Data expansion device, data expansion method, and program

Abstract

This invention provides a data expansion device that can increase the amount of historical performance data when sufficient historical performance data has not been accumulated. [Solution] The data expansion device includes means for acquiring time history data relating to the operation of the equipment, means for copying the time history data to generate duplicate data, means for changing the time information included in the duplicate data, means for varying the values ​​relating to the operation of the equipment included in the duplicate data within a predetermined range, and means for merging the time history data and the duplicate data.

Description

Technical Field

[0001] The present disclosure relates to a data augmentation device, a data augmentation method, and a program.

Background Art

[0002] A control method for operating an air conditioner aiming at achieving both economy and comfort within an allowable range of the target power consumption has been proposed (for example, Patent Document 1). In the control method described in Patent Document 1, after setting the annual target power consumption, the power consumption for each time period necessary to perform comfortable air conditioning control for the user within the range of the target power consumption is calculated, and the air conditioner is operated based on the calculated power consumption. In the control method described in Patent Document 1, if the initially set annual target power consumption is not appropriate, it is impossible to achieve both economy and comfort. For example, if the annual target power consumption is small, comfort is impaired, and if the annual target power consumption is large, costs may increase.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In order to achieve both economy and comfort, it is necessary to appropriately set the annual target power consumption. For this purpose, for example, a method of estimating the annual target power consumption based on past performance data or the like can be considered. However, if sufficient past performance data has not been accumulated, it may be impossible to accurately estimate the annual target power consumption.

[0005] Therefore, an object of the present disclosure is to provide a data augmentation device, a data augmentation method, and a program that can solve the above problems.

Means for Solving the Problems

[0006] According to one aspect of the present disclosure, the data expansion device includes means for acquiring time history data relating to the operation of equipment; means for copying the time history data to generate duplicate data; means for changing time information included in the duplicate data; means for varying values ​​relating to the operation of equipment included in the duplicate data within a predetermined range; and means for merging the time history data and the duplicate data.

[0007] According to one aspect of this disclosure, the data augmentation method involves a computer acquiring time history data relating to the operation of equipment, copying the time history data to generate duplicate data, modifying the time information contained in the duplicate data, varying the values ​​relating to the operation of the equipment contained in the duplicate data within a predetermined range, and merging the time history data and the duplicate data.

[0008] According to one aspect of the present disclosure, the program causes a computer to perform the following processes: acquire time history data relating to the operation of equipment, copy the time history data to generate duplicate data, modify the time information contained in the duplicate data, vary the values ​​relating to the operation of the equipment contained in the duplicate data within a predetermined range, and merge the time history data and the duplicate data. [Effects of the Invention]

[0009] According to this disclosure, if sufficient historical performance data has not been accumulated, the amount of performance data can be increased. [Brief explanation of the drawing]

[0010] [Figure 1] This diagram shows the overall configuration of the air conditioning system according to the first embodiment. [Figure 2] This figure shows a schematic configuration of the power consumption estimation unit according to the first embodiment. [Figure 3] This figure shows an example of actual power consumption data according to the first embodiment. [Figure 4]This figure shows an example of a predictive model according to the first embodiment. [Figure 5] This is a first flowchart showing an example of the annual power consumption estimation process according to the first embodiment. [Figure 6] This is a second flowchart showing an example of the annual power consumption estimation process according to the first embodiment. [Figure 7] This is a third flowchart showing an example of the annual power consumption estimation process according to the first embodiment. [Figure 8] This diagram shows a schematic configuration of the power consumption control unit according to the first embodiment. [Figure 9] This figure shows an example of the coefficients set for each month according to the first embodiment. [Figure 10] This figure shows an example of the cooling evaluation coefficient characteristics and heating evaluation coefficient characteristics according to the first embodiment. [Figure 11] This figure shows an example of monthly target power consumption according to the first embodiment. [Figure 12] This is the first diagram illustrating the method for calculating the target power consumption per unit of time according to the first embodiment. [Figure 13] This is a second diagram illustrating the method for calculating the target power consumption per unit of time according to the first embodiment. [Figure 14] This diagram illustrates the power consumption control process according to the first embodiment. [Figure 15] This is a flowchart showing an example of power consumption control processing according to the first embodiment. [Figure 16] This is a flowchart illustrating an example of the initial processing according to the first embodiment. [Figure 17] This figure shows an example of the processing cycle for each process that constitutes the loop processing according to the first embodiment. [Figure 18] This is a flowchart showing an example of failure avoidance processing according to the first embodiment. [Figure 19] This figure shows a schematic configuration of the power consumption estimation unit according to the second embodiment. [Figure 20]It is a flowchart showing an example of data expansion processing according to the second embodiment. [Figure 21] It is a diagram showing an example of the data after expansion according to the second embodiment. [Figure 22] It is a diagram showing a schematic configuration of the power consumption estimation unit according to the third embodiment. [Figure 23] It is a diagram for explaining a method of selecting similar properties according to the third embodiment. [Figure 24] It is a flowchart showing an example of the estimation process of the annual power consumption according to the third embodiment. [Figure 25] It is a diagram showing an example of the hardware configuration of the centralized monitoring device according to each embodiment.

Mode for Carrying Out the Invention

[0011] <Embodiment> Hereinafter, the air conditioning system according to each embodiment will be described with reference to the drawings. FIG. 1 is a diagram showing the overall configuration of the air conditioning system according to the first embodiment. The air conditioning system 1 includes an air conditioner 3 and a centralized monitoring device 8 that controls the air conditioner 3. In the present embodiment, the air conditioner 3 is a packaged air conditioner for buildings, and includes one outdoor unit 4 and a plurality of indoor units 5. The outdoor unit 4 and each indoor unit 5 are connected via a refrigerant pipe 6. The number of outdoor units 4 and indoor units 5 shown in FIG. 1 is an example, and is not limited to the number shown. Also, there may be a plurality of air conditioners 3 monitored and controlled by the centralized monitoring device 8. In this case, the number of outdoor units and indoor units included in one air conditioner 3 may be different from the number of outdoor units and indoor units included in other air conditioners 3.

[0012] The air conditioner 3 is provided, for example, in the building 2, the outdoor unit 4 is installed on the rooftop or the like, and the indoor unit 5 is installed inside the ceiling of the room on each floor. Each outdoor unit 4 and each indoor unit 5 has a built-in controller (control unit), such as a microcomputer. The outdoor unit 4, indoor unit 5, and central monitoring device 8 are connected by an air conditioning control network 9 for communicating control command values ​​and other information from the controllers. For example, the communication standard used in the air conditioning control network 9 is a proprietary communication protocol unique to each air conditioner manufacturer.

[0013] The centralized monitoring device 8 includes a power consumption control unit 10, a power consumption estimation unit 30 that estimates the annual power consumption, an operation control unit 40 that controls the air conditioner 3 so as not to exceed the power consumption upper limit generated by the power consumption control unit 10, and a communication unit 50 for communicating with the outside world. The communication unit 50 has functions for communicating with the air conditioner 3 via the air conditioning control network 9 and for communicating with external servers via the internet. The centralized monitoring device 8 is composed of a computer.

[0014] (Configuration of the power consumption estimation unit) Figure 2 is a diagram showing the schematic configuration of the power consumption estimation unit according to the first embodiment. The power consumption estimation unit 30 comprises a data acquisition unit 31, a learning unit 32, an estimation unit 33, and a storage unit 34. The data acquisition unit 31 acquires data necessary for estimating the annual power consumption. For example, the data acquisition unit 31 acquires actual power consumption data as illustrated in Figure 3. The actual data records the amount of power consumed by operating the air conditioner 3 installed in building 2, associated with the date, time of day, holiday status, temperature, etc. The date, time of day, holiday status, temperature, etc., are parameters related to the amount of power consumed by the air conditioner. These parameters may also include other items such as humidity and weather. The data acquisition unit 31 records the acquired actual data in the storage unit 34.

[0015] The learning unit 32 uses the actual data acquired by the data acquisition unit 31 as training data, performs learning using machine learning, etc., and constructs a predictive model that predicts the amount of electricity required for air conditioning. An example of a predictive model is shown in Figure 4. For example, the learning unit 32 constructs a predictive model 35 with the month and the average monthly temperature (maximum and minimum temperatures may also be added) as explanatory variables, and the monthly electricity consumption as the dependent variable. Alternatively, for example, the learning unit 32 constructs a predictive model 35 with the daily temperature, day of the week, holiday category, and the month or season to which the target day belongs as explanatory variables, and the daily electricity consumption as the dependent variable. Or, for example, the learning unit 32 constructs a predictive model 35 with the temperature for each time period (for example, the average temperature for that time period), day of the week, holiday category, and the month or season to which the target time period belongs as explanatory variables, and the electricity consumption for that time period as the dependent variable.

[0016] The estimation unit 33 predicts the monthly, daily, or hourly energy consumption required for air conditioning based on the prediction model 35 and the predicted values ​​of the explanatory variables, and estimates the annual energy consumption by aggregating the predicted energy consumption. The estimation unit 33 also calculates the monthly energy consumption predicted based on the prediction model 35 and an apportionment coefficient for calculating the monthly energy consumption from the annual energy consumption. The estimation unit 33 uses the daily energy consumption predicted based on the prediction model 35 to calculate an apportionment coefficient for calculating the daily energy consumption from the monthly energy consumption. The estimation unit 33 uses the hourly energy consumption predicted based on the prediction model 35 to calculate an apportionment coefficient for calculating the hourly energy consumption from the daily energy consumption.

[0017] The memory unit 34 stores various types of information. For example, the memory unit 34 stores actual data acquired by the data acquisition unit 31, the prediction model 35 constructed by the learning unit 32, and so on.

