JP2016181060A - Electric power demand prediction system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an electric power demand prediction system with which it is possible to predict a total demand of electric power with high accuracy even while processing a massive amount of track record data quickly.SOLUTION: An electric power demand prediction system 1 comprises: a database 13 in which track record data indicating the amount of power consumption per prescribed time of a plurality of users are recorded; as group classification unit 14 for classifying the users into a plurality of groups on the basis of the track record data; and an electric power demand prediction unit 15 for predicting, on the basis of the track record data, a total demand of electric power that is a sum total of the electric power demand of the users in successively arriving future unit periods of a prescribed duration with the timing that precedes the starting point of time of the respective unit periods. The electric power demand prediction unit 15 selectively extracts some of the users from each of the groups, and predicts the total demand of electric power on the basis of the track record data of the extracted users.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a power demand prediction system that predicts power demand.

  In recent years, due to the deregulation of the electricity retail business, not only general electricity companies (businesses that concurrently serve the power generation business, power transmission and distribution business, and power retail business), but also other electric power companies represented by specific scale electric power companies ( (Hereinafter referred to as “new power operator”) can also operate a power retail business. In addition to the businesses that concurrently serve as the power generation business and the power retail business, the new power business operators include business operators that conduct only the power retail business (power producers that supply power to customers' customers through the power market, etc.) Etc.).

  The new electric power company uses the transmission / distribution network of a general electric company to consign and supply power to customer customers. For example, in the case of the plan value simultaneous equal amount system, the new electric power company will calculate the total power demand (hereinafter referred to as “total”) of customer customers in the future unit period (for example, a period of 30 minutes) that arrives in sequence. The demand plan indicating the total power demand is submitted to a general electric utility in advance (for example, one hour before the unit period arrives). Then, when the above unit period arrives, the new power company supplies the same amount of power as the total power demand indicated by the demand plan submitted to the general electric power company to the transmission and distribution network (or power market etc. To secure supply in the unit period in advance).

  At this time, the total power demand indicated by the demand plan formulated by the new power company does not always coincide with the actual total power demand, and may naturally deviate. However, this divergence is absorbed by adjusting (balancing) the power supplied by the general electric utility to the transmission / distribution network, so that the power in the transmission / distribution network is maintained in a stable state (for example, the voltage or frequency is Kept constant).

  When the above divergence occurs, the new electric power company and the general electric utility company settle the imbalance fee determined according to the magnitude of the divergence. Since the imbalance fee is a fee generated by the above-mentioned divergence generated by the new electric power company using the transmission and distribution network of the general electric power company, the new electric power business is more than the burden of the general electric power company. Is set so as to increase the burden on the user.

  Specifically, when the actual power demand is larger than the total power demand indicated by the demand plan formulated by the new power company, an imbalance fee is paid from the new power company to the general electricity company. This imbalance fee is set to be high because it is paid not only as compensation for the purchase of insufficient power but also as a penalty for threatening the stability of the transmission and distribution network.

  On the other hand, when the actual power demand is smaller than the total power demand indicated by the demand plan formulated by the new electric power company, an imbalance fee is paid from the general electric company to the new electric power company. However, since this imbalance fee is simply paid as a consideration for taking over surplus power, it is set to be low, and an imbalance fee is paid for surplus power exceeding the specified range. I will not.

  Thus, in order to operate the power retail business and gain profits efficiently, the new power company predicts the total power demand that approximates the actual total power demand without excess or deficiency, and makes a demand plan. It is necessary to formulate. Even if the above-mentioned system is not assumed, in order for general electric utilities and new electric utilities to supply stable power to society, general electric utilities and new electric utilities It is imperative that each of these accurately predict total power demand.

  The total power demand for each unit period indicated by the demand plan can be predicted based on the actual data indicating the past power consumption of the consumer. However, in order to predict the total power demand for each unit period, the actual data must be data indicating temporal fluctuations in the power consumption of the consumer.

  Therefore, Patent Document 1 proposes a power consumption amount estimation method for estimating a power consumption amount per predetermined time in a specific consumer in a pipe by a general electric utility. This power consumption estimation method is based on the assumption that the temporal variation pattern of power consumption is similar for consumers with the same “attributes” such as location, contract type, contract power, and industry. This makes it possible to estimate temporal fluctuations in the amount of power consumed by individual consumers who are not installed with an interval meter.

