comparison of the electricity consumption, seasonality can be removed, that is, the influence caused by seasonal fluctuations in electricity consumption can be suppressed. Therefore, it is possible to effectively detect households with children entering elementary school.

[0139] (6-3) Households in which it is confirmed that there are infants are extracted in advance based on the attribute information of each household. By paying attention to the change in the electricity consumption during the period from early morning to afternoon in August for the extracted households, it is possible to more efficiently detect households with children entering elementary school.

[0140] (6-4) It is possible to detect households with children entering elementary school as a change in household composition. This detection result can be utilized in marketing or sales activities in a company. For example, due to the change in household composition accompanying the entry of children into elementary school, there may be a mismatch between the property where one lives and the household composition. Therefore, the change in household composition accompanying the entry of children into elementary school can be grasped as a sign of property renovation or relocation.

Explanation of Signs

[0141] 10… Detection device (Household composition change detection device) 10C… Control device 20…Communication network 30…Smart meter

Claims

1. A household composition change detection device having a control device that obtains the power consumption of each household via a communication network from a power meter having a communication function, wherein the control device is configured to detect a change in the household composition based on a change trend of power consumption common to the change in the household composition to be captured, the change in the household composition is a birth, the control device is configured to detect a birth household, which is a household whose household composition has changed due to a birth, based on a change in daytime power consumption over at least a continuous six-month period, the control device further calculates, as a birth score indicating the probability of being a birth household, the maximum value over the year of the sum of the squares of the upward deviations from the median of the same month of the previous year of the daytime power consumption over at least a continuous six-month period, and is configured to detect a birth household based on the calculated birth score. A household composition change detection device.

2. It further has a storage device configured to store attribute information of each household, wherein the control device is configured to decrease the value of the birth score of a household having an attribute with a low probability of giving birth based on the attribute information of each household stored in the storage device, according to claim 1. The household composition change detection device described.

3. A household composition change detection device having a control device that obtains the power consumption of each household via a communication network from a power meter having a communication function, and a storage device configured to store attribute information of each household, wherein the control device is configured to detect a change in the household composition based on a change trend of power consumption common to the change in the household composition to be captured, the change in the household composition is a decrease in the number of household members, the control device is based on the attribute information of each household stored in the storage device and the change in power consumption in a time period set as having a characteristic change in power consumption according to the age of the household members leaving the household. It is configured to detect a household in which the number of household members has decreased due to the departure of household members of a specific age group, the control device is configured to detect a household in which the number of household members has decreased due to the departure of the child generation from their 20s to 30s based on the change in power consumption on weekdays at night. A household composition change detection device.

4. A household composition change detection device having a control device that acquires the electricity consumption of each household via a communication network from an electricity meter having a communication function, and a storage device configured to store the attribute information of each household, The control device is configured to detect a change in the household composition based on a change trend in electricity consumption common to the change in the household composition to be captured, The change in the household composition is a decrease in the number of household members, The control device is configured to detect a household in which the number of household members has decreased due to the departure of household members of a specific age group based on the attribute information of each household stored in the storage device and the change in electricity consumption in a time period set as having a characteristic change in electricity consumption according to the age of the household members leaving the household, The control device is configured to detect a household in which the number of household members has decreased due to the departure of elderly people aged 60 or above based on the change in electricity consumption during the morning on weekdays. A household composition change detection device.

5. A household composition change detection method in which a control device acquires the electricity consumption of each household via a communication network from an electricity meter having a communication function, and detects a change in the household composition based on a change trend in electricity consumption common to the change in the household composition to be captured, The change in the household composition is a birth, The control device includes detecting a birth household, which is a household whose household composition has changed due to a birth, based on the change in daytime electricity consumption over at least a continuous six-month period, The control device further includes calculating, as a birth score indicating the probability of being a birth household, the maximum value over the year of the sum of the squares of the upward deviations from the median of the same month of the previous year of the daytime electricity consumption over at least a continuous six-month period, and detecting a birth household based on the calculated birth score. A household composition change detection method.

Images