JPS60102822A - Method of predicting power demand amount - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a power demand forecasting method for creating a power demand forecast curve used for operation planning of a power generation system.

[Background of the invention]

Prediction of power demand is usually carried out every day at the central power dispatch center of the power system in order to balance the five essential power sources and supply. Conventionally, as a prediction method, an operator at a command center inputs the following day's weather, maximum and minimum temperatures, day of the week, and other factors that change the amount of electric power, and then uses a computer to predict the next day's forecast while referring to past performance data. A method is being used to predict changes in demand over a 24-hour period. As a specific prediction method, a method is known that searches for and uses similar past actual power demand curves.

However, the conventional methods described above are dependent on the experience and proficiency of the operator, and are difficult for the operator to understand. The problem was that the forecast did not match the actual situation because the increase was not properly reflected.

[Purpose of the invention]

An object of the present invention is to provide a method for predicting power demand amount that is easy for an operator to understand and allows highly accurate prediction of power demand M:6(1).

[Summary of the invention]

The present invention provides a performance data table in which the form of the variation factor of 'electricity'1', +3 requirement is a column element and the demand quantity is a row element, and a day is divided into a set number of time periods, and each time period is The actual value of electric power demand is stored as an occurrence frequency in the corresponding element of the actual performance data table, and an occurrence frequency vector consisting of column elements corresponding to the time period to be predicted and the aspect of the predicted variation factor in that time period is stored. On the basis of the,
The power demand amount corresponding to the maximum frequency in the time period to be predicted is set as the predicted value, and the prediction is easy for the operator to use and has high accuracy.

That is, the present invention is structured so that it is easy to use and can obtain highly accurate predicted values when making predictions when organizing past performance figures of electricity demand, and this performance data table is It is characterized by the predictive calculations performed using the method.

[Embodiments of the invention]

Hereinafter, the present invention will be explained based on examples. First, a method for creating a performance data table will be explained below. 1st
As shown in the figure, the performance data table 1 includes time periods divided into fixed time periods (for example, every hour) and variable factors (for example, weather, temperature,
day of the week, etc.). Each performance data table 1 is
It has a matrix configuration, and the row elements have unit power demand ΔX.
The mental power demand level Xi (i=1.2゜・
), the column elements correspond to the aspects of each variation factor, respectively.

Specific examples of the performance data table 1 are shown in FIGS. 2(A), (13), and (C). The performance data table 1a shown in FIG. 2(A) uses the weather W as a variable factor, and the mode we is, for example, WI: sunny, W2: cloudy, W3
: Light rain, W4: Rain, W5: Snow. In Figure 2 (B), 7J< Actual data table 1b shows the temperature T
is a variable factor, and its aspect Tn1 is flJ
For example, Tl: U~5Cs '1'2: 5~10t
Z'XT3: 10~15tll'x '1'4:
15~20C1T5:20~25C1T6:25
-3°C, respectively. The actual data table IC shown in FIG.
February 11i1st, D2: Tue fa Sun, ]) 3: Wednesday, 1) 4: Thursday 11N1st, D5: Fri u wa Sun, 1) 6
: Saturday 11+l, D7: Set for fEt eyebrow 8 and holiday, respectively. In addition, the power demand level Xi, which is a row element of each actual cheetah table shown in FIG. 2 (A) to (Q), is set as shown in the following equation (1) with respect to the standard power demand Xo. ing.

For example, to give a specific numerical example, Xo=100QMW
When ΔX=100MW, X1=1000MW
~I100MW.

X2=1100MW~1200MW.

X1o=1900 MW to 2000 MW is set.

The matrix elements A and B corresponding to the weather, average temperature, and time of the actual data table 1 created in this way are
, C, each is expressed by the following equation (2).

Each element of the performance data table 1 set in this way is cleared to zero in the initial state, and thereafter cleared to 1 in accordance with the aspect of the variation factor corresponding to each matrix element and the power demand level.
1 eagle cuff) Every time a δ required achievement value occurs, the content is +
1 is added. That is, the contents of each matrix element indicate the frequency of occurrence in the past. For example, to give a concrete example, yesterday's performance of time Iku T+ is 1250MνXl, that is, X3
．． Assuming that the weather is Earthquake = W2, L temperature is 18C: 'T4, and the day of the week is Tuesday - D2, matrix element a32 + b
1 is added to each of the contents of B4 + C32.

