JP2015201972A - Demand prediction device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a demand prediction device and a program that can determine a demand power amount for a predetermined prediction period by using a proper prediction model every branch circuit to thereby enhance the prediction precision of the demand power amount.SOLUTION: A demand prediction device 10 has a first achieving part 111 and a prediction part 12. The first achieving part 111 achieves first information as a variation factor of power to be consumed by an electrical load 24. The prediction part 12 determines a demand power amount required during a predetermined prediction period for each of branch circuits branched to plural systems in a distribution board 21 by using the first information achieved by the achieving part 111 before the predetermined prediction period is started. The prediction part 12 further applies the first information achieved by the first achieving part 111 to a prediction model set for each branch circuit 22 in which the first information corresponds to the demand power amount, thereby determining the demand power amount for the prediction period for each branch circuit 22.

Description

  The present invention relates to a demand prediction device that predicts and obtains a demand power amount, and a program for causing a computer to function as a demand prediction device.

  Conventionally, a technique for predicting power demand for each house and encouraging power saving behavior has been proposed. In addition, a technique for predicting electric power demand using an arithmetic expression defined using environmental information and energy consumption measurement information stored in a database has been proposed (for example, see Patent Document 1). The environmental information described in Patent Document 1 includes weather forecast data and the like. Furthermore, Patent Document 1 also describes a technique for switching between a weekday database and a holiday database when predicting energy consumption.

JP2013-109550A

  By the way, the technique described in Patent Document 1 calculates an energy consumption predicted value in units of houses. That is, the prediction is performed based on the prediction that there is a correlation between environmental information such as weather conditions and the energy consumption of the entire house.

  However, since there are many types of equipment that consumes energy in homes, even if energy consumption is predicted in units of homes, it cannot be said that there is a sufficiently strong correlation between environmental information and energy consumption. I want.

  An object of this invention is to provide the demand prediction apparatus which improved the prediction precision of the demand electric energy in a predetermined prediction period. Furthermore, an object of this invention is to provide the program for functioning a computer as this demand prediction apparatus.

  The demand prediction apparatus according to the present invention includes a first acquisition unit that acquires first information that becomes a variation factor of power consumed by an electric load, and the first acquisition unit before a predetermined prediction period is started. Using the first information acquired by the power distribution board, a prediction unit for obtaining a demand power amount required for the prediction period for each branch circuit branched into a plurality of systems by a distribution board, and the prediction unit, By fitting the first information acquired by the first acquisition unit to the prediction model in which the first information corresponds to the amount of power demand and is set for each of the branch circuits, The demand power amount is obtained for each branch circuit.

  In the present invention, a prediction model is set for each branch circuit, and a demand power amount is obtained for each branch circuit. With this configuration, the present invention makes it possible to determine the amount of power demand in a predetermined prediction period using an appropriate prediction model for each branch circuit, and as a result, has the advantage that the prediction accuracy of the amount of power demand increases. is there.

It is a block diagram which shows an example of embodiment. It is a figure which shows the example of a relationship between external temperature and power consumption value in embodiment. It is a figure which shows the example of the distribution regarding the power consumption value in embodiment. It is a figure which shows the example of the frequency which drives an electric load in embodiment. It is a figure which shows the transition example of a power consumption value in embodiment. It is a figure which shows the example of a relationship between external temperature and the peak value of a power consumption value in embodiment. It is a figure which shows the transition example of a power consumption value in embodiment. It is a figure which shows the transition example of the difference of a power consumption value in embodiment.

  As shown in FIG. 1, the demand prediction apparatus 10 of this embodiment includes a first acquisition unit 111 and a prediction unit 12. The 1st acquisition part 111 acquires the 1st information used as the fluctuation factor of the electric power which electric load 24 consumes. The prediction unit 12 uses the first information acquired by the first acquisition unit 111 before the predetermined prediction period starts to predict each branch circuit 22 branched into a plurality of systems by the distribution board 21. Obtain the amount of power demanded during the period. Furthermore, the prediction unit 12 applies the first information acquired by the first acquisition unit 111 to the prediction model in which the first information corresponds to the amount of power demand and is set for each branch circuit 22. The amount of power demand in the prediction period is obtained for each branch circuit 22. In the following, the case where the target for predicting the amount of power demand is assumed to be the house 20, but the target for predicting the amount of power demand may be a store, an office, an educational facility (school, cram school), or the like. .

  In the demand prediction device 10 of the present embodiment, the power received from the power supply 25 is branched into a plurality of branch circuits 22 by the distribution board 21 and an electric load 24 is connected to the branch circuit 22. There are no particular restrictions. The power supply 25 may include not only a system power supply (commercial power supply) supplied by an electric utility but also a distributed power supply such as a solar power generation device, a power storage device, and a fuel cell system. The power consumption of the electrical load 24 is measured by the measuring device 23 for each branch circuit 22.

  The measurement device 23 employs either a configuration built in the distribution board 21 or a configuration arranged outside the distribution board 21. The branch circuit 22 and the electrical load 24 may correspond one-to-one or one-to-many. For example, an electrical load 24 with relatively large power consumption such as an air conditioner, an IH cooking heater (IH: Induction Heating), and a microwave oven may be connected to the branch circuit 22 on a one-to-one basis. When the branch circuit 22 and the electrical loads 24 are connected in a one-to-many manner, the branch circuit 22 is often assigned in units of rooms in the house 20. In the present embodiment, in order to simplify the description, no consideration is given to the case where a single electrical load 24 is connected to the branch circuit 22 and the case where the branch circuit 22 corresponds to a room of the house 20.