[0018] Next, we will explain the process of estimating annual power consumption when the learning unit 32 constructs a predictive model 35 that predicts monthly power consumption. Figure 5 is a first flowchart showing an example of the annual power consumption estimation process according to the first embodiment. The estimation unit 33 predicts the monthly power consumption (step S101). The user inputs the predicted average temperature for each month from January to December and the month to be predicted into the power consumption estimation unit 30 and instructs it to predict the power consumption. For example, in the power consumption estimation unit 30, the estimation unit 33 inputs the average temperature for January and the month of January to be predicted into the prediction model 35. The prediction model 35 outputs a predicted value for the power consumption of the air conditioner 3 for January. Similarly, the estimation unit 33 inputs the average temperature for each month from February to December into the prediction model 35 to obtain predicted values ​​for the power consumption of the air conditioner 3 for each month from February to December.

[0019] Next, the estimation unit 33 calculates the annual power consumption by summing up the predicted monthly power consumption values ​​(step S102). Next, the estimation unit 33 calculates the monthly apportionment coefficient (step S103). For example, the estimation unit 33 calculates the January apportionment coefficient by dividing the predicted power consumption for January calculated in step S101 by the estimated annual power consumption calculated in step S102. Similarly, the estimation unit 33 calculates the apportionment coefficients for February to December. The estimation unit 33 outputs the apportionment coefficients for each month to the power consumption control unit 10. The power consumption control unit 10 receives and stores the apportionment coefficients for each month. Next, the estimation unit 33 outputs the annual power consumption calculated in step S102 to a display device or electronic file, etc. (step S104). The user checks the outputted annual power consumption and sets the annual target power consumption based on this value (step S1 in Figure 16).

[0020] Next, we will explain the process of estimating the annual power consumption when the learning unit 32 constructs a predictive model 35 that predicts the daily power consumption. Figure 6 is a second flowchart showing an example of the annual power consumption estimation process according to the first embodiment. The estimation unit 33 predicts the daily power consumption (step S111). The user inputs the predicted temperature (e.g., average temperature), day of the week, holiday category, and month or season to which the prediction target day belongs to the power consumption estimation unit 30, and instructs it to predict the power consumption. In the power consumption estimation unit 30, the estimation unit 33 inputs the temperature, day of the week, holiday category, and month to which the prediction target day belongs for January 1 to the prediction model 35. The prediction model 35 outputs a predicted value for the power consumption of the air conditioner 3 on January 1. Similarly, the estimation unit 33 inputs the temperature, day of the week, holiday category, and month to which the prediction target day belongs for each day from January 2 to December 31 to the prediction model 35, thereby obtaining a predicted value for the daily power consumption of the air conditioner 3 for one year.

[0021] Next, the estimation unit 33 calculates the monthly power consumption by summing the predicted daily power consumption values ​​(step S112). For example, the estimation unit 33 calculates the power consumption for January by summing the predicted power consumption values ​​from January 1st to January 31st. Similarly, the estimation unit 33 calculates the power consumption for each month from February to December. Next, the estimation unit 33 calculates the annual power consumption by summing the predicted monthly power consumption values ​​calculated in step S112 (step S113).

[0022] Next, the estimation unit 33 calculates the monthly and daily allocation coefficients (step S114). For example, the estimation unit 33 calculates the January allocation coefficient by dividing the predicted value of the January power consumption calculated in step S112 by the estimated value of the annual power consumption calculated in step S113. Similarly, the estimation unit 33 calculates the allocation coefficients for February through December. The estimation unit 33 outputs the allocation coefficients for each month to the power consumption control unit 10. The power consumption control unit 10 receives and stores the allocation coefficients for each month.

[0023] Furthermore, the estimation unit 33 calculates the apportionment coefficient for January 1st by dividing the predicted power consumption for January 1st, calculated in step S111, by the estimated power consumption for January, calculated in step S112. Similarly, the estimation unit 33 calculates the apportionment coefficient for each day from January 2nd to 31st. The estimation unit 33 also calculates the apportionment coefficient for each day from February to December in the same manner. The estimation unit 33 outputs the apportionment coefficient for each day from January 1st to December 31st to the power consumption control unit 10. The power consumption control unit 10 receives and stores the apportionment coefficient for each day. Next, the estimation unit 33 outputs the annual power consumption calculated in step S113 to a display device or electronic file, etc. (step S115). The user checks the outputted annual power consumption and sets the annual target power consumption based on this value (step S1 in Figure 16).

[0024] Next, we will explain the process of estimating annual power consumption when the learning unit 32 constructs a predictive model 35 that predicts power consumption for each time period. Figure 7 is a third flowchart showing an example of the annual power consumption estimation process according to the first embodiment. The estimation unit 33 predicts the amount of power consumed for each time period (step S121). The user inputs the predicted temperature for each time period, day of the week, holiday category, month to which the predicted time period belongs, etc., for each day from January 1 to December 31 into the power consumption estimation unit 30 and instructs it to predict the amount of power consumed. For example, if the time period is set to 1 hour, the user inputs the predicted temperature for each hour from January 1 to December 31. In the power consumption estimation unit 30, the estimation unit 33 inputs the temperature for each time period on January 1, day of the week, holiday category, month to which the predicted time period belongs, etc., into the prediction model 35. The prediction model 35 outputs the predicted amount of power consumed by the air conditioner 3 for each time period on January 1. Similarly, the estimation unit 33 inputs the temperature for each time period on each day from January 2 to December 31, day of the week, holiday category, month to which the predicted time period belongs, etc., into the prediction model 35, thereby obtaining a year's worth of predicted power consumption values ​​for each time period by the air conditioner 3.

[0025] Next, the estimation unit 33 predicts the daily power consumption (step S122). The user aggregates the power consumption for each time period predicted in step S121 for each day to calculate the predicted power consumption of the air conditioner 3 for each day from January 1 to December 31. Next, the estimation unit 33 calculates the monthly power consumption by summing the predicted daily power consumption values ​​(step S123). For example, the estimation unit 33 calculates the power consumption for January by summing the predicted power consumption values ​​for January 1 to January 31 calculated in step S122. Similarly, the estimation unit 33 calculates the power consumption for each month from February to December. Next, the estimation unit 33 calculates the annual power consumption by summing the predicted monthly power consumption values ​​calculated in step S123 (step S124).

[0026] Next, the estimation unit 33 calculates the apportionment coefficients for each month, day, and time period (step S125). For example, the estimation unit 33 calculates the apportionment coefficient for January by dividing the predicted value of the January power consumption calculated in step S123 by the estimated value of the annual power consumption calculated in step S124. Similarly, the estimation unit 33 calculates the apportionment coefficients for February through December. The estimation unit 33 outputs the apportionment coefficients for each month to the power consumption control unit 10. The power consumption control unit 10 receives and stores the apportionment coefficients for each month.

[0027] Furthermore, the estimation unit 33 calculates the apportionment coefficient for January 1st by dividing the predicted power consumption for January 1st, calculated in step S122, by the estimated power consumption for January, calculated in step S123. Similarly, the estimation unit 33 calculates the apportionment coefficient for each day from January 2nd to December 31st. The estimation unit 33 outputs the apportionment coefficient for each day from January 1st to December 31st to the power consumption control unit 10. The power consumption control unit 10 receives and stores the apportionment coefficient for each day.

[0028] Furthermore, the estimation unit 33 calculates the apportionment coefficient for each time period on January 1st by dividing the predicted power consumption values ​​for each time period on January 1st, calculated in step S121, by the estimated power consumption values ​​for January 1st, calculated in step S122. Similarly, the estimation unit 33 calculates the apportionment coefficient for each time period on each day from January 2nd to December 31st. The estimation unit 33 outputs the apportionment coefficient for each time period on each day from January 1st to December 31st to the power consumption control unit 10. The power consumption control unit 10 receives and stores the apportionment coefficient for each time period.

[0029] Next, the estimation unit 33 outputs the annual power consumption calculated in step S124 to a display device or electronic file (step S126). The user checks the outputted annual power consumption and sets the annual target power consumption based on this value (step S1 in Figure 16).

[0030] (Configuration of the power consumption control unit) Figure 8 is a diagram showing the schematic configuration of the power consumption control unit according to the first embodiment. As shown in Figure 8, the power consumption control unit 10 includes an input information acquisition unit 11, a storage unit 12, a setting unit 13, a first determination unit 15, a second determination unit 16, a power upper limit adjustment unit 17, a correction unit 18, and a failure avoidance unit 20.

[0031] The input information acquisition unit 11 receives input from the user and the power consumption estimation unit 30. For example, the input information acquisition unit 11 receives the annual target power consumption entered by the user, and the selection information of the operating mode, specifying either the comfort-focused mode or the economy-focused mode. The input information acquisition unit 11 also receives monthly, daily, and time-of-day apportionment coefficients entered by the power consumption estimation unit 30. The user enters the annual target power consumption, referring to the annual power consumption presented in the processes shown in Figures 5 to 7. The apportionment coefficients may also be entered by the user. If the power consumption estimation unit 30 calculates monthly apportionment coefficients, the user enters the monthly apportionment coefficients. If the power consumption estimation unit 30 calculates monthly and daily apportionment coefficients, the user enters the monthly and daily apportionment coefficients. Alternatively, if the power consumption estimation unit 30 calculates monthly, daily, and time-of-day apportionment coefficients, the user inputs these coefficients. The user may input this information using an input unit such as a keyboard provided on the central monitoring device 8, or they may input it from an input unit installed in a remote location to the input information acquisition unit 11 via the network and communication unit 50.