  Specifically, in the power consumption amount estimation method proposed in Patent Document 1, consumers whose power consumption per predetermined time is known are grouped according to “attributes”, and then the power consumption A standard load curve showing a temporal variation pattern of the quantity is calculated for each group. Then, for a consumer whose power consumption, which is the estimation target, is unknown, the group to which it belongs belongs based on its attributes, and the consumer's consumption based on the standard load curve of the identified group. Estimate temporal fluctuations in power consumption.

JP 2006-11715 A

  The power consumption estimation method proposed in Patent Document 1 is based on the premise that there are few consumers on which smart meters are installed. The purpose is to estimate the power consumption per hour. However, in recent years, the spread of smart meters is rapidly progressing due to the influence of deregulation of the power retail business. In particular, the government and general electric utilities are planning to install smart meters for all consumers.

  If smart meters are installed in a large number (and all) of consumers, the new power company can accurately predict the total power demand based on abundant performance data. However, on the other hand, there is a problem that the actual data to be processed by the new electric power company becomes enormous. Specifically, since the smart meter records and transmits the power consumption sequentially every predetermined time, a large number of performance data is generated per day (24 hours) in one consumer. For example, if the smart meter records the amount of power consumed every 30 minutes and there are 100,000 customers of new electric power company customers, less than 500,000 performance data will be generated in one day.

  In this way, the new power company can predict the total power demand with high accuracy by using the performance data sequentially generated by the smart meter. A new problem has arisen that data must be processed, which has never been imagined before.

  In particular, the prediction of the total power demand needs to be performed just before the deadline for submitting the demand plan because the accuracy can be improved more by using the latest performance data. Therefore, the prediction of the total power demand is repeatedly performed every unit period, and the time for processing the actual data for the prediction is inevitably less than the unit period. That is, the new power company is not only able to accurately predict the total power demand, but also needs to process a huge amount of performance data quickly for the prediction, which is a problem.

  Therefore, an object of the present invention is to provide a power demand prediction system capable of accurately predicting the total power demand while processing a huge amount of performance data quickly.

  In order to achieve the above object, the present invention provides a database in which actual data indicating power consumption per predetermined time in each of a plurality of consumers is sequentially recorded, and a plurality of consumers based on the actual data. Based on the actual data, the group classification unit for classifying the total power demand, which is the sum of the power demands of the customer in the future unit period of the predetermined time length that sequentially arrives, for each unit period A power demand prediction unit that sequentially predicts at each timing before the start time, and the power demand prediction unit selectively extracted and extracted some of the consumers from each of the groups Provided is a power demand prediction system characterized by predicting the total power demand based on the actual data of the consumer.

  In this power demand prediction system, the power demand prediction unit predicts the total power demand in the future unit period using the customer's performance data selectively extracted from each group. For this reason, in this power demand forecasting system, the actual data recorded in the database is used without any bias, and the decrease in the accuracy of forecasting the total power demand is suppressed, while the number of actual data used for forecasting the total power demand is reduced. It is possible to reduce and realize a quick process.

  Further, in the power demand prediction system having the above characteristics, the power demand prediction unit predicts a provisional total power demand that is a sum of power demands of the consumers in the unit period in the first period, The start time of the first period of some of the consumers selectively extracted from each of the groups in the second period starting after the start time and ending before the start time of the unit period The total power demand for the unit period is predicted by correcting the provisional total power demand based on at least a part of the actual data recorded in the database before and after the start of the second period. ,preferable.

  According to this power demand prediction system, the calculation for calculating the total power demand can be simplified by correcting the provisional total power demand afterwards to calculate the total power demand. As a result, the total power demand can be calculated quickly.

  Moreover, in the power demand prediction system having the above characteristics, the power demand prediction unit predicts the provisional total power demand of the plurality of unit periods in the first period, and at least one unit in the second period. The number of unit periods for predicting the total power demand for a period and predicting the total power demand for the second period is less than the number of unit periods for predicting the provisional total power demand for the first period It is preferable.

  According to this power demand forecasting system, the latest power data is used to accurately predict the total power demand, and the provisional total power demand for a plurality of unit periods is collectively estimated to reduce the load on the power demand forecasting unit. It becomes possible to reduce.