  The measuring device 23 monitors the passing current of each branch circuit 22 with a sensor selected from a Rogowski coil, a clamp-type current sensor, and the like, and calculates the product of the monitored current value and the voltage value between the lines of the branch circuit 22. The integrated value is calculated as the power consumption value. That is, the power consumption value measured by the measuring device 23 is not actually an instantaneous value but an amount of power per predetermined unit time. The unit time is selected, for example, in the range of about 1 second to 10 minutes, and preferably 30 seconds or 1 minute. The unit time may be about 1 second.

  Although the instantaneous power for each branch circuit 22 fluctuates with time even within the unit time, in this embodiment, the fluctuation of the instantaneous power within the unit time is not taken into consideration, and the integrated power amount per unit time is calculated. The power consumption value. By dividing this power consumption value by unit time, an average value of power consumption in unit time can be obtained. As the unit time is shorter, the change in the power consumption value is monitored in detail. However, the amount of information to be processed increases, and more detailed information increases than necessary, so that the load increases. The unit time only needs to be set to such an extent that the time change tendency of the power consumption value can be recognized.

  The demand prediction apparatus 10 includes a computer that realizes the following functions by executing a program as a main hardware configuration. In other words, a program for causing a computer to function as the demand prediction apparatus 10 is provided. This type of computer may be a portable terminal device such as a smartphone or a tablet terminal in addition to a personal computer. In addition, the computer may have a configuration in which a processor and a memory are integrally provided, such as a microcomputer.

  When the computer which comprises the demand prediction apparatus 10 is selected from a personal computer, a smart phone, a tablet terminal, etc., it is desirable to provide the relay apparatus for receiving a power consumption value by communication from the measuring device 23. FIG. As this type of relay device, a HEMS controller (HEMS: Home Energy Management System) is preferably used, but an access point of a wireless LAN (Local Area Network) may be used for relay.

  The HEMS controller is configured to obtain the power consumption value from the measuring device 23 and to visualize the power consumption value by presenting the power consumption value to an appropriate presentation device. The HEMS controller can monitor the operation of the electrical load 24 and instruct the operation of the electrical load 24 by communicating with a specific electrical load 24 having a communication function. Since the HEMS controller also includes a computer (microcomputer), the demand prediction device 10 can also be realized as a partial function of the HEMS controller.

  The demand prediction device 10 may be configured by a server and a terminal device that communicate via an electric communication line such as the Internet. In this case, the terminal device may be configured to transfer the power consumption value acquired from the measurement device 23 to the server. In this configuration, a prediction model is generated in the server, and the demand power amount is predicted using the prediction model in the server. The demand prediction device 10 may be configured to directly acquire the power consumption value from the measurement device 23 without going through the relay device.

  In short, there are various modes for realizing the present embodiment, and the task as the demand forecasting apparatus 10 is performed according to hardware resources such as CPU (Central Processing Unit) processing capacity, memory capacity, or usage mode. It is configured to process with an appropriate device. Note that the program may be provided through a telecommunication line such as the Internet, in addition to being written in advance in a ROM (Read Only Memory). The program may be provided by a computer-readable recording medium.

  Below, the structure and operation | movement of the demand prediction apparatus 10 are demonstrated. Below, as shown in FIG. 1, the case where the demand prediction apparatus 10 is implement | achieved as one function of the server 40 is demonstrated. The demand prediction apparatus 10 of the present embodiment includes a first acquisition unit 111, a second acquisition unit 112, a prediction unit 12, a storage unit 13, an analysis unit 14, a model generation unit 15, and an output unit 16. Further, the demand prediction apparatus 10 includes a built-in clock 18.

  The first acquisition unit 111 acquires first information that becomes a variation factor of the power consumption value. Further, the second acquisition unit 112 acquires the power consumption value as second information. The first acquisition unit 111 may be configured to acquire only the variation factor of the power consumption value corresponding to the electric load 24 connected to the branch circuit 22 as the first information. It is desirable to collectively obtain the variation factors of the corresponding power consumption values.

  For example, if the electric load 24 connected to the branch circuit 22 is a cooling / heating device or a refrigerator, the variation factor of the power consumption value may be the outside air temperature, the weather, or the like. The outside air temperature means the outdoor air temperature announced by public authorities. Since the weather changes the amount of solar radiation incident on the room and changes the room temperature, it eventually becomes a variable factor of the power consumption value. That is, for this type of electrical load 24, at least the outside air temperature and the weather are the first information.

  The amount of solar radiation that enters the room also changes depending on factors such as the solar altitude and the open / close state of the curtains or blinds. However, the influence of these factors on the power consumption value is affected by the influence of the weather on the basis of human habits. Basically, the weather should be considered. In other words, if it is a time zone when the amount of solar radiation increases indoors and it is clear, people will try to reduce the amount of solar radiation by curtains or blinds, and if it is cloudy, people will not use curtains or Operate the blinds. Therefore, the influence on the room temperature, such as the open / close state of the curtain or blind, the solar altitude, etc. is folded into the influence of the weather.

  When the electric load 24 connected to the branch circuit 22 is a water heater, the outside air temperature, the water temperature, and the like are factors that vary the power consumption value. When the electric load 24 is an IH cooking heater, the room temperature, the temperature of the tap water, etc. It becomes a fluctuation factor of the power consumption value. In short, in both the hot water heater and the IH cooking heater, when boiling water, the power consumption value varies depending on the water temperature.

  Further, when the electric load 24 connected to the branch circuit 22 is a lighting device, the season, the weather, and the like become the fluctuation factors of the power consumption value. When a washing machine is connected to the branch circuit 22 as the electric load 24, the judgment as to whether or not to perform washing changes according to the weather, and the amount of washing is also considered to change. Become a factor.