[0032] The memory unit 12 stores various data that the setting unit 13 (described later) references when setting monthly, daily, and hourly target power consumption amounts, as well as the monthly, daily, and hourly target power consumption amounts and power consumption upper limits set by the setting unit 13. For example, as shown in Figure 9, the memory unit 12 stores the apportionment coefficients ax (x=1 to 12) set for each month. The apportionment coefficients a1 to a12 are set so that their sum equals 1 (a1 + a2 + ... + a12 = 1). The apportionment coefficients a1 to a12 are calculated in advance by the power consumption estimation unit 30.

[0033] The memory unit 12 stores cooling and heating evaluation coefficient characteristics, which are used to set daily or hourly target power consumption amounts, as illustrated in Figures 10(a) and (b). These are used when the prediction model 35 is constructed to predict monthly power consumption amounts. The cooling performance coefficient characteristics shown in Figure 10(a) are an example of a function derived from, for example, the relationship between past maximum temperatures and daily power consumption, where the higher the maximum temperature, the larger the cooling performance coefficient α_cool is set to. The heating evaluation coefficient characteristics shown in Figure 10(b) are an example of a function derived from, for example, the relationship between the lowest temperature in the past and the amount of electricity consumed in a day, where the lower the lowest temperature, the larger the value of the heating evaluation coefficient α_heat is set to.

[0034] The cooling performance coefficient characteristics shown in Figure 10(a) and the heating performance coefficient characteristics shown in Figure 10(b) may be used to set the power consumption for each time period. Alternatively, a function derived from the relationship between past maximum and minimum temperatures and power consumption for each time period may be provided. The following explanation will use the functions shown in Figures 10(a) and 10(b) as an example.

[0035] Furthermore, the memory unit 12 stores calculation formulas for setting target power consumption amounts for each month, day, and time period using the various data mentioned above.

[0036] The setting unit 13 calculates monthly, daily, and hourly target power consumption based on various data stored in the memory unit 12 (coefficients a1 to a12, cooling evaluation coefficient α_cool, heating evaluation coefficient α_heat, etc.), calculation formulas, the annual target power consumption Porg_y acquired by the input information acquisition unit 11, and the monthly apportionment coefficients (or monthly and daily apportionment coefficients, or monthly, daily and time-of-day apportionment coefficients) calculated by the power consumption estimation unit 30. The procedure for calculating the target power consumption by the setting unit 13 will be described below.

[0037] [Regarding monthly target power consumption for Porg_m] The setting unit 13 calculates the target power consumption Porg_m for each month by multiplying the annual target power consumption Porg_y by the apportionment coefficients a1 to a12 (monthly apportionment coefficients calculated by the power consumption estimation unit 30) shown in Figure 9. Figure 11 shows an example of the target power consumption Porg_m for each month. Figure 11 is a diagram showing an example of the monthly target power consumption according to the first embodiment. As shown in the diagram, for example, a large amount of power consumption is allocated in July and August when power consumption due to cooling increases, and a small amount of power consumption is allocated in months when there is little need for heating or cooling.

[0038] [Regarding the daily target power consumption for Porg_d] (1) When the power consumption estimation unit 30 calculates the daily apportionment coefficient. The setting unit 13 calculates the target power consumption Porg_d for each day by multiplying the daily apportionment coefficient calculated by the power consumption estimation unit 30 by the target power consumption Porg_m for each month. For example, the target power consumption Porg_d for January 1st is calculated by multiplying the apportionment coefficient for January 1st by the target power consumption Porg_m for January.

[0039] (2) When the power consumption estimation unit 30 has not calculated the daily apportionment coefficient. The setting unit 13 calculates the daily target power consumption Porg_d using the annual daily predicted maximum temperature data and predicted minimum temperature data, the cooling evaluation coefficient characteristics and heating evaluation coefficient characteristics shown in Figure 10, and the monthly target power consumption Porg_m. Annual forecast maximum temperature data and forecast minimum temperature data can be obtained, for example, from a predetermined server on the internet via the communication unit 50. The setting unit 13 obtains the cooling evaluation coefficient α_cool corresponding to the predicted maximum temperature in the cooling evaluation coefficient characteristics shown in Figure 10(a) for the cooling period (for example, from May to October). More specifically, it obtains daily predicted maximum temperature data from an external server, etc., and obtains the cooling evaluation coefficient α_cool for all days of the target month for the points on the horizontal axis of the cooling evaluation coefficient characteristics graph shown in Figure 10(a) that correspond to the obtained predicted maximum temperature.

[0040] Then, by substituting the acquired daily cooling performance coefficient α_cool(i), the target energy consumption for the month Porg_m, and the sum of the cooling performance coefficients for the month Σα_cool into the following equation (1), the daily target energy consumption for the month Porg_d is set.

[0041] Porg_d=Porg_m×(α_cool(i) / Σα_cool) ...(1)

[0042] In equation (1), (i) is the calculation date, Porg_m is the target power consumption for the month to which the calculation date belongs, α_cool(i) is the cooling performance coefficient for the calculation date, and Σα_cool is the sum of the cooling performance coefficients for the month to which the calculation date belongs.

[0043] Alternatively, the above-mentioned cooling performance coefficient α_cool may be corrected by multiplying it by a coefficient corresponding to the air conditioning operating rate in building 2, and the daily target power consumption Porg_d may be calculated using the corrected cooling performance coefficient α_cool. For example, in offices, the air conditioning utilization rate is lower on Saturdays, Sundays, and public holidays compared to weekdays. Therefore, in this case, the cooling evaluation coefficient α_cool for those days is corrected by multiplying it by a coefficient of less than 1 (for example, 1 / 4). Note that the daily apportionment coefficient predicted by prediction model 35 already takes into account whether it is a holiday or a weekday.

[0044] For the heating period (for example, from November to April), the setting unit 13 obtains a heating evaluation coefficient α_heat corresponding to the daily predicted minimum temperature obtained from an external server or the like, based on the heating evaluation coefficient characteristics shown in Figure 10(b).

[0045] Then, by substituting the acquired daily heating evaluation coefficient α_heat(i), the monthly target power consumption Porg_m, and the sum of the monthly cooling evaluation coefficients Σα_heat into the following equation (2), the daily target power consumption Porg_d for that month is set.

[0046] Porg_d=Porg_m×(α_heat(i) / Σα_heat) ...(2)

[0047] In equation (2), (i) is the calculation day, Porg_m is the target energy consumption for the month to which the calculation day belongs, α_heat(i) is the heating evaluation coefficient for the calculation day, and Σα_heat is the sum of the heating evaluation coefficients for the month to which the calculation day belongs.

[0048] Furthermore, for the heating period, similar to the cooling period, the heating evaluation coefficient α_heat may be corrected using the air conditioning utilization rate, and the daily target power consumption may be calculated using the corrected heating evaluation coefficient α_heat.

[0049] If the predicted temperature data described above cannot be obtained, the daily target power consumption Porg_d may be set by dividing the monthly target power consumption Porg_m by the number of days in that month. Alternatively, historical annual daily predicted maximum temperature data and predicted minimum temperature data may be used as substitutes. In this case, the utilization rate for each day of the week may also be taken into consideration.

[0050] [Regarding the target power consumption for each time period (Porg_h)] (1) When the power consumption estimation unit 30 calculates the apportionment coefficient for each time period. The setting unit 13 calculates the target power consumption Porg_h for each time period by multiplying the time-based apportionment coefficient calculated by the power consumption estimation unit 30 by the target power consumption Porg_d for each day. For example, the target power consumption Porg_h for January 1st from 0 to 1 am is calculated by multiplying the apportionment coefficient for January 1st from 0 to 1 am by the target power consumption Porg_d for January 1st.

[0051] (2) When the power consumption estimation unit 30 has not calculated the daily apportionment coefficient. The setting unit 13 calculates the target power consumption Porg_h for each time period using the annual hourly predicted maximum temperature data and predicted minimum temperature data, the cooling evaluation coefficient characteristics and heating evaluation coefficient characteristics shown in Figure 10, and the daily target power consumption Porg_d.

[0052] Annual forecast maximum temperature data and forecast minimum temperature data for each time period can be obtained, for example, from an external server via the communication unit 50. For the cooling period, the setting unit 13 obtains the cooling evaluation coefficient α_cool(k) corresponding to the expected maximum temperature for each time period in the cooling evaluation coefficient characteristics shown in Figure 10(a). For example, if the expected maximum temperature at 6:00 is X1℃, the setting unit 13 obtains the cooling evaluation coefficient α_cool corresponding to X1℃ as the cooling evaluation coefficient α_cool for 6:00. Similarly, the setting unit 13 obtains 24 cooling evaluation coefficients α_cool(k) for 24 hours. k is an integer from 1 to 24.

[0053] Here, we will explain the estimation accuracy when the hourly target power consumption is determined using equations (1') and (2') below, which are similar to equations (1) and (2) used to calculate the daily target power consumption Porg_d. (Cooling season) Porg_h´=Porg_d×(α_cool(k) / Σα_cool) ...(1') (Heating season) Porg_h´=Porg_d×(α_heat(k) / Σα_heat) ...(2')

[0054] Figure 12 is the first diagram illustrating the calculation method for hourly target power consumption according to the first embodiment. Figure 12 shows the 24-hour value of Porg_h'(j) calculated using equation (1') (target power consumption) and the actual power consumption when the air conditioner 3 is operated at the same set temperature. In the figure, points plotted with squares represent the target power consumption, and points plotted with diamonds represent the actual power consumption. Referring to the graph in Figure 12, the actual power consumption significantly exceeds the target power consumption from 6:00 to 9:00. In this example, the air conditioner 3 was not operating before 6:00, and the temperature of the entire building 2 was high. As a result, because the amount of heat contained in building 2 was large, the power consumption after operation started significantly exceeded the target power consumption. If the air conditioner 3 had been operating until dawn, it is thought that the actual power consumption would have been closer to the target power consumption.

