JP2712688B2 - Demand forecasting device for water distribution - Google Patents

Description

The present invention relates to an apparatus for estimating a demand for water distribution in a water distribution system of a water supply system.

B. Summary of the Invention The present invention provides a device for predicting the demand for water distribution in a predetermined time unit, and calculates a weather correction coefficient based on weather data,
If the forecast date is before or after a holiday with reference to the calendar, a holiday correction coefficient is obtained, and the predicted water distribution is calculated by correcting the past average water distribution using the obtained weather correction coefficient and the weather correction coefficient. If the forecast date is a holiday, correction will be made by delaying in the morning, and the calculation rules for the weather correction coefficient and holiday correction coefficient and the degree of delay when the forecast date is a holiday will be corrected.・ By making changes, the know-how of operators and the uniqueness of the region can be easily reflected in the prediction results.

C. Prior Art In general, water distribution is predicted in a distribution system when performing water distribution control such as water level management of a reservoir, water distribution pump control, and motorized valve control. This forecast includes a demand forecast in units of hours and a short-term demand forecast.

 FIG. 16 shows a conventional water distribution amount prediction process.

Collect water distribution data from the distribution system, and collect data on days of the week, holidays, weather, temperature, etc. Then, the data in the past predetermined period is analyzed, and the prediction calculation is performed from the data such as the day of the week of the prediction day. That is, the undetermined coefficient of the prediction calculation formula is determined so that the error of the entire predicted value is minimized, and the hourly demand forecast and the daily demand forecast are performed from the result.

The distribution system was controlled based on the hourly demand and daily demand thus obtained.

D. Problems to be Solved by the Invention However, in the above-described conventional technology, since prediction is performed by various prediction calculation methods, it is difficult to reflect intuition and know-how based on the experience of the operator in the prediction result.

Further, without special knowledge of a computer, it has been difficult to construct or maintain prediction rules and algorithms.

Furthermore, there was a problem that it was difficult to package general-purpose software because it was not possible to freely incorporate features unique to the region or region into the demand forecasting algorithm.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a device that can easily change a prediction rule and the like and that can predict a water distribution amount in consideration of an operator's know-how and regional uniqueness.

E. Means for Solving the Problems In order to achieve the above object, the present invention provides a water distribution demand forecasting device that predicts a demand for water distribution in a predetermined time unit, provided with the following means. .

Water distribution data collection unit that collects water distribution data from the water distribution system.

A water distribution data storage unit that stores past water distribution data.

A weather data input unit to which the weather data of the forecast day is input.
“Weather data” refers to data indicating weather conditions such as the weather and temperature of the day.

A weather correction coefficient calculation unit that performs a calculation based on weather data and obtains a weather correction coefficient indicating a change in demand for water distribution amount due to weather conditions.

A calendar storage unit that stores a calendar.

A holiday correction coefficient calculation unit that performs calculations with reference to the calendar and calculates a holiday correction coefficient indicating a change in demand for water distribution when the predicted date is before or after a holiday.

A water distribution amount prediction unit that calculates an average water distribution amount from past water distribution amount data, and corrects the average water distribution amount based on a weather correction coefficient and a holiday correction coefficient to obtain a predicted water distribution amount.

A delay correction unit that performs an operation with reference to the calendar and delays the predicted water distribution amount in a time zone centered on the morning when the predicted date is a holiday. Here, "delay"
It is to correct the water distribution amount of each time so that the curve of the water distribution amount in the predetermined time zone is shifted in the delay direction, for example, for each time, the intermediate value of the water distribution amount of that time and the water distribution amount of the previous time Thus, it is possible to take an aspect in which the water distribution amount after correction is used.

F. Action In the water demand forecasting device according to the present invention, the water distribution data from the water distribution system is collected by the water distribution data collection unit,
Over a predetermined period in the past, for example, water distribution data in hourly units is stored in the water distribution storage unit. Then, a past average water distribution amount is obtained from the data, and a predetermined correction is made to the average water distribution amount, thereby predicting the demand of the water distribution amount on the prediction day.

In making a prediction, first, weather data such as the predicted weather and the predicted maximum temperature on the prediction day is input by the weather data input unit.

Using this weather data, a weather correction coefficient calculation unit performs a calculation according to a predetermined calculation rule. That is, the weather correction coefficient is determined in consideration of the influence of the expected weather and the expected maximum temperature on the demand for water distribution.

Further, based on a calendar stored in the calendar storage unit in advance, the holiday correction coefficient calculation unit determines whether the predicted date is before or after the holiday, and if the predicted date corresponds to this, a predetermined calculation rule The holiday correction coefficient is obtained according to the following.