  Furthermore, a change in the number of people in the house 20, a change in behavior, and the like over the entire electric load 24 become the fluctuation factors of the power consumption value. The number and behavior of the house 20 are generally determined by the characteristics of the day such as weekdays, holidays, and long vacations. The term “long vacation” means a holiday period for school or the like in the house 20 where children, school children, students, students and the like live together.

  As described above, the elements of the first information that become the fluctuation factors of the power consumption value include the outside air temperature, weather, day characteristics, water temperature, room temperature, and the like. It becomes a fluctuation factor of the value. Therefore, the first acquisition unit 111 acquires first information including elements according to the type of the electrical load 24.

  When the first information is the outside air temperature and the weather, the first acquisition unit 111 may acquire the first information through an electric communication line such as the Internet. When the air conditioner is an air conditioner, the first acquisition unit 111 may acquire the outside air temperature from a temperature sensor built in the outdoor unit. Further, if the water heater includes a heat pump type heat source device and the heat source device includes a temperature sensor, the first acquisition unit 111 may acquire the outside air temperature from the temperature sensor.

  The date characteristics can be acquired by using the function of the built-in clock 18 as a calendar. However, it is desirable that the demand prediction apparatus 10 is configured to be able to set holidays and long vacations separately from the function of the built-in clock 18 as a calendar. In other words, the demand prediction device 10 preferably includes the input unit 17 that receives input information from an appropriate input device 31, and the storage unit 13 stores the characteristics of the day that the input unit 17 receives as input information. . In this configuration, the first acquisition unit 111 can acquire the characteristics of the corresponding day by comparing the date and time counted by the built-in clock 18 with the input information stored in the storage unit 13. For the built-in clock 18, for example, a real time clock provided in the demand prediction device 10 is used.

  The characteristics of the day become a fluctuation factor of the power consumption value for most electric loads 24. Further, as described above, the branch circuit 22 does not necessarily correspond to the electric load 24 on a one-to-one basis, and the branch circuit 22 and the electric load 24 may be connected on a one-to-many basis. In this case, the power consumption value is measured in units of rooms, and the power consumption value measured for each branch circuit 22 may include a mixture of power consumption values of the plurality of electric loads 24. Thus, even if the power consumption value is in units of rooms, since the power consumption value generally differs between weekdays and holidays, the characteristics of the day become a variable factor.

  When the first acquisition unit 111 acquires the water temperature or room temperature as the first information, if it is intended to accurately measure the water temperature or room temperature, a temperature sensor for monitoring the water temperature or room temperature is required. May be estimated from information such as outside temperature and weather. That is, it is desirable to monitor the water temperature or room temperature using a dedicated temperature sensor, but a value estimated from the outside air temperature and weather may be substituted. After all, the first information may be meteorological information including the outside temperature, excluding the characteristics of the day. The first information acquired by the first acquisition unit 111 is not limited to the example described above, and is appropriately determined according to the type of the electrical load 24.

  The second acquisition unit 112 acquires the power consumption value for each branch circuit 22 measured by the measurement device 23. The second acquisition unit 112 is also used as an interface unit for connecting the measurement device 23. As described above, since the power consumption value as the second information is calculated by the measuring device 23 every unit time, the second acquisition unit 112 acquires the power consumption value for each branch circuit 22 every unit time. Note that the second acquisition unit 112 can also be configured to use the function of calculating the power consumption value per unit time in the measurement device 23.

  The storage unit 13 stores information that associates the first information acquired by the first acquisition unit 111, the second information acquired by the second acquisition unit 112, and the date and time that the built-in clock 18 measures. Remember. In other words, when the second acquisition unit 112 acquires the power consumption value for each branch circuit 22 as the second information, the first information acquired by the first acquisition unit 111 and the second information A combination with the date and time that the built-in clock 18 is measuring at the time of acquiring is stored in the storage unit 13. In short, the storage unit 13 collectively stores the first information, the second information obtained for each branch circuit 22, and the date and time.

  When the first acquisition unit 111 acquires the outside temperature and weather as the first information through the telecommunication line, the first information at the time when the power consumption value is acquired can be acquired in units of 10 minutes, for example. Also, the first information of the next day can be acquired, for example, in units of one hour. The time interval for acquiring the first information is not limited to the exemplified time.

  With the above-described operation, the storage unit 13 of the demand prediction device 10 stores a history of transition between the second information and the first information for each branch circuit 22. The storage unit 13 desirably has a capacity that can store the first information and the second information in a period selected from one week, one month, six months, one year, two years, and the like. The longer the period for storing the first information and the second information in the storage unit 13, the greater the amount of information and the possibility of generating a prediction model with high prediction accuracy. When the second information is collected, the start of use is delayed. Therefore, it is desirable to generate a prediction model temporarily using the first information and the second information of a relatively short period, and then correct the prediction model as necessary in order to improve prediction accuracy.

  By the way, in the present embodiment, when the first information is appropriately determined according to the type of the electrical load 24 connected to the branch circuit 22, the demand power amount can be estimated from the first information. It is structured based on the expectation that it will be possible. That is, if it is going to obtain | require the power demand amount in a predetermined prediction period, it is necessary to know the value in a prediction period regarding applicable 1st information. The value related to the first information is a temperature value if it is an outside temperature, means a clear day or a cloudy day if it is weather, and means a weekday or holiday if it is a day characteristic.

  The fact that the first information is defined for each branch circuit 22 means that a prediction model is set for each branch circuit 22. Therefore, if the value related to the first information in the prediction period is known, it is possible to obtain the amount of power demand in the prediction period by applying this value to the prediction model. That is, if the 1st acquisition part 111 has acquired the 1st information in a prediction period, it will become possible for prediction part 12 to ask for the amount of demand electric power in a prediction period.