[0055] Similarly, at 6:00 AM when heating is started in winter, the temperature of the entire building tends to be low. Therefore, the target power consumption calculated using equation (2') above based on the predicted minimum temperature at 6:00 AM may be lower than the actual power consumption at the same time. Thus, the target power consumption calculated based on predicted temperature may deviate from the actual power consumption. Therefore, in this embodiment, in addition to the predicted temperature for each time period, the target power consumption for each time period is calculated by considering the assumed air conditioning load conditions for each time period (for example, the amount of heat the building possesses). For example, in summer, the temperature of the entire building warms up, and in winter, the morning hours when the temperature of the entire building tends to cool down tend to be lower, so a larger target power consumption is allocated to each hour.

[0056] Figure 13 is a second diagram illustrating the calculation method for hourly target power consumption according to the first embodiment. Figure 13 shows an example of air conditioning load coefficients that can be set for each time period. For example, the air conditioning load coefficients b1, b2, and b3 for the cooling period are set for 6:00-9:00, 9:00-24:00, and 24:00-5:00, respectively, and values ​​are set such that, for example, b1>b2>b3. Similarly, the air conditioning load coefficients b4, b5, and b6 for the heating period are set for 6:00-9:00, 9:00-24:00, and 24:00-5:00, respectively, and values ​​are set such that, for example, b4>b6>b5.

[0057] Here, the air conditioning load coefficients b1 to b6 may be determined according to the building's usage. For example, if Building 2 is an office building, the air conditioning is turned off at night. Therefore, relatively large values ​​are set for the air conditioning load coefficients b1 and b4 during the morning hours. For example, if Building 2 is a hospital or a 24-hour store, the air conditioning is running even at night. Therefore, relatively small values ​​are set for the air conditioning load coefficients b1 and b4 during the morning hours. By setting them in this way, a large amount of power can be allocated to the time of day when Building 2 has a large amount of heat, and the air conditioner 3 can be operated without compromising the comfort of the air conditioning. In other examples, the air conditioning load coefficients b1, etc. may be set according to the location environment of Building 2 and the structural characteristics of the building. For example, if Building 2 is an office building located in an environment where buildings are densely packed and heat tends to accumulate, or if it has characteristics that do not release heat due to the number of windows or the characteristics of the wall materials, an even larger value may be set for the air conditioning load coefficient b1. Furthermore, if Building 2 is located in an area with severe morning chills or is a building with poor insulation, a larger value may be set for the air conditioning load coefficient b4. Alternatively, in buildings with many heat-generating devices, such as data centers with numerous computers, a relatively large value may also be set for the nighttime air conditioning load coefficient b6.

[0058] Note that the air conditioning load coefficients b1 to b6 and the time periods corresponding to each coefficient shown in Figure 13 are examples only. For example, the air conditioning load coefficient b1, etc., could be set every hour, or the air conditioning load coefficient b1, etc., could be set for two time periods: the morning time period (6:00 to 9:00) and other time periods. Also, in the example in Figure 13, the morning time period is set to 6:00 to 9:00, but other time periods (for example, 4:00 to 7:00) could be set according to the sunrise time, which varies depending on the season and region (latitude and longitude). Note that the setting table of air conditioning load coefficients shown in Figure 13 is stored in the storage unit 12.

[0059] Next, the specific calculation method for hourly target power consumption will be explained. First, the setting unit 13 obtains α_cool(k) based on the cooling evaluation coefficient characteristics and the maximum predicted temperature for each hour shown in Figure 10(a). Next, the setting unit 13 obtains the air conditioning load coefficient β_cool(k) based on the setting table shown in Figure 13. Then, the setting unit 13 sets the hourly target power consumption Porg_h for the day by substituting the hourly cooling evaluation coefficient α_cool(k), the hourly air conditioning load coefficient β_cool(k), the target power consumption Porg_d for the day, and the sum of the values ​​obtained by multiplying the hourly cooling evaluation coefficient by the air conditioning load coefficient Σ(α_cool×β_cool) into the following equation (3).

[0060] Porg_h=Porg_d×(α_cool(k)×β_cool(k)) / Σ(α_cool×β_cool) ···(3)

[0061] In equation (3), (k) is the calculation time, and Porg_d is the target power consumption for the day to which the calculation time belongs.

[0062] Similarly, for the heating period, the setting unit 13 obtains a heating evaluation coefficient α_heat(k) corresponding to the hourly predicted minimum temperature obtained from an external server or the like, based on the heating evaluation coefficient characteristics shown in Figure 10(b). The setting unit 13 also obtains an air conditioning load coefficient β_heat(k) based on the setting table shown in Figure 13.

[0063] Then, by substituting the acquired hourly heating evaluation coefficient α_heat(k), hourly air conditioning load coefficient β_heat(k), the target power consumption for the day Porg_d, and the sum of the values ​​obtained by multiplying the hourly heating evaluation coefficient by the air conditioning load coefficient Σ(α_heat×β_heat) into the following equation (4), the hourly target power consumption Porg_h for the day is set.

[0064] Porg_h=Porg_d×(α_heat(k)×β_heat(k)) / Σ(α_heat×β_heat) ···(4)

[0065] [Regarding the target power consumption every 30 minutes for Porg_j] The setting unit 13 divides the target power consumption Porg_h for each time period and sets the target power consumption Porg_j for every 30 minutes. For example, if the time period is 1 hour, the setting unit 13 divides the target power consumption Porg_h for each hour into two and sets the target power consumption Porg_j for every 30 minutes.

[0066] [Regarding setting the maximum power consumption limit] The setting unit 13 sets the power consumption limit per sampling period (instantaneous power consumption limit) by, for example, dividing the target power consumption Porg_j every 30 minutes by the sampling period of the air conditioning control. The power consumption limit is set so that when the air conditioning is operated at this power consumption limit, the power consumption over a 30-minute period will be less than or equal to the target power consumption associated with that 30-minute period.

[0067] The setting unit 13 stores the monthly target power consumption Porg_m[Wh], the daily target power consumption Porg_d[Wh], the time-of-day target power consumption Porg_h[Wh], the 30-minute target power consumption Porg_j[Wh], and the power consumption upper limit Pt[W] set at 30-minute intervals (in 30-minute units) in the storage unit 12.

[0068] The first determination unit 15 determines, during the cooling period, at a predetermined determination cycle, whether the value obtained by subtracting the set temperature Ts from the indoor temperature Ta (hereinafter referred to as "temperature difference ΔT_cool") is greater than or equal to the cooling deviation threshold Tt_cool. Here, the indoor temperature Ta is, for example, the indoor intake temperature of the indoor unit 5.

[0069] The above determination period is set to the period of the minimum interval of power consumption set by the setting unit 13, that is, a time shorter than 30 minutes (for example, 5 minutes). In this embodiment, the determination period is set to 5 minutes, but it is not limited to this example, and may be 10 minutes, 15 minutes, for example. The set temperature Ts is, for example, a value obtained from the air conditioning schedule if the central monitoring device 8 controls the operation of the air conditioner 3 based on the air conditioning schedule. If such operation control based on an air conditioning schedule is not performed, the set temperature Ts may be a value set by the user in the remote controller, for example. Thus, the method of obtaining the set temperature Ts is not limited.

[0070] The cooling deviation threshold Tt_cool is initially set to 3°C. This value is predetermined based on, for example, the temperature difference between the indoor temperature at which a user begins to feel uncomfortable and the set temperature. Furthermore, this cooling deviation threshold Tt_cool can be changed by the economy adjustment unit 22, which will be described later.

[0071] The second determination unit 16 determines, during the heating period, at a predetermined determination cycle, whether the value obtained by subtracting the indoor temperature Ta from the set temperature Ts (hereinafter referred to as "temperature difference ΔT_heat") is greater than or equal to the heating deviation threshold Tt_heat. The determination cycle and the set temperature Ts are the same as those of the first determination unit 15 described above. The heating deviation threshold Tt_heat is initially set to, for example, 3°C. This heating deviation threshold Tt_heat can be changed by the economy adjustment unit 22, which will be described later. Furthermore, the heating deviation threshold Tt_heat and the cooling deviation threshold Tt_cool may be set to different values.

[0072] The power limit adjustment unit 17 increases the power consumption limit Pt for the next determination cycle (for example, the next 5 minutes) by a predetermined amount if the first determination unit 15 determines that the temperature difference ΔT_cool is greater than or equal to the cooling deviation threshold Tt_cool during the cooling period. Similarly, if the second determination unit 16 determines that the temperature difference ΔT_heat is greater than or equal to the heating deviation threshold Tt_heat during the heating period, the power limit adjustment unit 17 increases the power consumption limit Pt for the next determination cycle (for example, the next 5 minutes) by a predetermined amount.

[0073] Next, the power consumption control process of this embodiment will be explained using Figure 14. Figure 14 shows an example of the relationship between the set temperature Ts, the room temperature Ta, the actual power consumption Pr, and the power consumption upper limit Pt from 9:50 to 11:05 on a day during the cooling period. For example, at 10:30 and 10:35, the temperature difference ΔT_cool between the set temperature Ts and the room temperature Ta is greater than or equal to the cooling deviation threshold Tt_cool. Therefore, the power consumption upper limit Pt at 10:30 and 10:35 is increased by a predetermined amount. The white circles in Figure 14 represent the power consumption upper limit Pt set based on the target power consumption Porg_j every 30 minutes, and indicate the power consumption upper limit Pt before the power increase by the power upper limit adjustment unit 17.