  Moreover, in the power demand prediction system having the above characteristics, the power demand prediction unit is configured to obtain an average power demand in the unit period of the consumers belonging to the group based on the actual data of the consumers belonging to the group. A group average power demand is calculated for each group, and for each group, a group total power demand is calculated by multiplying the group average power demand by the number of consumers belonging to the group, and all the groups Preferably, the provisional total power demand is calculated based on a value obtained by adding the group total power demands.

  In this electricity demand forecasting system, when calculating the provisional total electricity demand, the group average electricity demand is used to calculate the total including the customers with inadequate performance data, such as those who have newly signed an electricity supply contract. Electric power demand can be calculated.

  Moreover, in the power demand prediction system having the above characteristics, it is preferable that the power demand prediction unit sequentially predicts the total power demand in the future unit periods that arrive sequentially one by one at different timings.

  In this power demand prediction system, it is possible to accurately predict the total power demand by always using the latest performance data.

  Moreover, in the power demand prediction system having the above characteristics, the power demand prediction unit selectively extracts the consumers at an equal ratio from each of the groups, and the total based on the extracted actual data of the consumers. Predicting power demand is preferable.

  In this power demand prediction system, it is possible to effectively suppress the deviation of the actual data used for the prediction of the total power demand. Therefore, it is possible to effectively suppress a decrease in the prediction accuracy of the total power demand.

  Moreover, in the power demand prediction system having the above characteristics, it is preferable that the group classification unit reclassifies the consumer at every predetermined timing.

  In this power demand prediction system, fluctuations in power consumption patterns of consumers (for example, fluctuations in power consumption patterns due to fluctuations in family structure and home environment) and fluctuations in consumers (for example, conclusion and cancellation of power supply contracts) It becomes possible to reclassify customers into appropriate groups.

  The power demand forecasting system of the above feature uses the actual data recorded in the database without any bias, and suppresses the decrease in the prediction accuracy of the total power demand, while reducing the number of actual data used for the prediction of the total power demand. It is possible to reduce and realize a quick process. Therefore, the power demand prediction system having the above characteristics can accurately predict the total power demand while rapidly processing a large amount of performance data.

The block diagram shown about the structural example of the electric power demand prediction system which concerns on embodiment of this invention. The figure which shows typically about the classification method of the customer by a group classification | category part. The figure which shows typically about the selection method of the performance data used when an electric power demand prediction part estimates a total electric power demand. The figure shown about the specific operation example of the electric power demand prediction system which concerns on embodiment of this invention.

  Hereinafter, a power demand prediction system according to an embodiment of the present invention will be described with reference to the drawings. In the following, for the sake of concrete description, a power demand prediction system used by a new power company to formulate a demand plan is illustrated as a power demand prediction system according to an embodiment of the present invention.

<Configuration example of power demand forecasting system>
First, a configuration example of a power demand prediction system according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram illustrating a configuration example of a power demand prediction system according to an embodiment of the present invention.

  As shown in FIG. 1, a power demand prediction system 1 according to an embodiment of the present invention includes a performance data acquisition unit 10, a weather data acquisition unit 11, a customer data registration unit 12, a database 13, and a group classification unit. 14 and a power demand prediction unit 15.

  The performance data acquisition unit 10 is composed of a communication device for connecting to a network such as the Internet, for example, and results data indicating the power consumption per predetermined time in each of a plurality of consumers who are customers of the new power company Are acquired sequentially. The actual data is sequentially generated by, for example, a smart meter installed at each consumer. Moreover, the performance data acquisition part 10 receives the performance data transmitted from the smart meter installed in each consumer, or receives the performance data transmitted via a general electric utility from these smart meters. By doing so, the result data is acquired sequentially.

  The meteorological data acquisition unit 11 includes a communication device for connecting to a network such as the Internet, for example, past actual weather information (for example, temperature, weather, wind direction, wind speed, etc .; the same applies hereinafter) and the actual actual data. Meteorological data indicating weather information and further forecasted meteorological information is sequentially acquired. The meteorological data acquisition unit 11 sequentially acquires meteorological data by receiving meteorological data distributed from a predetermined organization (for example, the Japan Meteorological Agency) or receiving meteorological data transmitted from a meteorological observation device. Note that the performance data acquisition unit 10 and the weather data acquisition unit 11 may be configured by the same communication device.