  As is clear from the above description, the prediction model is generated so that the first information corresponds to the amount of power demand. The prediction model is expressed by a mathematical expression having the first information as an independent variable or a data table in which the first information is collated. The first information can be acquired in units of one hour for the outside air temperature or weather, and in units of one day for the characteristics of the day.

  For example, if the electrical load 24 is an air conditioner and the corresponding first information is the outside air temperature and the weather, the prediction model has a form in which the amount of power demand is linked to the combination of the outside air temperature and the weather. . Here, if the prediction period is the next day, the outside temperature and weather, which are the first information, can be acquired in units of one hour, so that the demand power amount can be predicted in units of one hour.

  However, since the electric load 24 such as an air conditioner is generally operated only during a part of the day, it is necessary to obtain the operating time in order to obtain the amount of power demand. Furthermore, since this type of electrical load 24 greatly varies in power consumption during operation, it is necessary to predict the demand power amount in consideration of fluctuations in the power consumption value. A technique for predicting a time zone during which this type of electric load 24 operates and predicting a demand power amount in anticipation of fluctuations in power consumption will be described later.

  Next, a procedure in which the model generation unit 15 generates a prediction model will be described. As described above, when the second acquisition unit 112 acquires the power consumption value for each branch circuit 22 as the second information, the storage unit 13 acquires the first acquisition unit 111 that has acquired the first power. A set of information and the date and time that the built-in clock 18 measures when the second information is acquired is stored. That is, when the history of the first information and the second information stored in the storage unit 13 is analyzed, there is a possibility that the relationship between the first information and the second information can be found for each branch circuit 22. .

  The analysis unit 14 extracts the relationship between the first information and the second information by evaluating the correlation between the two. The first information stored in the storage unit 13 includes outside air temperature, weather, day characteristics, and the like. Therefore, when these elements are used alone or in combination to obtain a correlation with the second information, it is predicted that a strong correlation with the second information appears.

  For the sake of simplicity, the correlation between the power consumption value during the cooling period and the outside air temperature is calculated for the branch circuit 22 connected to the air conditioner as the electric load 24 using the data stored in the storage unit 13. Suppose you want to. Note that the time interval at which the second acquisition unit 112 acquires the power consumption value as the second information is 1 minute. That is, it is assumed that the storage unit 13 stores the data of the power consumption value and the outside air temperature for each branch circuit 22 at intervals of 1 minute.

  A scatter diagram in which data obtained by combining power consumption values stored in the storage unit 13 and outside air temperature is plotted is, for example, as shown in FIG. The analysis unit 14 obtains a correlation coefficient between the outside air temperature and the power consumption value for the corresponding branch circuit 22. Since FIG. 2 is schematically described, the number of data is as small as about several tens, but actually, 1440 data can be plotted in one day.

  If the absolute value of the correlation coefficient is equal to or greater than an appropriate reference value in the relationship as in the illustrated example, the analysis unit 14 determines that a prediction model that links the first information and the second information can be generated. . The reference value is appropriately determined within a range of 0.6 to 0.8, for example. When the criteria set by the analysis unit 14 are satisfied, a prediction model is generated by the model generation unit 15. That is, the model generation unit 15 generates a prediction model in which the first information is associated with the amount of power demand, based on the combination of the first information stored in the storage unit 13 and the power consumption value. However, the power consumption value stored in the storage unit 13 does not correspond one-to-one with the first information, and the power consumption value varies with respect to the first information. Therefore, even in the prediction model, it is difficult to uniquely associate the first information with the demand power amount.

  Therefore, the model generation unit 15 includes a first prediction model in which the standard value of the demand power amount is associated with the first information, and a second value in which the boundary value in the allowable range of the demand power amount is associated with the first information. It is desirable to generate a prediction model of

  The first prediction model is obtained by approximating the power consumption value with a straight line with respect to the outside air temperature, for example, as shown by the solid line in FIG. That is, in order to generate the first prediction model, a prediction function of a linear function is set so as to obtain the amount of power demand using the first information as an independent variable. That is, the prediction formula is expressed in a format of a × (outside air temperature) + b × (demand power amount) + c = 0. When the coefficients a, b, and c in the prediction formula are determined by the least square method or the like using the power consumption value and the outside air temperature stored in the storage unit 13, a prediction formula serving as a first prediction model is obtained. The coefficients a, b, and c determined in this way are set as a standard value V0.

  The boundary value in the second prediction model is generated based on the distribution of the power consumption value with respect to the first information stored in the storage unit 13. Here, as shown in FIG. 3, the distribution of the power consumption value when the outside air temperatures are equal is applied to the normal distribution with the standard value V0 as an average value. Since the variance is obtained when the distribution of the power consumption value is fitted to the normal distribution, the boundary value can be determined by the size of the variance. For example, the upper boundary value is defined as μ + σ, the lower boundary value is defined as μ−σ, and the like using the average value μ and variance σ of the normal distribution. Since the boundary value is obtained for each outside air temperature, the boundary value obtained for each outside air temperature in the entire outside air temperature range is approximated by a straight line parallel to a prediction formula in which the coefficients a, b, and c are standard values V0. It is desirable to do. An example of the boundary value thus determined is shown by a one-dot chain line in FIG. That is, the second prediction model is a region between two straight lines indicated by a one-dot chain line in FIG.

  When the first prediction model and the second prediction model are generated as described above, the prediction unit 12 uses the first prediction model and the second prediction model to calculate the amount of power demand in the prediction period. Can be obtained. As a result, the prediction unit 12 can obtain the standard value of the demand power amount using the first prediction model, and can obtain the upper limit value and the lower limit value of the demand power amount using the second prediction model.