[0074] The modification section 18 includes a first modification section 25, a second modification section 26, a third modification section 27, and a fourth modification section 28. The first modification unit 25 calculates the difference between the actual power consumption [Wh] for the first 30 minutes of an hour and the target power consumption Porg_j [Wh], and modifies the target power consumption Porg_j [Wh] by adding this difference to the target power consumption for the latter 30 minutes of that hour.

[0075] The second modification unit 26 compares the actual power consumption [Wh] over the past hour with the target power consumption Porg_h [Wh], and modifies the target power consumption Porg_h [Wh] for each hour of the same day by adding the difference equally to the target power consumption Porg_h [Wh] for each subsequent hour of the same day.

[0076] The third modification unit 27 compares the actual power consumption [Wh] for the past day with the target power consumption Porg_d [Wh], and modifies the daily target power consumption Porg_d [Wh] for the same month by equally dividing the difference and adding it to the target power consumption Porg_d [Wh] for the following days in the same month.

[0077] The fourth modification unit 28 compares the actual power consumption for the past month with the target power consumption Porg_m[Wh], and modifies the monthly target power consumption Porg_m[Wh] for the same year by equally dividing the difference and adding it to the target power consumption Porg_m[Wh] for the following months in the same year.

[0078] The failure avoidance unit 20 includes a mode determination unit 21, an economy adjustment unit 22, and a comfort adjustment unit 23. For example, if a user sets a strict annual power consumption target, or if actual power consumption increases significantly compared to previous years due to unexpected weather anomalies, the actual power consumption may significantly exceed the monthly target power consumption Porg_m[Wh] set by the setting unit 13. In such cases, the correction unit 18, for example, the fourth correction unit 28, corrects the monthly target power consumption Porg_m[Wh], which can then be heavily burdened on the target power consumption Porg_m[Wh] for subsequent months, potentially causing the power consumption control to fail. The failure avoidance unit 20 makes adjustments to prevent the power consumption control from failing in such cases. This will be explained in detail below.

[0079] First, the mode determination unit 21 determines the driving mode selected by the user, that is, whether the economy-focused mode or the comfort-focused mode has been selected.

[0080] [Regarding the economy-focused mode] When the ratio (ΣPorg_m´ / ΣPorg_m) of the sum ΣPorg_m´ [Wh] of the target power consumption amounts Porg_m´ [Wh] for each month after the next month of this year after being corrected by the fourth correction part 28 to the sum ΣPorg_m [Wh] of the target power consumption amounts Porg_m [Wh] for each month after the next month of this year before being corrected by the fourth correction part 28 is less than or equal to a predetermined economic standard value Ke (0 < Ke < 1), the cooling deviation threshold value Tt_cool and the heating deviation threshold value Tt_heat are increased. As an example, these are adjusted using the following formulas (5) and (6).

[0081] Tt_cool´ = Tt_cool × c (5) Tt_heat´ = Tt_heat × d (6)

[0082] In formulas (5) and (6), Tt_cool´ and Tt_heat´ are the values after the change, and the coefficients c and d are coefficients set to values greater than 1. For example, c and d are set to values that are inversely proportional to the ratio (ΣPorg_m´ / ΣPorg_m), and the smaller the ratio, the larger the values of c and d are set. For example, a function or table having such characteristics is prepared in advance, and using this function or table, the cooling deviation threshold value Tt_cool and the heating deviation threshold value Tt_heat are adjusted using the coefficients c and d corresponding to the ratio.

[0083] For example, when the economic standard value Ke is 1 / 3 and ΣPorg_m´ / ΣPorg_m is 1 / 3 or less, it is assumed that the cooling deviation threshold value is increased by 2°C. When the economic standard value Ke is 2 / 3 and ΣPorg_m´ / ΣPorg_m is less than 2 / 3 and greater than or equal to 1 / 3, it is assumed that the cooling deviation threshold value is increased by 1°C, and the cooling deviation threshold value is adjusted according to the economic standard value. Since the cooling deviation threshold value is often about 3°C, in the above example, the cooling deviation threshold value Tt_cool changes between 3°C and 5°C.

[0084] In this manner, when the cooling deviation threshold Tt_cool and the heating deviation threshold Tt_heat are adjusted, the first determination unit 15 and the second determination unit 16 perform the above determination using the adjusted cooling deviation threshold Tt_cool' and the adjusted heating deviation threshold Tt_cool'.

[0085] In economy-prioritizing mode, the target power consumption cannot be increased from an economy-prioritizing perspective, so the economy adjustment unit 22 adjusts in a direction that increases the cooling (heating) deviation threshold. This makes it possible to ensure economy even at the expense of some comfort.

[0086] [About the comfort-focused mode] The comfort adjustment unit 23 multiplies the target power consumption Porg_m for each month from this month onward by γ (γ>1) if the ratio (ΣPr_m / ΣPorg_m) of the total actual monthly power consumption ΣPr_m for the current year up to the previous month to the total monthly target power consumption ΣPorg_m for the current year up to the previous month is equal to or greater than a predetermined comfort standard value Kc (Kc>1). For example, if the comfort standard value Kc is 1.2 and (ΣPr_m / ΣPorg_m) is 1.2 or higher, then γ is set to 1.1, and the target power consumption amount Porg_m for this month and beyond is multiplied by 1.1 to relax the target power consumption amount.

[0087] In the comfort-first mode, an upper limit can be set on the number of times the target power consumption Porg_m for each month is adjusted. If this limit is exceeded, even if the above conditions are met, no further adjustments to the target power consumption Porg_m may be made. For example, you could set the upper limit to two adjustments, and not adjust the target power consumption Porg_m from the third time onward. Thus, in the comfort-prioritizing mode, the cooling (heating) deviation threshold cannot be increased from the perspective of prioritizing comfort, so the comfort adjustment unit 23 increases the target power consumption. This relaxes the constraint on the target power consumption, avoids control failure, and ensures comfort.

[0088] Next, the procedure for power consumption control processing performed by the power consumption control unit 10 having the above-described configuration will be explained with reference to Figures 15 to 18. As shown in Figure 15, the power consumption control process includes an initial process and a loop process. The initial process is executed, for example, when the user inputs an annual target power consumption or when the input annual target power consumption is changed, and is mainly performed by the setting unit 13. The initial process will be described below with reference to Figure 16. Figure 16 is a flowchart of an example of the initial process according to the first embodiment.

[0089] First, the input information acquisition unit 11 receives the annual target power consumption Porg_y (step S1) and stores the received annual target power consumption Porg_y in the storage unit 12 (step S2). The annual target power consumption Porg_y is set by the user, but the user sets the annual target power consumption Porg_y by referring to the annual power consumption estimated by the power consumption estimation unit 30. The annual power consumption estimated by the power consumption estimation unit 30 is an estimated value based on actual data, so it is unlikely to be far off and is considered to be a highly accurate estimate. By setting the annual target power consumption Porg_y by referring to the annual power consumption estimated by the power consumption estimation unit 30, a highly accurate power consumption upper limit Pt can be calculated by the following process.

[0090] Next, the setting unit 13 sets the monthly target power consumption Porg_m by multiplying the annual target power consumption Porg_y by a monthly apportionment coefficient (step S3). Next, the setting unit 13 sets the daily target power consumption Porg_d by multiplying the target power consumption Porg_m by a daily apportionment coefficient, etc. (step S4). Next, the setting unit 13 sets the time-of-day target power consumption Porg_h using the daily target power consumption Porg_d, etc. (step S5). As described above, the setting unit 13 calculates the time-of-day target power consumption Porg_h by multiplying the target power consumption Porg_d by a time-of-day apportionment coefficient. Alternatively, the setting unit 13 calculates the hourly target power consumption Porg_h based on the time-of-day air conditioning load conditions β_cool or β_heat and the hourly predicted temperature. Next, the setting unit 13 sets the 30-minute target power consumption Porg_j using the time-of-day target power consumption Porg_h (step S6). For example, if the time period is 2 hours long, the setting unit 13 divides the target power consumption Porg_h into 4 equal parts and sets the target power consumption Porg_j for each 30-minute interval. If the time period is 2 hours long, the setting unit 13 divides the target power consumption Porg_h into 6 equal parts and sets the target power consumption Porg_j for each 30-minute interval.

[0091] Furthermore, the setting unit 13 sets the power consumption upper limit Pt[W] in 30-minute increments based on the target power consumption Porg_j every 30 minutes (step S7). The setting unit 13 stores the monthly target power consumption Porg_m[Wh], the daily target power consumption Porg_d[Wh], the hourly target power consumption Porg_h[Wh], the 30-minute target power consumption Porg_j[Wh], and the power consumption upper limit Pt[W] set in 30-minute increments, which have been set in this manner, in the storage unit 12 (step S8).

[0092] Once the target power consumption and power consumption upper limit Pt[W] are stored in the memory unit 12, the air conditioner 3 is monitored and controlled based on the latest target power consumption and other information stored in the memory unit 12. Specifically, the operation control unit 40 acquires the power consumption upper limit Pt for 30-minute intervals set by the power consumption control unit 10, and controls the air conditioner 3 so that the instantaneous power consumption is less than or equal to the acquired power consumption upper limit Pt.

[0093] Next, we will explain the loop processing in the power consumption control process. In the loop processing, as shown in Figure 17, the power limit adjustment unit 17 performs power limit adjustment processing at 5-minute intervals. The first correction unit 25 performs the first correction processing at the end of the first 30 minutes of each hour (for example, at 0:30, 1:30, etc.), and the second correction unit 26 performs the second correction processing at 1-hour intervals (for example, at 55 minutes past the hour). In addition, the third correction unit 27 performs the third correction processing at 1-day intervals (for example, at 23:55 every day), and the fourth correction unit 28 performs the fourth correction processing at 1-month intervals (for example, at 23:55 on the last day of each month). Furthermore, once the fourth correction processing is completed, the failure avoidance unit 20 performs failure avoidance processing. The following describes each process in detail using the cooling season as an example.