In this way, the power for each time period is +9r7.1.
The performance data table is revised sequentially based on the performance values to improve the reliability of the performance data.

In the case of a variable factor that changes continuously, such as the average temperature, it is necessary to not only revise only the matrix element corresponding to the actual value, but also to add 1 to each adjacent column element. Reliability can be further improved by adding appropriate weights.Next, the procedure for creating the next day's demand forecast curve using the performance data table 1 created as described above will be explained. First, the operator predicts the next day's variable factors based on the information obtained. For example, the next day's weather is light rain: W3,
Assuming that the average temperature is 17'c: T4 and the day of the week is Tuesday = D2, first, from the occurrence frequency vector table shown in the following equations (3) to (5), which is composed of column element vectors corresponding to the aspects of these fluctuation factors. Take it out.

Next, the vectors shown in the previous equations (3) to (5) are normalized by dividing them by the sum of each vector element, and the following equation (6) is obtained.
~ Random variable vector W3+ T< + D shown in (8)
Get 2.

Next, assuming that each element of each variation factor is independent, the elements of each occurrence frequency vector are multiplied together to obtain a total vector P shown in the following equation (9).

・・・・・・・・・(9) Furthermore, equation (9) is normalized and the total random variable P is expressed as the following equation (1
0) Fill in the formula.

This comprehensive random variable V is a random variable p that integrates variable factors such as weather, temperature, and day of the week, and the expected value Lt of the predicted value of electricity demand for the time period to be predicted from now on is expressed by the circle of the following equation. Become.

By performing the above calculation for each time period 11''t24, a predicted power demand curve for the next day -01 can be obtained.

The above prediction calculation procedure is explained using the same 1>j (7) communal numerical system as described above. If so, each power iM of the actual data table corresponding to the pattern W3 of the sky pattern Required level X
The number of occurrences of l is i =
1 (1000~1100MW) three times, 1=2(1
100~1200MW) 5 times, 1=3(1200~1
300MW) has appeared seven times, the occurrence frequency vector W3 is expressed by the following equation.

In addition, the random variable vector w3 obtained in this way as shown in the quadratic equation of the random variable vector Wst is shown in FIG. 3 (a
) as shown in X5, i.e. 1400~
This shows that the probability of reaching the power demand level of 1500 MW is highest. Note that the sum of the hatched parts in the figure is 1 because it is a random variable.

Similarly, the random variable vectors T4 and D2 are shown in Figure 3(b) and (C), respectively.When the average temperature is 17C, the probability is that the power will be 1300MW to 1400MW, and on Tuesday it will be 1000Mw to 14ooMw. This indicates that the probability of Furthermore, from the curves shown, it can be seen that there is no unevenness in the distribution of the power demand level Xi with respect to the days of the week.

The total random variable vector P obtained from
It can be seen that the probability is the highest. By determining the power demand level that indicates the maximum probability for each time period TI-'l''24 in this way, a prediction curve 147 of the power demand and demand as shown in FIG. 4 is obtained.

In the above prediction calculation, each variation factor was assumed to be independent, but if, for example, two variation factors are in a dependent relationship, they should be weighted to create a composite vector of them as described below. It is desirable to do so. In other words,
For variable itb factors such as weather and average temperature, which are in a dependent relationship with each other, for example, if the specific weight of weather is dogs and its weighting coefficient μ is created, a vector (μW3+'I' 4 ) is created and this is Normalization may be performed to create a random variable vector, and this may be used as a random variable vector for one variation factor to perform predictive calculations in the same manner as described above. This further improves prediction accuracy.

As described above, according to this embodiment, the actual demand values corresponding to each time period, each variable factor, and each aspect are classified by power demand level, and the frequency of occurrence is stored. Based on this actual data table, we select only the data corresponding to specific variable factors to perform predictive calculations, making it easy for operators to understand and This has the effect of making it possible to predict demand with high accuracy.