The water distribution amount prediction unit performs an operation based on the weather correction coefficient thus obtained (when the holiday correction coefficient is obtained, the weather correction coefficient and the holiday correction coefficient), and corrects the past average water distribution amount. Obtain the estimated water distribution.

Further, when the predicted date is a holiday, the delay correction unit performs an operation to compensate for the tendency of the morning water distribution curve to be delayed, and delays the predicted water distribution in a time zone centered on the morning.

Since the predicted water distribution is calculated by the above calculation, the degree to which the influence of weather and temperature is taken into account in the calculation of the weather correction coefficient, the degree to take into account the influence of the holiday in the calculation of the holiday correction coefficient, and the forecast date is In this case, the degree of delay and the like can be easily changed.

That is, it is possible to easily correct and change the handling of each factor such as weather data in the prediction of the water distribution amount.

G. Examples Hereinafter, examples of the present invention will be described with reference to the drawings.

FIG. 1 shows a water distribution demand forecasting apparatus according to one embodiment of the present invention.

The water distribution data collection unit 1 detects and collects water distribution data from the water distribution system 2. The collected water distribution amount data is stored in the water distribution amount data file 3.

Predicted water distribution calculation unit 4, based on the meteorological correction factor K ₁ holiday correction coefficient K ₂ to be described later, by correcting the average distribution amount data _t, and requests the predicted water distribution amount Q _t.

The data input unit 5 is for inputting the predicted weather and the predicted maximum temperature on the predicted date. The expected weather / expected maximum temperature data file 6 stores data such as expected weather in a predetermined period in the past.

Meteorological correction coefficient fuzzy inference section 7 by performing a fuzzy inference based on the expected weather, the expected maximum temperature data, and requests meteorological correction coefficient K _1. The weather correction factor K ₁ is a value indicating the effect of weather conditions such as weather and temperature has on the water distribution amount.

The calendar file 8 stores a calendar for a predetermined period.

The calendar input unit 9 is for setting a calendar by inputting a holiday such as a Sunday or a public holiday or a special day.

Holidays correction coefficient fuzzy inference section 10 refers to the calendar, and requests holidays correction factor K ₂ from the relationship between holiday predicted day. Holidays correction factor K ₂ in the case before and after the predicted day or predicted day is a holiday, it is a value indicating the influence of the water distribution amount.

The time delay correction unit 11 performs time delay correction when the prediction date is a holiday.

The display unit 12 displays the obtained estimated water distribution amount.

 Next, the operation of this device will be described.

The distribution amount data collection unit 1 fetches distribution amount data from the distribution system 2 and creates a distribution amount data file 3. Water distribution amount data Q _t the device is handled is 1 hour (t-1 to t time) distribution amount per the day.

 FIG. 2 shows average water distribution data.

The water distribution amount data file 3 has water distribution amount data per hour on the past n days (n is an arbitrary numerical value of about 20 to 60). From this data, the average water distribution data can be obtained by the following equation. Here, _t is the average amount of water distribution from (t-1) to t.

Next, the calculation of the predicted water distribution amount Q _t in the prediction distribution calculation unit 4.

Predicted water distribution amount Q _t is the weather correction factor K ₁ holiday correction coefficient K _2, as shown in the following equation, by correcting the average distribution amount data _t, obtains a predicted water distribution amount Q _t.

Q _t = (1 + K ₁ + K ₂ ) × _t (2) Next, the calculation of the weather correction coefficient K ₁ performed by the weather correction coefficient fuzzy inference unit 7 will be described.

In obtaining weather correction factor K _1, first, the weather coefficient
Determine the K _3.

Fluctuations in demand for water distribution are affected not only by the weather of the day, but also by the weather the day before and two days before. Weather factor K ₃ is a value for evaluating the weather conditions for several days, it is determined from the value K ₅ Metropolitan showing the weather changes and numbers K ₄ showing the expected weather.

Expected weather K ₄ is input a numerical value 0 to 2.0. Weather change K
₅ can be obtained from the expected weather K ₄ by the following equation. Where K ₄ (0) is the expected weather on the forecast date, K ₄ (-1) is the expected weather on the day before the forecast date, K ₄ (-2) is the expected weather on the day before the forecast date, and α is the coefficient is there.

K ₅ = K ₄ (0) − ｛K ₄ (−1) · α + K ₄ (−2) · (1−
α)｝ (3) FIG. 3 shows a fuzzy rule used in inferring a weather coefficient. These fuzzy rules are nine rules in IF-THEN format. For example, if the expected weather is rain and the weather change is NB (small), the weather coefficient is NB (small).