[0094] In the power limit adjustment process, during the cooling period, the first determination unit 15 calculates the temperature difference ΔT_cool between the indoor temperature Ta and the set temperature Ts at a predetermined determination cycle (5-minute intervals). If the temperature difference ΔT_cool is greater than or equal to the cooling deviation threshold Tt_cool, the power limit adjustment unit 17 increases the power consumption limit Pt[W] for the next determination cycle (5-minute intervals) by a predetermined amount (see times 10:30 and 10:35 in Figure 14).

[0095] In the first correction process, the first correction unit 25 calculates the difference between the actual power consumption ΣPr[Wh] for the first 30 minutes and the target power consumption Porg_j for those 30 minutes. The first correction unit 25 adds this difference to the target power consumption Porg_j for the latter 30 minutes of that time. For example, if the actual power consumption P_j[Wh] for the first 30 minutes of a given hour is greater than the target power consumption Porg_j for that time, the target power consumption Porg_j for the latter 30 minutes is reduced by that difference, and the power consumption upper limit Pr[W] for the latter 30 minutes is reset based on the modified target power consumption Porg_j.

[0096] For example, in the example in Figure 14, the first modification unit 25 calculates the actual power consumption Pr_j from 9:55 to 10:25, and calculates the difference between this actual power consumption Pr_j and the target power consumption Porg_j for the 30 minutes from 9:55 to 10:25. The first modification unit 25 then adds this difference to the target power consumption Porg_j for the 30 minutes from 10:25 to 10:55. In the example in Figure 14, the actual power consumption Pr_j for the 30 minutes from 9:55 to 10:25 is greater than the target power consumption Porg_j, so the difference (hatched area in Figure 14) is reflected in the target power consumption Porg_j for the 30 minutes from 10:25 to 10:55, and the target power consumption Porg_j is modified to a value smaller than the original value. In Figure 14, the power consumption upper limit value based on the target power consumption after modification by the first modification process is represented by a white circle.

[0097] In the second correction process, the second correction unit 26 calculates the difference between the target power consumption Porg_h and the actual power consumption Pr_h[Wh] for the past hour (the length of the time period is assumed to be one hour), adds this difference equally to the target power consumption Porg_h for each subsequent hour on the same day, and stores the corrected target power consumption Porg_h in the storage unit 12. As a result, if the actual power consumption Pr_h[Wh] for the past hour is less than the target power consumption Porg_h, the target power consumption Porg_h for each subsequent hour on the same day will increase. Conversely, if the actual power consumption Pr_h[Wh] for the past hour exceeds the target power consumption Porg_h, the target power consumption Porg_h for each subsequent hour on the same day will decrease.

[0098] In the third correction process, the third correction unit 27 calculates the difference between the target power consumption Porg_d for the past day and the actual power consumption P_d[Wh], adds this difference equally to the daily target power consumption Porg_d for the same month thereafter, and stores the corrected daily target power consumption Porg_d in the storage unit 12. As a result, if the actual power consumption P_d[Wh] for the past day is less than the target power consumption Porg_d for that day, the daily target power consumption Porg_d for the same month thereafter will increase. Conversely, if it exceeds the target power consumption Porg_d, the daily target power consumption Porg_d for the same month thereafter will decrease.

[0099] In the fourth correction process, the fourth correction unit 28 calculates the difference between the target power consumption Porg_m for the past month and the actual power consumption P_m[Wh], adds this difference equally to the monthly target power consumption Porg_m for the same year thereafter, and stores the corrected monthly target power consumption Porg_m in the storage unit 12. As a result, if the actual power consumption P_m[Wh] for the past month is less than the target power consumption Porg_m for that month, the monthly target power consumption Porg_m for the same year thereafter will increase. Conversely, if it exceeds the target power consumption Porg_m, the monthly target power consumption Porg_m for the same year thereafter will decrease.

[0100] In the breakdown avoidance process, as shown in FIG. 18, the breakdown avoidance unit 20 determines whether the user has selected the economy - priority mode (step S11). When the economy - priority mode is selected (step S11; YES), the breakdown avoidance unit 20 determines whether the avoidance processing conditions for the economy - priority mode are satisfied. Specifically, the breakdown avoidance unit 20 determines whether the ratio (ΣPorg_m´ / ΣPorg_m) of the sum ΣPorg_m´[Wh] of the target power consumption amounts Porg_m´[Wh] for each month after the next month in the current year after being corrected by the fourth correction unit 28 to the sum ΣPorg_m[Wh] of the target power consumption amounts Porg_m[Wh] for each month after the next month in the current year immediately before being corrected by the fourth correction unit 28 is less than or equal to a predetermined economy criterion value Ke (0 < Ke < 1) (step S12).

[0101] When the above ratio is less than or equal to the economy criterion value Ke (step S12; YES), the breakdown avoidance unit 20 changes the cooling temperature deviation ΔTt_cool and the heating temperature deviation ΔTt_heat according to the ratio (step S13), and stores the changed values in the storage unit 12. The stored values are used in subsequent processing.

[0102] On the other hand, in step S11, when the comfort - priority mode is selected (step S11; NO), the breakdown avoidance unit 20 determines whether the avoidance processing conditions for the comfort - priority mode are satisfied. Specifically, the breakdown avoidance unit 20 determines whether the ratio (ΣPr_m / ΣPorg_m) of the sum ΣPr_m of the actual power consumption amounts for each month up to the previous month in the current year to the sum ΣPorg_m of the target power consumption amounts for each month up to the previous month in the current year is greater than or equal to a comfort criterion value Kc (Kc > 1) (step S14).

[0103] When the above ratio is greater than or equal to the comfort criterion value Kc (step S14; YES), the breakdown avoidance unit 20 determines whether the number of correction times of the target power consumption amount has reached the upper limit value (step S15). If it has not reached the upper limit value (step S15; YES), the breakdown avoidance unit 20 multiplies the latest target power consumption amount Porg_m for each month stored in the storage unit 12 by γ (γ > 1) (step S16).

[0104] If the conditions for avoiding the economy-first mode are not met in step S12 (step S12; NO), if the conditions for avoiding the comfort-first mode are not met in step S14 (step S14; NO), or if the number of corrections has already reached the upper limit in step S15 (step S15; NO), the process ends there.

[0105] As explained above, according to the power consumption control unit 10 of this embodiment, the annual target power consumption Porg_y[Wh] set based on actual data and the apportionment coefficient set based on actual data are used to set monthly, daily, hourly (every 1 hour), and 30-minute target power consumption Porg_m[Wh], Porg_d[Wh], Porg_h[Wh], and Porg_j[Wh]. The power consumption upper limit Pt[W] is then set from the 30-minute target power consumption Porg_j[Wh]. Control is then performed so that the instantaneous power does not exceed this power consumption upper limit Pt[W]. At the same time, the difference between the set temperature Ts and the room temperature Ta is calculated at a predetermined judgment cycle (for example, 5 minutes), and if these differences are greater than or equal to a predetermined temperature deviation threshold, the power consumption upper limit for the next judgment cycle is adjusted to increase.

[0106] In this way, by adjusting the power consumption limit according to the relationship between the set temperature and the room temperature, it is possible to control power consumption to a target level while maintaining a certain degree of comfort. In particular, by considering the predicted temperature for each hour and the air conditioning load conditions according to the time of day, it is possible to accurately estimate the target power consumption for each hour, thereby suppressing the decline in comfort.

[0107] Furthermore, even if the annual target power consumption is set appropriately, if the allocation to each month, day, and time slot is not appropriate, there is a possibility that there will be times when users feel uncomfortable. In contrast, according to this embodiment, since the allocation coefficient is calculated based on actual data, it becomes possible to allocate power consumption to time slots according to the actual situation. This allows for allocating more power consumption to time slots with high air conditioning load and reducing the amount of power consumption allocated to time slots with low air conditioning load, enabling control that balances economy and comfort.

[0108] <Second Embodiment> In the first embodiment, a predictive model 35 is constructed by learning from the actual performance data of the air conditioner 3. However, it is possible that sufficient performance data may not be accumulated. In the second embodiment, the actual performance data is increased using a technique called data augmentation or data inflation.

[0109] Figure 19 shows a schematic configuration of the power consumption estimation unit according to the second embodiment. The power consumption estimation unit 30A according to the second embodiment includes a data acquisition unit 31, a data expansion unit 36, a learning unit 32, an estimation unit 33, and a storage unit 34. Components of the power consumption estimation unit 30A that are the same as those of the power consumption estimation unit 30 are denoted by the same reference numerals, and their respective descriptions are omitted. The data expansion unit 36 ​​creates dummy training data based on actual data. The storage unit 34 stores the dummy training data created by the data expansion unit 36. The other configurations are the same as in the first embodiment.