  The customer data registration unit 12 is configured by a personal computer, for example, and is based on information obtained through an operator's operation or a network such as the Internet, and the customer who has signed a power supply contract with a new power company. Customer data indicating various types of information (for example, an ID (identification) number for identifying a consumer, contract details, address, etc.) is generated. In addition, the customer data registration unit 12 not only generates customer data of a customer who has newly signed a power supply contract, but also edits customer data of a customer who has signed a power supply contract in the past (for example, (Cancellation of power supply contract and change of contract details).

  The database 13 is composed of a recording device capable of recording a large amount of data such as a hard disk, for example, and results data acquired via the results data acquisition unit 10 and weather data acquired via the weather data acquisition unit 11. The customer data generated by the customer data registration unit 12 and the group classification data generated by the group classification unit 14 are recorded. The database 13 may record record data, group classification data, and customer data in association with each other.

  The group classification unit 14 includes an arithmetic device such as a CPU (Central Processing Unit) and a storage device such as a semiconductor memory, and operates when the arithmetic device executes a predetermined program. Specifically, the group classification unit 14 classifies consumers into a plurality of groups based on the actual data and customer data read from the database 13. Further, the group classification unit 14 generates group classification data indicating the classification result of the customer.

  Here, an example of a customer classification method by the group classification unit 14 will be described with reference to the drawings. FIG. 2 is a diagram schematically illustrating a consumer classification method by the group classification unit. The white circles shown in FIG. 2 schematically show the customer C.

  As illustrated in FIG. 2, the group classification unit 14 classifies the customer C into a plurality of groups 1 to 4 based on the actual data of the customer C. Specifically, for example, the group classification unit 14 averages the performance data in the past predetermined period (for example, the latest one week, one month, several months) for each customer C, so that the average day Calculate the power consumption pattern. And the group classification | category part 14 classify | categorizes the consumer C with which the calculated power consumption pattern is similar into the same group, and classifies the consumer C with the said similar power consumption pattern into a different group, and the consumer C Are classified into a plurality of groups 1 to 4.

  In FIG. 2, the customer C is divided into a group 1 (detached house type), a group 2 (condominium energy-saving type), a group 3 (small business type), and a group 4 (night activity type). The case where it is classified is illustrated. However, FIG. 2 is merely an example, and it is not always necessary to configure the group classification unit 14 so as to classify the customer C into groups 1 to 4.

  For example, the consumer C is simply unspecified based on the similarity of the power consumption patterns without learning in advance the specific group to be classified and the power consumption patterns (that is, teacher data) of the consumers belonging to the group. The group classification unit 14 may be configured so as to be classified into groups. That is, the group classification unit 14 may be configured to classify the customer C into an unspecified group by unsupervised learning clustering.

  In addition, for example, the group classification unit 14 may calculate an average daily power consumption pattern for each customer C using actual data for all days included in a predetermined period in the past. The average daily power consumption pattern for each customer C is calculated by selectively extracting performance data on days (for example, weekdays, holidays, specific days of the week, etc.) that satisfy specific conditions included in the period. May be. In this case, the group classification unit 14 may generate a plurality of group classification data. Specifically, for example, the group classification unit 14 includes group classification data for weekdays indicating the classification result of the customer C based on the actual data of weekdays and a holiday indicating the classification result of the customer C based on the actual data of holidays. And group classification data for use may be generated separately. In this case, the power demand prediction unit 15 to be described later may select the group classification data to be used according to the day (for example, weekday or holiday) to which the unit period for which the total power demand is predicted belongs.

  Further, the group classification unit 14 sets the customer C at every predetermined timing (for example, every time a predetermined period such as one week or one month elapses, or every time a customer customer changes by a predetermined number). Redo the classification. As a result, fluctuations in the power consumption pattern of the consumer C (for example, fluctuations in the power consumption pattern due to fluctuations in the family structure and home environment), increase / decrease in the consumer C (for example, conclusion or cancellation of the power supply contract), etc. Following this, it becomes possible to reclassify consumers into appropriate groups.