  The demand prediction device 10 includes an output unit 16 for outputting the demand power amount obtained by the prediction unit 12 to an appropriate presentation device 30. The presentation device 30 desirably employs an operation display device that is integrally provided with a display device that is a flat panel display such as a liquid crystal display device and an operation device such as a touch panel or a push button switch. This type of presentation device 30 may be a dedicated device for the demand prediction device 10, but if the output unit 16 has a function of communicating with the terminal device, the terminal device is adopted as the presentation device 30. It is possible. When a general-purpose terminal device selected from a personal computer, a smartphone, a tablet terminal, or the like is adopted as this type of terminal device, information can be exchanged with the demand prediction device 10 without using a dedicated presentation device 30. become.

  As the input device 31 that gives input information to the input unit 17 described above, it is desirable to use an operation display device that is integrally provided with a display device and an operation device, like the presentation device 30. That is, the input device 31 may be a dedicated device for the demand prediction device 10, but if the function of communicating with the terminal device is added to the input unit 17, the terminal device can be adopted as the input device 31. Is possible. The terminal device is desirably a general-purpose terminal device selected from a personal computer, a smartphone, a tablet terminal, and the like.

  The presentation device 30 and the input device 31 may share hardware. However, when the hardware of the presentation device 30 and the input device 31 is shared, the functions of the presentation device 30 and the input device 31 are realized by an application program (so-called application) executed by the terminal device.

  As described above, the second acquisition unit 112 acquires the power consumption value of the electric load 24 for each branch circuit 22 from the measurement device 23 as the second information. The storage unit 13 stores the first information and the second information in association with the date and time when the second information is acquired. The model generation unit 15 generates a prediction model using the first information and the second information stored in the storage unit 13.

  Moreover, the demand prediction apparatus 10 may include an analysis unit 14 that employs corresponding first information when the absolute value of the correlation coefficient between the first information and the second information is equal to or greater than a predetermined value. desirable. In this case, the model generation unit 15 generates a prediction model in which the demand power value is associated with the first information adopted by the analysis unit 14.

  Furthermore, it is desirable that the model generation unit 15 generates a first prediction model and a second prediction model. The first prediction model associates the standard value of the demand power amount with the first information. Based on the distribution of the second information with respect to the first information stored in the storage unit 13, the second prediction model associates the boundary value in the allowable range of the demand power amount with the first information.

  In addition, the model generation unit stores, in the storage unit 13, the probability that the demand power amount becomes a standard value with respect to the first information and the probability that the power consumption amount becomes a boundary value with respect to the first information. You may obtain | require based on distribution of the 2nd information with respect to 1 information.

  Furthermore, it is desirable that the prediction unit 12 is configured to obtain the standard value and the boundary value of the demand power amount by fitting the first prediction model and the second prediction model to the first information, respectively. . In this case, the demand prediction device 10 further includes an output unit 16 that associates the probability with the standard value and the boundary value of the demand power amount obtained by the prediction unit 12 and outputs the associated probability to the presentation device 30.

  The prediction model described above does not consider the period during which the electric load 24 is operated. Therefore, when only the above-described prediction model is used, the demand power amount of the electric load 24 operated all day or the demand power amount when the time zone for operating the electric load 24 is designated is obtained. Hereinafter, a technique for predicting a time zone in which the electric load 24 is operated and predicting a demand power amount related to the electric load 24 in which the power consumption value varies regardless of the first information will be described.

  The prediction model described here includes a time zone in which the electric load 24 is used for each branch circuit 22. In order to generate this prediction model, the analysis unit 14 starts operation of the electrical load 24 for each branch circuit 22 using data for a plurality of days (for example, one month or more) stored in the storage unit 13. And the time from the start of operation of the electric load 24 to the stop of operation is obtained. Hereinafter, the time when the operation of the electric load 24 is started is referred to as a start time, and the time from the start of operation of the electric load 24 to the operation stop is referred to as an operation time.

  Here, it is assumed that one electric load 24 is connected to one branch circuit 22. When the electric load 24 is, for example, an air conditioner, an IH cooking heater, an electric floor heater or the like, one branch circuit 22 is often dedicated to one electric load 24. In addition, an electric load 24 that hardly moves from the installation location such as a washing machine may be dedicated to one branch circuit 22.

  In this embodiment, the reference value is determined for each branch circuit 22 based on the standby power value, and the analysis unit 14 is connected to the corresponding branch circuit 22 when the power consumption value rises and exceeds the reference value. It is determined that the operation of the electric load 24 has started. In addition, when the power consumption value falls to the reference value, the analysis unit 14 determines that the operation of the electrical load 24 connected to the corresponding branch circuit 22 has stopped.

  By the way, the start time for each branch circuit 22 shows a similar tendency if the day characteristics in the first information are the same, and the operation time for each branch circuit 22 has the same day characteristics in the first information. Show similar trends. Therefore, when data having the same day characteristics in the first information is extracted from the storage unit 13 and the frequency of the start time and the frequency of the operation time are obtained for each branch circuit 22, the frequency is biased. It was seen.

  FIG. 4 shows an example of the frequency of the start time in the branch circuit 22 to which only the air conditioner is connected. In the three sections of 9:00, 12:00, and 22:00, the frequency is higher than the other sections. I understand. In FIG. 4, one section is set to one hour, but it is also possible to set a shorter section such as 30 minutes or 15 minutes. Each time given to a section represents the start time of the section. Therefore, the section of 9:00 means one hour from 9:00 to 10:00.

  After obtaining the start time frequency as shown in FIG. 4, the analysis unit 14 extracts a start time having a relatively high frequency. The number of start times to be extracted is not particularly limited, but the analysis unit 14 may treat it as an exception when the frequency is 1. In short, the analysis unit 14 may not use a start time with a frequency of 1 to generate a prediction model. In the example of FIG. 4, the start times with the third highest frequency are extracted. Accordingly, the extracted start times are 9:00, 12:00, and 22:00.