[0110] Figure 20 shows an example of data augmentation processing by the data augmentation unit 36. The data expansion unit 36 ​​acquires the settings for the increase parameters (step S201). The user sets in the data expansion unit 36 ​​how many times to increase the actual data (e.g., by 2 times), which parameters to vary (e.g., temperature and power consumption), and the range of variation (e.g., ±1 degree for temperature, ±2 kWh for power consumption). Next, the data acquisition unit 31 acquires actual data (step S202). The data acquisition unit 31 records the acquired actual data in the storage unit 34. Next, the data expansion unit 36 ​​copies the actual data acquired by the data acquisition unit 31 (step S203) and generates dummy training data. For example, if it is set to 2x in step S201, the data expansion unit 36 ​​copies the actual data once to double the amount of data, and if it is set to 3x, it copies it twice to triple the amount of data. Next, the data expansion unit 36 ​​changes the date of the copied dummy training data (step S204). For example, changing the actual data from January 1, 2023 to August 1, 2023 is not appropriate given the nature of the air conditioner 3, so only the year is changed, for example, to January 1, 2022. Next, the data expansion unit 36 ​​adds noise to the dummy training data based on the parameter to be varied and its variation range set in step S201 (step S205). In the example above, the data augmentation unit 36 ​​adds noise to the temperature and power consumption of the copied dummy training data. For temperature, the data augmentation unit 36 ​​generates random numbers within a range of ±1 degree and adds the random number to the original temperature. Similarly, for power consumption, the data augmentation unit 36 ​​generates random numbers within a range of ±2 kWh and adds the random number to the original power consumption. Next, the data augmentation unit 36 ​​records the increased dummy training data in the storage unit 34 and merges the dummy training data with the actual data (step S206). The learning unit 32 uses the data after adding the dummy training data to the actual data as training data to construct the prediction model 35.

[0111] Figure 21 shows an example of the increased performance data. The lower part of Figure 21 shows the expanded training data. The 2023 data is the original performance data, and the 2022 data is dummy data. Increase category = 0 is the original data, and increase category = 1 is the increased dummy data. For the data with increase category = 1, the weights may be reduced during training to construct the predictive model 35.

[0112] According to the second embodiment, by expanding the data, a prediction model 35 can be constructed that enables highly accurate monthly, daily, and hourly power consumption predictions even when there is little accumulated historical data. By constructing a highly accurate prediction model 35, it is possible to set a highly accurate and realistic power consumption upper limit Pt[W], and as a result, it is possible to perform air conditioning control that suppresses power consumption to the target power consumption while maintaining comfort.

[0113] Furthermore, the data augmentation process according to the second embodiment can be applied to data other than training data for a prediction model that predicts the power consumption of the air conditioner 3. For example, it can be applied to the augmentation of time history data related to the operation of any equipment. The time history data may include information such as date and time, values ​​indicating the state of the equipment at that time, values ​​indicating the state of the operating environment, values ​​input to the equipment (e.g., command values), values ​​output from the equipment (e.g., values ​​calculated and output by the equipment, engine or turbine output, CO2 emissions, etc.), energy required for the operation of the equipment (power consumption, fuel supply amount, etc.), and changes resulting from the operation of the equipment (e.g., changes occurring in other equipment). The augmented data can be used as training data for machine learning. For example, training data can be prepared by copying the operation data of a plant, changing the time information, varying parameters such as pressure and temperature included in the operation data within a predetermined range, and labeling whether the plant's state is normal or abnormal. Then, by adding this training data to the original operation data and performing machine learning, a model for anomaly detection can be constructed.

[0114] <Third Embodiment> In the first embodiment, a predictive model 35 is constructed by learning actual data from the air conditioner 3. In the second embodiment, the predictive model 35 is constructed by expanding (increasing) the limited actual data from the air conditioner 3. However, there are cases where actual data from the air conditioner 3 is unavailable. In the third embodiment, in such cases, a property in a similar environment is selected, and the actual data from the selected property is used to estimate the monthly, daily, and hourly power consumption.

[0115] Figure 22 shows a schematic configuration of the power consumption estimation unit according to the third embodiment. The power consumption estimation unit 30B according to the third embodiment includes a data acquisition unit 31, an object selection unit 37, a learning unit 32, an estimation unit 33B, and a storage unit 34. Components of the power consumption estimation unit 30B that are the same as those of the power consumption estimation unit 30 are denoted by the same reference numerals, and their respective descriptions are omitted. The data acquisition unit 31 acquires property information and stores it in the storage unit 34. A property is a facility, either owned or owned, in which an air conditioner is installed. Property information includes, for example, the location of the building in which the air conditioner is installed, the type of business (shop, factory, office, home, etc.), capacity (kW), window orientation and size, wall material, sunlight exposure, volume or floor area of ​​the space to be air-conditioned, and the year of construction.

[0116] The property selection unit 37 selects a property that operates under similar environmental and conditional conditions to the air conditioner 3 in building 2. The property selection unit 37 refers to the property information acquired by the data acquisition unit 31 and selects a property with a similar property profile. The process by which the property selection unit 37 selects a similar property will be explained with reference to Figure 23. (Step 1) Assume the property set is as follows: A set of properties = {Property 1, Property 2, Property 3, Property 4, Property 5, Property 6} (Step 2) From these, a subset is selected, for example, from the following perspectives: The property profile of each property (location, type of business, capacity, window orientation and size, wall material, sunlight, volume and floor area of ​​the space to be air-conditioned, year of construction, etc.) is compared using the distance D_i (i is the property number) on the feature space, and the top k properties with the highest similarity are selected. Figure 23 shows, as an example, a diagram in which three parameters from the property profile are taken as coordinate axes in a 3D coordinate space, and each property is plotted in this coordinate space. The property selection unit 37 calculates the distance (Euclidean distance) between the coordinate information of the own property (building 2) and the coordinate information of other properties, for example, and selects the property with the shortest distance. As shown in the figure, properties 3 and 6 are located close to the own property, and property 5 is located a little further away. The property selection unit 37 calculates the distance D_i between the own property and each property, and selects the k properties in order from the shortest distance. If k=3, then in the case of Figure 23, {Property 3, Property 6, Property 5} are selected. Note that in Figure 23, only three parameters are used for illustrative purposes, but the property selection unit 37 may also calculate the distance D_i from its own property when each property is plotted in multidimensional space using all the parameters of the property profile (for example, the nine types of parameters mentioned above).

[0117] The estimation unit 33B estimates the monthly, daily, and hourly power consumption based on the actual data of the property selected by the property selection unit 37 (here, it is assumed that at least one year's worth of actual data, as illustrated in Figure 3, is available for each property). The estimation unit 33B aggregates the actual data of property 3 on a monthly basis to calculate the power consumption for each month. Similarly, the estimation unit 33B aggregates the actual data of property 5 and property 6 on a monthly basis to calculate the power consumption for each month. The estimation unit 33B calculates the average power consumption of property 3, property 5, and property 6 for each month, and uses the calculated average value for each month as the power consumption of its own property for each month.

[0118] Furthermore, the estimation unit 33B aggregates the actual data for property 3 on a daily basis to calculate the daily power consumption for each day from January 1st to December 31st. Similarly, the estimation unit 33B aggregates the actual data for properties 5 and 6 on a daily basis to calculate the daily power consumption for each of them. The estimation unit 33B may also calculate the average daily power consumption for properties 3, 5, and 6 and use the calculated daily average as the daily power consumption for its own property.

[0119] Furthermore, the estimation unit 33B aggregates the actual data of property 3 by time period to calculate the amount of electricity consumed for each time period on each day from January 1st to December 31st. Similarly, the estimation unit 33B aggregates the actual data of properties 5 and 6 by time period to calculate the amount of electricity consumed for each time period on each day. The estimation unit 33B may also calculate the average value of the electricity consumed for each time period of property 3, property 5, and property 6, and use the calculated average value as the electricity consumed for each time period of its own property.

[0120] Next, we will explain the process for estimating annual power consumption when the estimation unit 33B calculates power consumption on a monthly, daily, and hourly basis. Figure 24 is a flowchart showing an example of the annual power consumption estimation process according to one embodiment. As a prerequisite, it is assumed that the data acquisition unit 31 has previously acquired property information for its own property (building 2) and other facilities, and that this data is recorded in the storage unit 34. The property selection unit 37 selects properties similar to its own property (step S301). For example, the property selection unit 37 may select properties whose distance from its own property is shorter than a predetermined value when each property is plotted in a multidimensional space, or it may select a predetermined number of properties in order of increasing distance, or it may select a predetermined number of properties in order of increasing distance from its own property from among properties whose distance from its own property is shorter than a predetermined value. Next, the data acquisition unit 31 acquires actual data for the selected properties (step S302). The data acquisition unit 31 records the acquired actual data in the storage unit 34.

[0121] Next, the estimation unit 33B calculates the power consumption for each time period (step S303). The estimation unit 33B calculates the average power consumption for each time period by summing the power consumption for each time period for each day from January 1st to December 31st using the actual data for each property, and dividing by the number of properties. At this time, a weighted average (with a larger weight assigned to shorter distances) may be calculated by assigning weights according to the distance between the current property and similar properties in multidimensional space. Alternatively, instead of the average or weighted average, the estimation unit 33B may calculate the median, maximum, or mode of the power consumption for each time period for multiple properties. The same applies to the following steps S304 and S305.

[0122] Next, the estimation unit 33B calculates the daily power consumption (step S304). The estimation unit 33B calculates the average daily power consumption by summing the daily power consumption data for each property from January 1st to December 31st and dividing by the number of properties. At this time, a weighted average (with a larger weight assigned to shorter distances) may be calculated by assigning weights according to the distance between the current property and similar properties in multidimensional space. Alternatively, the estimation unit 33B may calculate the daily power consumption by aggregating the average power consumption for each time period calculated in step S303 on a daily basis.

[0123] Next, the estimation unit 33B calculates the monthly power consumption (step S305). The estimation unit 33B aggregates the actual data for each property for each month. Then, the estimation unit 33B calculates the average monthly power consumption by summing the daily power consumption for each property on a monthly basis and dividing by the number of properties. At this time, a weighted average (with a larger weight assigned to shorter distances) may be calculated by assigning weights according to the distance between the current property and similar properties in multidimensional space. Alternatively, the estimation unit 33B may calculate the monthly power consumption by aggregating the average daily power consumption calculated in step S304 on a monthly basis.