FIG. 4 shows the membership function used in inferring the weather factor. (A) shows the predicted weather, (b) shows the weather change, and (c) shows the membership function of the weather coefficient.

Weather factor K ₃ is expected weather and K ₄ and the weather change K ₅ the condition part can be determined by a technique such as the maximum minimization method using fuzzy rules and membership functions described above.

Next, the weather correction coefficient fuzzy inference unit 7 calculates the weather coefficient K ₃
Use expected maximum temperature change K ₆ and obtains a weather change coefficient K _7. The weather change coefficient K ₇ is a value for evaluating the change in weather conditions and temperature.

It expected maximum temperature change K ₆ described above can be obtained by the following equation. Where K ₈ is the expected maximum temperature on the forecast day, K
₈ (n) is the average maximum temperature for the past n days.

K ₆ = K ₈ −K ₈ (n) (4) FIG. 5 shows a fuzzy rule used in inferring a weather change coefficient. This figure also shows nine fuzzy rules, as in FIG.

FIG. 6 shows a membership function used in inferring a weather change coefficient. (A) shows the weather coefficient, (b) shows the predicted maximum temperature coefficient, and (c) shows the weather change coefficient.

A weather factor K ₃ the expected maximum temperature change K ₆ a condition part, using fuzzy rules and membership functions described above, it is possible to infer weather change coefficient K _7.

Finally, weather correction coefficient fuzzy inference unit 7 obtains a weather correction factor K ₁ from the weather change coefficient K ₇ daytime coefficient K _9. Fluctuations in demand for water distribution due to weather conditions mainly occur during the daytime, and rarely at nighttime.

FIG. 7 shows the membership function of the time factor,
(A) The solid line is daytime, (b) the dashed line is nighttime, (c)
Indicates the morning.

Further, FIG. 8 is a daytime shows the value of the coefficient K _9.

The daytime using the coefficients K _9, weather correction factor K ₁ can be obtained by the following equation.

K ₁ = K ₇ × K ₉ (5) Next, the calculation performed by the holiday correction coefficient fuzzy inference unit 10 will be described.

Generally, during the daytime on holidays, there is a tendency for the demand for water distribution to decrease. To deal with this, if the predicted day is a holiday, it provided -K ₉ · beta becomes correction term. However, β is 0 to
It is a coefficient appropriately set between 0.5.

In addition, when holidays are consecutive, the demand for water distribution tends to decrease during the night (afternoon) of the previous weekday. If the consecutive holidays are long, this tendency becomes remarkable. To cope with this, if the predicted date is continuous holiday previous day, to deduce the correction term K ₁₀ from holiday continuous degree and time factor.

FIG. 9 shows the fuzzy rules used for this inference. This figure shows six rules.

FIG. 10 shows the membership function used for this inference. FIG. 7A shows a continuous holiday (continuous degree), and FIG. 7B shows a correction term. Weekends (Saturday / Sunday) are treated as 1.5 consecutive days off.

On weekdays after consecutive holidays, the demand for water distribution in the morning tends to increase. To cope with this, if the predicted day is weekday after continuous holiday, inferring correction term K ₁₁ from holiday continuous degree and time factor.

FIG. 11 shows the fuzzy rules used for this inference. This figure also shows six rules, as in FIG.

FIG. 12 shows the membership function used for this inference. FIG. 7A shows a continuous holiday, and FIG. 7B shows a correction term.

As shown in the following equation, by summing these correction terms can be calculated holidays correction factor K _2.

K ₂ = K ₁₀ + K ₁₁ −K ₉ · β (6) After obtaining the weather correction coefficient K ₁ and the holiday correction coefficient K ₂ in this manner, the predicted water distribution amount calculation unit 4 performs the predicted water distribution amount as described above.
Find Q _t .

Here, the demand curve of the water distribution volume tends to be delayed in the morning on holidays. To deal with this, the time delay correction unit 11
By, subjected to a time delay correction to the predicted water distribution amount Q _t.

FIG. 13 shows the time delay correction. In this figure, the predicted water distribution amount (initial value) obtained by the predicted water distribution amount calculation unit 4 is indicated by a broken line, and the corrected value is indicated by a solid line.

This time delay correction is performed based on the following equation, targeting the time zone set in the morning (see FIG. 7 (c)). Here, d is a delay time (can be set arbitrarily in the range of 0 ≦ d ≦ 120), and Q _t ′
Is the value after correction of the estimated water distribution.