  The analysis unit 14 obtains the frequency for one operation time of the electric load 24 and extracts an operation time having a relatively high frequency. Here, when the operation time of one time is extracted for the branch circuit 22 in which only the same air conditioner as the branch circuit 22 for which the start time is obtained is extracted, the result is that the frequency of 1 hour, 3 hours and 4 hours is relatively high was gotten.

  Furthermore, when the probabilities of occurrence of three types of operation time for each of the three types of start times were obtained, the results shown in Table 1 were obtained. The probabilities in Table 1 indicate the ratios at which the respective operation times occur for the same start time.

  According to Table 1, when the start time is 9:00, the operation time is the largest for 3 hours, and when the start time is 12:00, the operation time is the most for 1 hour, and the start time is 22:00. And the driving time is 4 hours most. Here, whether or not to adopt as a prediction model is determined depending on the frequency of the start time, and a combination of operation times that maximizes the probability for each start time is adopted as the prediction model. Therefore, in the above-described example, the model generation unit 15 generates a prediction model in which one operation time is 3 hours when the start time is 9:00, and one time when the start time is 22:00. A prediction model with an operation time of 4 hours is generated.

  By the way, depending on the type of the electric load 24, the amount of power consumption may increase significantly immediately after the start of operation, and then the amount of power consumption may decrease to decrease the amount of fluctuation. Examples of the electric load 24 whose power consumption value changes in this way include an air conditioner such as an air conditioner, an IH cooking heater, a heat pump type water heater, and the like. Thus, the electrical load 24 mainly used for heating or cooling has a large difference between the temperature of the object to be heated or cooled and the target temperature immediately after the start of operation. Consumes a lot of power to get close to On the other hand, after the temperature of the object reaches the target temperature, only power for maintaining the target temperature is consumed, so that the power consumption value is lower than immediately after the start of operation.

  Therefore, the power consumption value in this type of electric load 24 tends to show a change that rises sharply from the start of operation, decreases after a peak value, and then becomes almost stable. Such a change is schematically shown in FIG. The change in the power consumption value on the left side in FIG. 5 indicates a prediction model with a start time of 9:00, and the operation time is 3 hours as described above. Further, the change in the power consumption value on the right side in FIG. 5 indicates a prediction model with a start time of 22:00, and the operation time is 4 hours as described above.

  When the power consumption value changes as shown in FIG. 5, it is known that a relatively strong correlation occurs between the peak value of the power consumption value and the outside air temperature as shown in FIG. The outside air temperature here is the temperature of the environment in which the electric load 24 is used, and includes the case of room temperature. However, the room temperature in this case is a temperature before using the air conditioner, and means a temperature that changes according to the outside air temperature. On the other hand, it is known that the power value in a state where the power consumption value is stable in FIG. 5 is determined by the peak value of the power consumption value and the outside air temperature.

  According to these findings, in the prediction model as shown in FIG. 5, if the relationship between the outside temperature and the peak value of the power consumption value is obtained in advance, the peak value of the power consumption value is obtained based on the outside temperature. . In addition, if the relationship for deriving the power value in a state where the power consumption value is stable from the peak value of the power consumption value and the outside air temperature is obtained in advance, the peak value of the power consumption value obtained from the outside air temperature and the outside air temperature Can be used to estimate the power value in a state where the power consumption value is stable.

  Now, the relationship between the outside temperature θ and the peak value Cp of the power consumption value is expressed by Cp = f (θ), and the power value Cs in a state where the power consumption value is stable is the peak value Cp of the power consumption value. It is assumed that it is expressed in the form of Cs = g (Cp, θ) using the outside temperature θ. When this relationship is established, it can be seen that Cs = g (f (θ), θ), and the power consumption value related to the prediction model eventually results in a function of the outside temperature θ.

  The analysis unit 14 divides the operation time in the prediction model into a period in which the power consumption value is stable and other periods based on the actual measurement in order to obtain an integrated value of power consumption with respect to the prediction model. Sum the integrated values of power consumption over the period.

  In order to divide the prediction model described above into two periods, data stored in the storage unit 13 is used as in the generation of the prediction model. In order to divide the prediction model into two periods, it is necessary to use data for a plurality of days, as in the case of generating the prediction model. Here, in order to briefly explain the procedure for dividing the two periods, 1 Use the data for the operation time of the operation.

  As shown in FIG. 7, when the electric load 24 is an air conditioner, the power consumption value rises rapidly immediately after the start of operation, and then decreases to a stable state. As time further elapses, the air conditioner repeats stopping and running intermittently. That is, when the room temperature reaches the target temperature by the air conditioner, the state where the operation is temporarily stopped and the state where the operation is resumed are intermittently repeated. This operation continues until the air conditioner is stopped. Note that, in the period in which the operation is temporarily stopped and restarted, the average value of the power consumption value can be regarded as the same level as the state where the power consumption value is stable.

  Here, since the operation is temporarily stopped within 10 minutes, it is necessary to consider that the air conditioner is in operation when the operation is resumed after the stop of this amount of time. Therefore, after the operation is stopped, the analysis unit 14 determines that the operation is stopped when the stopped state continues for 30 minutes, for example. On the other hand, the analysis unit 14 determines that the air conditioner has started operation when the power consumption value rises above the reference value. Thus, the analysis part 14 extracts the power consumption value in the period which operates an air conditioner once using the data which the memory | storage part 13 preserve | saved. For the electrical load 24 other than the air conditioner, the analysis unit 14 similarly determines the electrical load 24 that performs the same operation. However, the time for determining that the operation has been stopped is set in accordance with the target electric load 24.