It is known that the amount of electricity demanded increases year by year with economic growth. The rate of increase in electricity demand is said to be close to the economic growth rate of the country. Therefore, by modifying the contents of the performance data table 1 in the embodiment according to the growth rate, the accuracy of the prediction can be improved. In other words, when the annual demand increase rate is α,
Electricity demand level Xi, which is a row element of the actual data table
By making adjustments to increase each amount according to α, it becomes possible to forecast demand in consideration of the annual growth rate. In addition, since the prediction accuracy is improved by increasing the frequency of correction as much as possible, the following formula (12
Correct the power demand level Xt by 1 respectively.

Similarly, the equation circle is modified as shown in the following equation (15).

(13) With such a simple modification, it is possible to modify the power demand in line with economic growth.

In addition, since the above formula <121 and ([3) are repeated every N days, strictly speaking, the following formula (14) is applied to α.
However, in practice, by using α, the number of calculations becomes only 1Ili.

In addition, in the above example, the weather, average temperature, and day of the week were applied as variable factors, but this
For example, the season, month, daylight hours, etc.
By applying as many variable factors as possible, such as maximum temperature, minimum temperature, or social events, it is possible to predict power demand more accurately.

FIG. 5 shows a configuration diagram of an embodiment apparatus to which the method of the present invention is applied. As shown in FIG. 5, the actual values of temperature and demand are sequentially input to the actual data table creation device 12 via the data input device 11. In addition, the performance data table creation device 12 includes a keyboard 13,
Variable factors such as the weather and day of the week are input. Actual data table creation device 1
In step 2, a signal for revising the contents of the corresponding matrix element is sent to the performance data table storage device 14 to change the contents based on the input data and the variation factor as described above. ing. The electric power demand forecast calculation device 15 calculates, based on the condition of the variation factor in the time period to be predicted, which is input from the keyboard 13,
The above-mentioned actual data table storage device 14 is referred to, necessary data is imported, the above-mentioned prediction@] calculation is executed, data corresponding to the demand quantity forecast curve is sent to the display device 16, and the demand is forecasted. It is designed to display curves. Incidentally, the apparatus shown in FIG. 5 can be easily realized by applying a computer.

〔Effect of the invention〕

As described above, according to the present invention, there is an effect that it is easy for an operator to use and it is possible to predict the power requirement of 1jJ with high accuracy.

[Brief explanation of the drawing]

FIG. 1 is a diagram for explaining the concept of a performance data table according to an embodiment of the present invention, and FIGS. 2 (A) and (B). (C) is a configuration diagram of a performance data table according to an embodiment of the present invention, FIG. 3 is a diagram showing each random variable vector according to an embodiment of the present invention, and FIG. 4 is an example obtained by the present invention. FIG. 5 is a diagram showing a predicted curve of demand for one power supply, and FIG. 5 is a block diagram of an apparatus to which the method of the present invention is applied. Agent Patent Attorney Tatsuyuki Unuma 3 [1 Kaya 4 mouth t, tz i6 ttz t+8624$5 hard/l

Claims (1)

[Claims] 1. Divide one day into multiple time periods and record the actual value of the electricity demand in each time period in the corresponding performance data table in which the variation factor of the demand is the column element and the demand ° is the row element. The occurrence frequency is stored in the element as an occurrence frequency, and based on the occurrence frequency vector of the demand quantity in the actual data table described in ^1, which corresponds to the time period to be predicted and the aspect of the prediction variation factor in that time period, the relevant prediction should be made. A power demand forecasting method that uses the power demand corresponding to the maximum frequency in a time period as the predicted value. 2. In the invention as set forth in claim 1, the fluctuation factors of the demand amount include weather, average temperature, and day of the week. A power demand forecasting method comprising at least one of seasonal sunshine hours, maximum temperature, minimum temperature, and social events. 3. In the invention set forth in claim 1 or 2, the amount of power demand as a row element of the performance data table is corrected based on the annual growth rate of the amount of power demand. A power demand forecasting method characterized by the following.