The estimated water distribution _Qt (or _Qt ') obtained in this way
Is displayed on the display unit 12 and is used for various controls of the water distribution system.

In the case where the actual weather and temperature are significantly different from the forecast, the forecast weather and the forecast maximum temperature may be re-input and recalculated. In this case, the re-calculation is triggered by the operator.

 Next, a specific example of this embodiment will be described.

 FIG. 14 shows a hardware configuration of this specific example.

The main body of the device is an industrial computer 13 and its peripheral devices.
It consists of 14-17.

The industrial computer 13 performs main control of this device. The fuzzy controller 14 performs fuzzy inference. The CRT 15 is for displaying the estimated water distribution amount and other various types. Keyboard 16 and mouse 17
Is for inputting various data and instructions.

The process input / output device 18 is for inputting and outputting water distribution amount actual data and other data.
13 and connected via LAN (Local Area Network).

In the above configuration, the apparatus shown in FIG. 1 can be realized by constructing predetermined software. FIG. 15 shows the main procedure in this specific example.

H. Effects of the Invention As described above, the demand forecasting device for water distribution according to the present invention obtains a weather correction coefficient and a holiday correction coefficient from weather data and a calendar,
The average water distribution in the past is corrected by these coefficients, and if necessary, a time delay correction in the morning is performed to calculate the predicted water distribution.

In order to obtain the predicted water distribution amount by the above calculation, the degree of considering the influence of weather and temperature in the calculation of the weather correction coefficient, the degree of considering the influence of holidays in the calculation of the holiday correction coefficient, and the predicted date is a holiday. By changing the degree of delay in such a case, it is possible to easily correct / change the handling of each factor in predicting the water distribution amount.

Therefore, there is an advantage that the intuition and know-how of the operator can be easily reflected in the prediction result, and further, the prediction including the regional uniqueness peculiar to the target distribution system can be performed.

Further, even if there is no special knowledge of a computer, it is possible to construct a prediction rule and the like simply by determining the specific gravity of each factor in the calculation of the predicted water distribution amount, and there is an advantage that maintenance can be easily performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a water distribution demand forecasting apparatus according to one embodiment of the present invention, FIG. 2 is an explanatory diagram showing average water distribution data, and FIGS. 3 and 4 are used for inference of a weather coefficient. FIG. 5 and FIG. 6 are explanatory diagrams showing fuzzy rules and membership functions used in inference of weather change coefficients, and FIG. 7 is a membership diagram of time coefficients. FIG. 8 is an explanatory diagram showing a daytime coefficient value, FIG. 9 and FIG. 10 are explanatory diagrams showing a fuzzy rule and a membership function used for inferring correction before the holidays, FIG. And FIG. 12 is an explanatory diagram showing a fuzzy rule and a membership function used for inferring correction after consecutive holidays, FIG. 13 is an explanatory diagram showing time delay correction, and FIG. 14 is a hardware configuration of one specific example Block FIG. 15 is a flowchart showing a main procedure in the specific example of FIG. 14, and FIG. 16 is an explanatory view showing a conventional water distribution amount prediction process. 1 ... water distribution amount data collection unit, 3 ... water distribution amount data file,
4: Predicted water distribution calculation unit, 5: Data input unit, 6: Expected weather / expected maximum temperature data file, 7: Weather correction coefficient fuzzy inference unit, 8: Calendar file, 10: Holiday correction coefficient fuzzy inference unit, 11 ... Time delay correction unit.

Claims (1)

(57) [Claims] An apparatus for predicting the demand for water distribution in a predetermined time unit, comprising: a water distribution data collecting unit for collecting water distribution data from a water distribution system; and a water distribution data storage unit for storing past water distribution data. A weather data input unit to which weather data of the forecast day is input, and a weather correction calculation unit that performs a calculation based on the input weather data and obtains a weather correction coefficient indicating a change in demand of water distribution amount due to weather conditions. A calendar storage unit for storing a calendar; and a holiday correction operation unit for performing an operation with reference to the calendar and calculating a holiday correction coefficient indicating a change in demand for water distribution when the predicted date is before or after a holiday. Calculate the average water distribution from past water distribution data and correct the average water distribution based on the weather correction coefficient and holiday correction coefficient to obtain the predicted water distribution. A water amount prediction unit and a delay correction unit that performs an operation with reference to the calendar and delays the predicted water distribution amount for a time zone centered on the morning when the prediction day is a holiday, Demand forecasting device for water distribution.