  When the analysis unit 14 extracts the power consumption value during the period in which the air conditioner is operated once as described above, the model generation unit 15 generates a prediction model based on the extracted power consumption value. In order to divide the prediction model of the air conditioner into two periods, the difference between the data whose time is adjacent is obtained for the power consumption value extracted by the analysis unit 14. That is, if one power consumption value is P (n), the difference ΔP (n) is ΔP (n) = P (n) −P (n−1), and P (0) exceeds the reference value. The previous value is used. Note that n is a positive integer value.

  FIG. 8 shows the difference ΔP (n) obtained for the power consumption value shown in FIG. In FIG. 8, the positive and negative signs are not taken into consideration, and the difference ΔP (n) represents an absolute value. As is apparent from FIG. 8, the appearance frequency of a relatively large difference ΔP (n) increases during the period before and after the peak value Cp of the power consumption value. On the other hand, a relatively large difference ΔP (n) appears intermittently after a state in which the appearance frequency of a relatively small difference ΔP continues to be high during a period in which the power consumption value is stable.

  When the power consumption value changes as shown in FIG. 7, in order to divide the prediction model into two periods, it is necessary to determine the timing at which the power consumption value switches to a stable state. In order to determine this timing, for example, an appropriate threshold value is set for the difference ΔP (n) during the period in which the air conditioner is operated once, and the difference ΔP (n) is less than or equal to the threshold value in the time series of the difference ΔP (n). What is necessary is just to obtain | require the time of the state which becomes continuous over a predetermined time. This time is set to 30 minutes, for example, as in the case of extracting the timing when the operation of the air conditioner is stopped.

  The threshold value used for comparison with the difference ΔP (n) is obtained from the appearance frequency of the difference P (n) during the period in which the air conditioner is operated once. That is, when the difference ΔP (n) is divided into appropriate sections, a section where the appearance frequency is 0 occurs between sections having a relatively high appearance frequency, and the threshold is selected from this section.

  As described above, when the period during which the air conditioner is operated once is divided into two periods, the first period is a period including the peak value Cp of the power consumption value, and the latter period is a period in which the power consumption value is stable. become. Hereinafter, the first half period is referred to as a first period T11, and the second half period is referred to as a second period T12. The first period T11 and the second period T12 determined by the above procedure are shown in FIG. If the time for operating the air conditioner once is short and the power consumption value does not reach a stable state, the prediction model may not be divided into two periods.

  Here, if the change in the power consumption value in the prediction model is approximated as shown in FIG. 5, for example, the power consumption value in the first period T11 is applied to a Gaussian function, and the power consumption value in the second period T12 is approximately. This is true for a certain value. When fitting in this way, the peak value Cp in the first period T11 is expressed as a function of the outside air temperature, which is the first information, and the length of the first period T11 is known. An integrated value of electric power is obtained. Similarly, the power consumption value in the second period T12 is determined by the peak value Cp and the outside air temperature that is the first information, and the length of the second period T12 is known. A value is determined.

  As described above, if the integrated value of power consumption in the first period T11 and the integrated value of power consumption in the second period T12 are summed, the integrated value of power consumption in the period in which the air conditioner operates once can be obtained. It becomes possible. In addition, since the integrated value of power consumption obtained in this way is an approximate value, it is desirable to improve the accuracy of calculation by performing appropriate correction.

  As described above, when one operation time of the electric load 24 is divided, a set of a start time and an operation time for starting the operation of the electric load 24 is used as the prediction model. In addition, depending on the type of the electrical load 24 connected to the branch circuit 22, the model generation unit 15 generates a prediction model in which the period of one operation is divided into two periods. By using such a prediction model, it is possible to estimate an integrated value of power consumption during a period in which the electric load 24 is operated once.

  A result predicted using the prediction model generated by the model generation unit 15 is presented to the presentation device 30 through the output unit 16. Since the probability is determined in the prediction model, it is desirable to present not only the prediction result but also the predicted probability to the presentation device 30.

  The above-described demand prediction device 10 can obtain the amount of power demand for a predetermined prediction period in the house 20 for each branch circuit 22. Therefore, in the house 20, when receiving a power saving request based on DR information (DR: Demand Response), it is possible to obtain a guideline for selecting the branch circuit 22 that reduces the amount of power demand. Moreover, when controlling the electric power supplied from the power supply 25, it is possible to determine the supply electric energy based on the demand electric energy in the prediction period by using the electric power demand in the prediction period obtained by the demand prediction device 10. become.

  By the way, the demand prediction apparatus 10 presents the standard value, the upper limit value, and the lower limit value for the demand power amount to the presentation device 30, and gives a probability to each. Therefore, on the demand side that consumes power, it is possible to confirm the ratio of the standard value, the upper limit value, and the lower limit value of the demand power amount for each branch circuit 22. On the other hand, on the power supply side, when the demand power amount obtained by the demand prediction device 10 is aggregated, the probability of the shortage in the power supply amount due to the increase or decrease in the demand power amount with respect to the plan for the power supply amount is estimated. Is possible.

  As described above, the analysis unit 14 determines the operation start time and the operation time for which the operation continues for the electrical load 24 connected to the branch circuit 22 based on the second information stored in the storage unit 13. It may be estimated. In this case, the model generation unit 15 generates a prediction model for the corresponding electric load so as to include the start time and the operation time.

  Furthermore, any of the electric loads 24 has an electric power value that changes so that the power consumption value decreases to the peak value immediately after the start of operation and then decreases with time, and then the power consumption value becomes stable. The load 24 may be used. In this case, based on the second information stored in the storage unit 13, the analysis unit 14 has a first period in which the power consumption value decreases to the peak value and then decreases with time, and the power consumption value is It is desirable to extract the second period in which the state is stable. Then, the model generation unit 15 generates a prediction model for each of the first period and the second period.