  The power demand prediction unit 15 includes, for example, an arithmetic device such as a CPU and a storage device such as a semiconductor memory, and operates when the arithmetic device executes a predetermined program. Specifically, the power demand prediction unit 15 predicts the total power demand in the future unit periods that come sequentially based on the performance data, customer data, group classification data, and weather data read from the database 13. Furthermore, the power demand prediction unit 15 generates demand plan data indicating the predicted total power demand (ie, demand plan).

  The demand plan data is transmitted to a general electric utility via a predetermined communication device. As a result, a demand plan is submitted from the new electric power company to the general electric company. In order to supply the transmission and distribution network with the same amount of power as the total power demand indicated by the demand plan submitted to the general electric utility, the demand plan data is also available for the power generation facilities managed by the new electric utility. You may comprise so that it may be transmitted. Moreover, the group classification | category part 14 and the electric power demand prediction part 15 may be comprised with the same arithmetic unit and the memory | storage device.

  By the way, the electric power demand in the future unit period of each consumer can be predicted using the performance data of each consumer. Furthermore, the total power demand in the unit period can be calculated by adding up the power demands in the unit period of all consumers. However, since the record data recorded in the database 13 is enormous, problems such as time-consuming processing may occur.

  Therefore, the power demand prediction unit 15 enables a rapid process while maintaining the prediction accuracy of the total power demand by limiting the actual data to be used when predicting the total power demand. A method of processing actual data by the power demand prediction unit 15 will be described with reference to the drawings. FIG. 3 is a diagram schematically illustrating a method of selecting actual data used when the power demand prediction unit predicts the total power demand.

  As shown in FIG. 3, the power demand prediction unit 15 refers to the group classification data, grasps the groups 1 to 4 to which the customer C belongs, and then selects some of the consumers SC from each of the groups 1 to 4. (Black circles in the figure) are selectively extracted. At this time, it is preferable that the power demand prediction unit 15 selectively extracts the consumers SC from each of the groups 1 to 4 at an equal ratio (for example, 20%). And the electric power demand prediction part 15 estimates a total electric power demand using the performance data of some extracted consumers SC.

  As described above, in the power demand prediction system 1 according to the embodiment of the present invention, the power demand prediction unit 15 uses the performance data of the customer SC selectively extracted from each of the groups 1 to 4, Predict the total power demand for a unit period. For this reason, the power demand prediction system 1 uses the actual data recorded in the database 13 without bias to suppress the decrease in the prediction accuracy of the total power demand, and the number of the actual data used for the prediction of the total power demand. It is possible to realize a rapid process by reducing. Therefore, the power demand prediction system 1 can accurately predict the total power demand while rapidly processing a large amount of performance data.

  Moreover, the electric power demand prediction part 15 suppresses the bias | inclination of the performance data used for prediction of a total electric power demand effectively by selectively extracting the consumer SC from each of the groups 1-4 at an equal ratio. Can do. Therefore, it is possible to effectively suppress a decrease in the prediction accuracy of the total power demand.

  In addition, the electric power demand prediction part 15 may select the consumer SC at random in each group 1-4. In this case, the power demand prediction unit 15 may select the customer SC at the time of predicting the total power demand, or select the customer SC in advance before the time of predicting the total power demand. Also good.

  Further, the power demand prediction unit 15 may select the customer SC based on the customer data or the actual data. In this case, it is preferable that the power demand prediction unit 15 selects the customer SC so that the bias of the actual data used for the prediction of the total power demand is reduced. For example, the power demand prediction unit 15 may preferentially select the customer SC that is close to the average power consumption pattern of the customer C that constitutes the group in each of the groups 1 to 4, and the actual data is The consumer SC may be selected so that the scale of the power supply contract indicated by the amount of power consumption to be indicated and the consumer data varies. In this case, if the power demand prediction unit 15 can quickly select the customer SC, the customer SC may be selected when the total power demand is predicted. If a long time is required for the selection, it is preferable to select the customer SC in advance before the time when the total power demand is predicted.

<Specific operation example of power demand forecasting system>
Next, a specific operation example of the power demand prediction system 1 according to the embodiment of the present invention will be described with reference to the drawings. FIG. 4 is a diagram illustrating a specific operation example of the power demand prediction system according to the embodiment of the present invention.