[0124] Next, the estimation unit 33B calculates the annual power consumption (step S306). The estimation unit 33B aggregates the actual data for each property for each month. Then, the estimation unit 33B calculates the average annual power consumption by summing the monthly power consumption for each property and dividing by the number of properties. Alternatively, the estimation unit 33B may calculate the annual power consumption by summing the average monthly power consumption calculated in step S305.

[0125] Next, the estimation unit 33B calculates the apportionment coefficients by month, day, and time of day (step S307). For example, the estimation unit 33B calculates the apportionment coefficient for January by dividing the amount of electricity consumed in January, calculated in step S305, by the amount of electricity consumed for the year, calculated in step S306. Similarly, the estimation unit 33B calculates the apportionment coefficients for February through December. The estimation unit 33B outputs the apportionment coefficients for each month to the electricity consumption control unit 10. The electricity consumption control unit 10 receives and stores the apportionment coefficients for each month.

[0126] Furthermore, the estimation unit 33B calculates the apportionment coefficient for January 1st by dividing the amount of power consumption for January 1st, calculated in step S304, by the amount of power consumption for January, calculated in step S305. Similarly, the estimation unit 33B calculates the apportionment coefficient for each day from January 2nd to December 31st. The estimation unit 33B outputs the apportionment coefficient for each day from January 1st to December 31st to the power consumption control unit 10. The power consumption control unit 10 receives and stores the apportionment coefficient for each day.

[0127] Furthermore, the estimation unit 33B calculates the apportionment coefficient for each time period on January 1st by dividing the amount of power consumption for each time period on January 1st, calculated in step S303, by the amount of power consumption for January 1st, calculated in step S304. Similarly, the estimation unit 33B calculates the apportionment coefficient for each time period on each day from January 2nd to December 31st. The estimation unit 33B outputs the apportionment coefficient for each time period on each day from January 1st to December 31st to the power consumption control unit 10. The power consumption control unit 10 receives and stores the apportionment coefficient for each time period.

[0128] Next, the estimation unit 33 outputs the annual power consumption calculated in step S306 to a display device or electronic file (step S308). The user checks the outputted annual power consumption and sets the annual target power consumption based on this value (step S1 in Figure 16). Subsequent processing is the same as in the first embodiment. The power consumption control unit 10 sets the power consumption upper limit Pt based on the annual target power consumption (flowchart in Figure 16).

[0129] According to the third embodiment, the annual target power consumption and apportionment coefficient are calculated using actual data obtained in environments with similar air conditioner capacity and operating environment / conditions. This allows for the appropriate setting of the annual target power consumption and apportionment coefficient even if actual data for the property has not been accumulated, enabling air conditioning control that suppresses power consumption to the target amount while maintaining comfort.

[0130] In this case, the amount of electricity consumed per hour, per day, per month, and per year can be calculated from the actual data of similar properties, and the apportionment coefficient is calculated from these. However, the apportionment coefficient may be calculated in the same way as in the first embodiment.

[0131] Figure 25 shows an example of the hardware configuration of a centralized monitoring device according to each embodiment. The computer 900 includes a CPU 901, a main memory 902, an auxiliary memory 903, an input / output interface 904, and a communication interface 905. The aforementioned centralized monitoring device 8 is implemented in the computer 900. The functions described above are stored in auxiliary storage device 903 in the form of programs. The CPU 901 reads the programs from auxiliary storage device 903, loads them into main memory 902, and executes the above processes according to the programs. The CPU 901 also allocates memory space in main memory 902 according to the programs. The CPU 901 also allocates memory space in auxiliary storage device 903 to store data being processed according to the programs.

[0132] Furthermore, a program to implement all or part of the functions of the centralized monitoring device 8 may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be loaded into a computer system and executed to perform processing by each functional unit. Here, "computer system" includes hardware such as the OS and peripheral devices. Also, if a WWW system is used, "computer system" also includes the homepage provisioning environment (or display environment). Furthermore, "computer-readable recording medium" refers to portable media such as CDs, DVDs, USBs, and storage devices such as hard disks built into the computer system. Furthermore, if this program is distributed to computer 900 via a communication line, computer 900 that receives the distribution may load the program into main memory 902 and execute the above processing. Furthermore, the above program may be for implementing some of the functions described above, and may also be able to implement the above functions in combination with programs already recorded in the computer system. The centralized monitoring device 8 may be composed of multiple computers 900. Furthermore, the power consumption control unit 10 can also exist as a separate device from the central monitoring device 8, implemented on a different computer 900, and functioning as a power consumption control device. In this case, for example, the central monitoring device 8 and the power consumption control device are configured to communicate with each other, and the various processes described above are realized through this communication.

[0133] Furthermore, it is possible to replace the components in the above-described embodiments with well-known components as appropriate, without departing from the spirit of the present invention. Also, the technical scope of this invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention. The power consumption estimation unit 30A is an example of a data expansion device.

[0134] <Note> The data expansion device, data expansion method, and program described in each embodiment can be understood, for example, as follows:

[0135] (1) A data expansion device according to the first embodiment includes means for acquiring time history data relating to the operation of equipment, means for copying the time history data to generate duplicate data, means for changing time information included in the duplicate data, means for varying at least one of the values ​​relating to the operation of the equipment included in the duplicate data within a predetermined range, and means for merging the time history data and the duplicate data. This allows for an increase in the amount of time history data.

[0136] (2) The data expansion device according to the second embodiment is the data expansion device of (1), wherein the values ​​relating to the operation of the equipment include at least one of the following: a value indicating the state of the equipment, a value indicating the state of the operating environment of the equipment, a value input to the equipment, a value output from the equipment, and a value consumed by the equipment. This makes it possible to increase the amount of time-history data that includes varying values ​​related to the operation of the equipment.

[0137] (3) A data expansion device according to the third embodiment is a data expansion device according to any one of (1) to (2), further comprising means for receiving settings regarding how many times the copying will be performed, which of the values ​​relating to the operation of the device will be changed, and by how much the values ​​will be changed. This allows for the arbitrary generation of duplicate data.

[0138] (4) The data expansion device according to the fourth embodiment is the data expansion device described in any of (1) to (3), wherein the device is an air conditioner, and the time history data includes temperature and the amount of power consumed by the air conditioner. This allows for the appropriate setting of annual energy consumption targets.

[0139] (5) The data augmentation device according to the fifth embodiment is the data augmentation device described in any of (1) to (4), which is used to increase the amount of training data when constructing a machine learning model that uses the time history data as training data. By increasing the amount of time history data and securing a sufficient amount of training data, the accuracy of the learning process can be ensured.

[0140] (6) In the data augmentation method according to the sixth aspect, a computer acquires time history data relating to the operation of the equipment, copies the time history data to generate duplicate data, modifies the time information contained in the duplicate data, varies at least one of the values ​​relating to the operation of the equipment contained in the duplicate data within a predetermined range, and merges the time history data and the duplicate data.

[0141] (7) The program according to the seventh aspect causes a computer to perform the following processes: acquire time history data relating to the operation of the equipment, copy the time history data to generate duplicate data, modify the time information contained in the duplicate data, vary at least one of the values ​​relating to the operation of the equipment contained in the duplicate data within a predetermined range, and merge the time history data and the duplicate data. [Explanation of Symbols]

[0142] 1. Air conditioning system 3. Air conditioner 4...Outdoor unit 5...Indoor unit 8... Central monitoring device 10. Power Consumption Control Unit 11. Input Information Acquisition Unit 12...Storage section 13. Settings section 15...1st judgment section 16...Second judgment section 17. Power Limit Adjustment Section 18... Correction section 20. Bankruptcy Prevention Department 21. Mode determination unit 22. Economic Adjustment Department 23. Comfort adjustment section 25...1st amendment part 26...2nd revised part 27...Third Amendment Part 28...4th amendment part 30, 30A, 30B...Power consumption estimation section 31. Data Acquisition Unit 32. Learning Department 33, 33B...Estimation part 34...Storage section 35. Predictive Models 36. Data Expansion Section 37. Property Selection Section 40.. Operation Control Unit 50... Communications Department

Claims

1. A means for acquiring time history data related to the operation of equipment, A means for copying the aforementioned time history data to generate duplicate data, Means for changing the time information contained in the aforementioned replicated data, Means for varying values ​​related to the operation of the equipment included in the replicated data within a predetermined range, Means for merging the aforementioned time history data and the aforementioned duplicate data, A data expansion device equipped with the following features.

2. The values ​​relating to the operation of the aforementioned equipment include at least one of the following: a value indicating the status of the equipment, a value indicating the status of the operating environment of the equipment, a value input to the equipment, a value output from the equipment, and a value consumed by the equipment. The data expansion device according to claim 1.

3. A means for receiving settings regarding how many copies to make, which values ​​related to the operation of the equipment to change, and by how much to change them. The data expansion device according to claim 1 or claim 2, further comprising:

4. The aforementioned device is an air conditioner, and the time history data includes the temperature and the power consumption of the air conditioner. A data expansion device according to claim 1 or claim 2.

5. When constructing a machine learning model using the aforementioned time history data as training data, the following is used to increase the amount of training data: A data expansion device according to claim 1 or claim 2.

6. Computers We acquire time history data related to the operation of the equipment. The aforementioned time history data is copied to generate duplicate data, The time information included in the aforementioned duplicated data is modified, The values ​​related to the operation of the equipment included in the replicated data are varied within a predetermined range. Merging the aforementioned time history data and the aforementioned duplicate data, Data expansion method.

7. On the computer, We acquire time history data related to the operation of the equipment. The aforementioned time history data is copied to generate duplicate data, The time information included in the aforementioned duplicated data is modified, The values ​​related to the operation of the equipment included in the replicated data are varied within a predetermined range. A process of merging the aforementioned time history data and the aforementioned duplicate data. A program that executes the command.

Images