  Note that the prediction model may be set to predict the demand amount in the prediction period for at least one of gas and tap water.

  That is, in the above-described configuration example, the model generation unit 15 predicts only the demand power amount, but is set to obtain the demand amount in the prediction period for at least one of the gas and tap water of the house 20. May be. Even when the prediction model predicts the demand amount related to gas or tap water, the basic operation is the same as the case where the demand power amount is obtained in that the demand amount is obtained based on the first condition. In this case, the second acquisition unit 112 is also configured to acquire the consumption of at least one of gas and tap water.

  The demand prediction apparatus 10 described above can be installed for each house 20. Moreover, when the HEMS controller is installed in the house 20, the power value from the measurement device 23 may be acquired from the HEMS controller through an electric communication line such as the Internet. In this configuration, the demand prediction device 10 is provided in a server that communicates through an electric communication line. That is, the server collects the power value consumed for each branch circuit from the plurality of houses 20 and generates a prediction model for each house 20.

  With this configuration, the abundant hardware resources of the server can be used for the storage unit 13, the analysis unit 14, the model generation unit 15, and the like. As a result, the house 20 only needs to be provided with a HEMS controller for acquiring the second information from the measuring device 23 and notifying the server of the second information, and the initial cost for using the demand prediction device 10. The increase of is suppressed. In addition, what is necessary is just to implement | achieve the presentation apparatus 30 which presents the information output from the demand prediction apparatus 10, and the input device 31 which gives an instruction | indication to the demand prediction apparatus 10 with a general purpose terminal device. In this case, the terminal device functions as the presentation device 30 and the input device 31 by executing the application program. The application program may be provided through a telecommunication line such as the Internet. The application program may be provided by a computer-readable recording medium.

  The above-described embodiment is an example of the present invention. For this reason, the present invention is not limited to the above-described embodiment, and various modifications can be made according to design and the like as long as the technical idea according to the present invention is not deviated from this embodiment. Of course, it can be changed.

DESCRIPTION OF SYMBOLS 10 Demand prediction apparatus 12 Prediction part 13 Memory | storage part 14 Analysis part 15 Model generation part 16 Output part 17 Input part 18 Built-in clock 20 House 21 Distribution board 22 Branch circuit 23 Measuring device 24 Electric load 30 Presentation apparatus 31 Input apparatus 111 1st 1st Acquisition unit 112 Second acquisition unit

Claims (11)

A first acquisition unit that acquires first information that becomes a variation factor of power consumed by the electrical load;
By using the first information acquired by the first acquisition unit before the predetermined prediction period is started, the prediction period is required for each branch circuit branched into a plurality of systems by the distribution board. A forecasting unit for determining the amount of power demand,
The prediction unit
By fitting the first information acquired by the first acquisition unit to the prediction model in which the first information corresponds to the amount of power demand and is set for each of the branch circuits, The demand prediction apparatus, wherein the demand power amount is obtained for each branch circuit. A second acquisition unit that acquires, as second information, a power consumption value of the electric load for each branch circuit;
A storage unit that stores the first information and the second information in association with the date and time when the second information is acquired;
The demand prediction device according to claim 1, further comprising: a model generation unit that generates the prediction model using the first information and the second information stored in the storage unit. When the absolute value of the correlation coefficient between the first information and the second information is greater than or equal to a predetermined value, the analyzer further includes an analysis unit that adopts the corresponding first information,
The demand prediction device according to claim 2, wherein the model generation unit generates the prediction model in which the demand power value is associated with the first information adopted by the analysis unit. The model generation unit
Generating a first prediction model in which a standard value of the demand electric energy is associated with the first information;
Furthermore, based on the distribution of the second information with respect to the first information stored in the storage unit, a second value in which a boundary value in the allowable range of the demand power amount is associated with the first information. The demand prediction device according to claim 3 which generates a prediction model. The model generation unit
The probability that the demand power amount becomes the standard value for the first information and the probability that the demand power amount becomes the boundary value for the first information are stored in the storage unit. The demand prediction device according to claim 4, wherein the demand prediction device is obtained based on a distribution of the second information with respect to information. The prediction unit is configured to obtain the standard value and the boundary value of the demand electric energy by fitting the first prediction model and the second prediction model to the first information, respectively. ,
The demand prediction device according to claim 5, further comprising an output unit configured to associate the probability with the standard value and the boundary value of the demand power amount obtained by the prediction unit and output the probability to a presentation device. The analysis unit
For the electrical load connected to the branch circuit, based on the second information stored in the storage unit, estimate the operation start time and the operation time during which the operation continues,
The model generation unit
The demand prediction apparatus according to any one of claims 2 to 6, wherein the prediction model is generated so as to include the start time and the operation time for a corresponding electric load. Any of the electric loads is an electric load whose power consumption value decreases to the peak value immediately after the start of operation and then decreases with time, and thereafter changes so that the power consumption value becomes stable. There,
The analysis unit
Based on the second information stored in the storage unit, a first period in which the power consumption value decreases to the peak value after the power consumption value has increased to a peak value, and the power consumption value becomes stable. 2 periods are extracted,
The model generation unit
The demand prediction device according to claim 7, wherein the prediction model is generated for each of the first period and the second period.   The demand prediction device according to any one of claims 1 to 8, wherein the first information is weather information including an outside air temperature. The demand prediction device according to any one of claims 1 to 9, wherein the prediction model is set to predict a demand amount in the prediction period for at least one of gas and tap water.   The program for functioning a computer as a demand prediction apparatus of any one of Claims 1-10.

Images