  FIG. 4 shows a specific operation example of the power demand prediction system 1 when the total power demand of three consecutive unit periods T1 to T3 is predicted in order. In FIG. 4, the time length of the unit periods T1 to T3 is 30 minutes, and the time interval of the timing (arrow D in the figure) at which the result data acquisition unit 10 acquires the result data and records it in the database 13 is 30 minutes. It is assumed that the time interval of the timing (arrow W in the figure) when the weather data acquisition unit 11 acquires weather data and records it in the database 13 is 3 hours.

  Further, in the example shown in FIG. 4, a day before the day when the new electric power company supplies electric power to the consumer as a customer (hereinafter referred to as “power supply date”) Assuming a case where a provisional demand plan showing a prediction result of provisional total power demand (hereinafter referred to as “provisional total power demand”) in each of the 48 unit periods forming the power supply date is submitted by a predetermined time L1. doing. In the case of this example, the power demand prediction unit 15 generates provisional demand plan data indicating provisional total power demand. The temporary demand plan data is transmitted to a general electric power company (further, a power generation facility managed by a new electric power company) via a predetermined communication device in the same manner as the demand plan data.

  Further, in the example shown in FIG. 4, the new electric power company is the last in the unit periods T1 to T3 by the time L21 to L23 that is one hour before the unit periods T1 to T3 are started. It is assumed that a demand plan showing the predicted result of total power demand is submitted.

First, the power demand prediction unit 15 predicts provisional total power demand in each of the 48 unit periods forming the power supply date including the unit periods T1 to T3 in the first period P1 before the time L1. Specifically, the calculation shown in the following formula (1) is performed. In the following formula (1), Z _t is the provisional total power demand in the unit period t, K is an integer of 2 or more and the total number of groups into which the consumers are classified, and Z (k) _t is the group in the unit period t. group average power demand, which is the average power demand of k, N (k) is the number of consumers belonging to group k, τ (i, j) _t is the weather of unit period t in the prediction of provisional total power demand Z _t It is an adjustment coefficient that is a coefficient for reflecting information (temperature i and weather j). Z (k) _t × N (k) is a group total power demand that is the total power demand in the unit period t of consumers belonging to the group k.

For example, the group average power demand Z (k) _t can be calculated by averaging the power consumption indicated by the performance data of the consumers belonging to the group k for each unit period. At this time, as illustrated in FIG. 3, the group average power demand Z (k) _t is calculated using the actual data of the consumers SC selectively extracted from the consumers C belonging to each group. Also good. In addition, since the provisional total power demand only needs to be predicted once a day, the first period P1 can be set longer. Therefore, the group average power demand Z (k) _t may be calculated using the performance data of all the consumers C belonging to the group k.

Further, for example, the adjustment coefficient τ (i, j) _t is the past temperature and weather (sunny, rainy, cloudy, etc.) indicated by the weather data recorded in the database 13 and the consumer's power consumption indicated by the actual data. A function τ representing these correlations is calculated by performing regression analysis or the like on the basis of the meteorological data indicating the weather information of the future prediction recorded in the database 13 most recently. The temperature i and the weather j can be obtained and applied to the function τ. Here, the variable of the function τ for calculating the adjustment coefficient τ (i, j) _t is exemplified by the temperature and the weather, but in addition to (or instead of) other weather information as a variable It may be used. Moreover, you may employ | adopt the weather information with a strong correlation with the power consumption specified by performing regression analysis etc. of various weather information and power consumption as a variable of function (tau).

Moreover, as illustrated in FIG. 2 and FIG. 3, an arithmetic expression for calculating the provisional total power demand Z _t of the unit period t when the customer C is classified into four groups 1 to 4 is as follows. It shows in Formula (2). In the following formula (2), the total number of customers C who are customers of the new power company is 50,000, the number of customers belonging to Group 1 is 20000, and the number of consumers belonging to Group 2 is 15000. The number of customers belonging to the eaves and the group 3 is 10,000, and the number of customers belonging to the group 4 is 5,000.

  Next, the power demand prediction unit 15 predicts the total power demand in the unit period T1 in the second period P21 before the time L21. Thereafter, the power demand prediction unit 15 predicts the total power demand in the unit period T2 in the second period P22 before the time L22. Thereafter, the power demand prediction unit 15 predicts the total power demand in the unit period T3 in the second period P23 before the time L23.

When the power demand prediction unit 15 predicts the total power demand for each of the unit periods T1 to T3, the calculation shown in the following formula (3) is performed. In the following formula (3), H _t is the total power demand for the unit period t, Z _t is the provisional total power demand for the unit period t, and h _t is the provisional total power demand Z _t for the unit period _t and the actual unit period. It is a correction value that is a value corresponding to the difference from the power demand of t.

As shown in the above equation (3), the correction value h _t is a value for correcting the provisional total power demand Z _t predicted in the past so as to be brought close to the actual power demand after the fact. For example, the correction value h _t is the previous starting point at the start after and the second period P21~P23 provisional total power demand Z _t consumer actual data which could not have been acquired at the time of predicting the (first period P1 It can be calculated using at least a part of the actual data recorded in the database 13. Specifically, for example, the difference between the power consumption indicated by the actual data obtained on the power supply date and the provisional total power demand predicted with respect to the time when the power consumption indicated by the actual data is measured. based on the sheath variation trend, it is possible to calculate the correction value h _t of the unit period t. Incidentally, when calculating the correction value h _t, provisional total power demand Z _t meteorological data which could not have been acquired at the time of predicting the (beginning after the first period P1 and the start time earlier second period P21~P23 The weather information (for example, the latest or current actual weather information or the weather information predicted in the unit periods T1 to T3) indicated by at least a part of the weather data recorded in the database 13 may be used.

As shown in FIG. 4, it is necessary to predict the total power demand a plurality of times a day (48 times in the case of this example where the unit period has a time length of 30 minutes). Therefore, it is difficult to set the second period P21 to 23 long. Therefore, the power demand prediction unit 15 uses the actual data for consumer SC was selectively extracted from the customer C that belong to each group, as shown in FIG. 3, to calculate the correction value h _t.

  Thus, by calculating the total power demand by correcting the provisional total power demand after the fact, it is possible to simplify the calculation when calculating the total power demand. As a result, the total power demand can be calculated quickly.

  In addition, when calculating the provisional total power demand, the group average power demand is used to calculate the total power demand, including those with inadequate performance data, such as customers who have newly signed a power supply contract. It becomes possible. In this case, a customer for which there is no actual data may be tentatively treated as belonging to one of the groups (for example, the group having the largest number of customers belonging), You may treat as what belongs to the same group as the consumer whose contents approximate. However, such a customer with insufficient performance data is not extracted when the power demand prediction unit 15 calculates the total power demand (so as not to correspond to the customer SC shown in FIG. 3). ,preferable.

<Deformation, etc.>
[1] The specific operation example of the power demand prediction system 1 illustrated in FIG. 4 is to calculate the provisional total power demand for the unit periods T1 to T3, and then correct this to calculate the total power for the unit periods T1 to T3. Although the demand is calculated, the total power demand may be directly calculated without calculating the provisional total power demand. In this case, the power demand prediction unit 15 uses the actual data of the consumers SC selectively extracted from the consumers C belonging to each group as illustrated in FIG. The total power demand may be calculated by performing the same calculation as.

  Furthermore, in this case, the power demand prediction unit 15 calculates only the total power demand for one unit period in the period corresponding to the second period illustrated in FIG. 4 (that is, calculates the total power demand one by one at different timings). Prediction), or the total power demand of a plurality of unit periods may be calculated simultaneously.

  However, the accuracy of the total power demand can be improved by calculating using the latest performance data. For this reason, the total power demand is preferably calculated immediately before the arrival of the unit period. In particular, as illustrated in FIG. 4, it is preferable to calculate only the total power demand for one unit period in one second period, because the total power demand can always be predicted using the latest actual data.

[2] In addition, the specific operation example of the power demand prediction system 1 illustrated in FIG. 4 calculates the temporary total power demand for 48 unit periods corresponding to one day in the first period P1, and The total power demand for one unit period T1 to T3 is calculated in the second period P21 to P23, but the number of unit periods for predicting the provisional total power demand in the first period and the total power in one second period. The number of unit periods for forecasting demand is not limited to this example.

  For example, in the first period, provisional total power demand for 24 unit periods corresponding to half a day may be calculated, or provisional total power demand for 16 unit periods corresponding to 1/3 day may be calculated. Alternatively, provisional total power demand for 96 unit periods corresponding to two days may be calculated. Further, for example, in one second period, the total power demand of two unit periods may be calculated, the total power demand of three unit periods may be calculated, or the total power demand of four unit periods May be calculated.

  However, the accuracy of the total power demand can be improved by calculating using the latest performance data. For this reason, the total power demand is preferably calculated immediately before the arrival of the unit period. In particular, as illustrated in FIG. 4, calculating the total power demand for one unit period in one second period is preferable because the total power demand can always be predicted using the latest performance data. In addition, provisional total power demand only needs to ensure a certain degree of prediction accuracy. Therefore, from the viewpoint of reducing the load of the power demand prediction unit 15, it is preferable to collectively estimate the provisional total power demand for a plurality of unit periods.

  Further, due to the difference between the total power demand and the provisional total power demand as described above, the number of unit periods in which the power demand prediction unit 15 predicts the total power demand in the second period is determined by the power demand prediction unit 15 in the first period. It is preferable to make the number less than the number of unit periods for which the provisional total power demand is predicted.

[3] The present invention is not limited to the power demand prediction system 1 according to the embodiment described above. For example, the present invention may be implemented as any power demand forecasting system for forecasting total power demand, such as a power demand forecasting system used by a general electric utility to forecast the total power demand of consumers in the jurisdiction. It can be done.

  The power demand prediction system of the present invention can be suitably used for, for example, a power demand prediction system used by a new power company to formulate a demand plan.

DESCRIPTION OF SYMBOLS 1: Power demand prediction system 10: Performance data acquisition part 11: Weather data acquisition part 12: Customer data registration part 13: Database 14: Group classification part 15: Electric power demand prediction part P1: First period P21-P23: Second Period T1-T3: Unit period

Claims (7)

A database in which performance data indicating power consumption per predetermined time in each of a plurality of consumers is sequentially recorded;
A group classification unit for classifying the consumers into a plurality of groups based on the performance data;
Based on the performance data, the total power demand, which is the sum of the power demands of the consumers in the future unit period of the predetermined time length that sequentially arrives, at each timing before the start time of the unit period A power demand forecasting unit that sequentially forecasts
The power demand prediction unit selectively extracts a part of the consumers from each of the groups, and predicts the total power demand based on the extracted performance data of the consumers. Demand forecasting system. The power demand prediction unit
In the first period, a provisional total power demand that is the sum of the power demands of the consumers in the unit period is predicted,
The second part of the consumer selectively extracted from each of the groups in the second period starting after the start time of the first period and ending before the start time of the unit period. The total of the unit periods is corrected by correcting the provisional total power demand based on at least a part of the performance data recorded in the database after the start time of one period and before the start time of the second period. The power demand prediction system according to claim 1, wherein power demand is predicted. The power demand prediction unit
Predicting the provisional total power demand for a plurality of the unit periods in the first period;
Predicting the total power demand for at least one of the unit periods in the second period;
The number of the unit periods for which the total power demand is predicted in the second period is smaller than the number of the unit periods for which the provisional total power demand is predicted in the first period. Power demand forecasting system. The power demand prediction unit
Based on the actual data of the consumers belonging to the group, a group average power demand that is an average power demand in the unit period of the consumers belonging to the group is calculated for each group,
For each group, calculate the group total power demand by multiplying the group average power demand by the number of consumers belonging to the group,
The power demand prediction system according to claim 2 or 3, wherein the provisional total power demand is calculated based on a value obtained by adding the group total power demands of all the groups.   5. The power according to claim 1, wherein the power demand prediction unit sequentially predicts the total power demand of the future unit periods that arrive sequentially one by one at different timings. Demand forecasting system.   The power demand prediction unit selectively extracts the consumers at an equal ratio from each of the groups, and predicts the total power demand based on the extracted actual data of the consumers. The power demand prediction system according to any one of Items 1 to 5.   The power demand prediction system according to any one of claims 1 to 6, wherein the group classification unit reclassifies the consumer at every predetermined timing.

Images