JP3345506B2 - Water distribution forecast method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for predicting the amount of water distribution for improving the accuracy of a device for predicting the amount of water distribution, which is a support device for planning an optimal operation plan of a water purification plant.

[0002]

2. Description of the Related Art With the advancement and diversification of management in water treatment plants and the automation for reducing the burden on operators, the monitoring equipment of water treatment plants has also evolved from the function of simply monitoring the state of equipment. Planning support functions, which provide information for formulating optimal operation plans for water treatment plants, have become a major factor.

In order to formulate an operation plan for a water purification plant, it is important to predict the amount of demanded water in a demand network (water distribution system). This is the case when water is effectively distributed from a large municipal water distribution station, a high-ranked large water distribution station managed by a prefecture, or a booster water supply station to a wide area demand network.

[0004]

In many cases, the demand water amount is predicted by referring to the data of the previous day. However, the demand water volume is affected by natural phenomena such as weather and temperature, as well as by whimsical human factors, so that accurate prediction is considered impossible. An object of the present invention is to solve these difficult problems and to provide a highly accurate water distribution amount prediction method.

[0005]

SUMMARY OF THE INVENTION In view of the above, the present invention has a large variation in the daily water demand, but as a major trend, the demand water has a great correlation between the maximum temperature of the day and the weather. The focus was on things.

[0006] The solution of the present invention and the operation and effect thereof are as follows. That is, first, for an arbitrary period (about three years) from a certain past date to the present, based on a trend curve for each weather of the actual data of the water distribution amount with respect to the maximum temperature, excluding a predetermined special day, Generate an approximate expression to calculate the water distribution.

[0007] That is, in the water distribution amount prediction method for predicting the daily water distribution amount of the demand network, the water distribution amount is calculated based on the trend curve of the actual daily water distribution amount with respect to the maximum temperature in the first predetermined period from the past to the present excluding the specific day. By inputting the predicted maximum temperature into the approximate expression for calculating the water distribution amount incorporated in the arithmetic unit, the predicted value of the daily water distribution amount is obtained.

In addition, a calculation is performed to obtain an average error value between the predicted value of the daily water distribution amount and the predicted value for each weather in the latest second predetermined period within the first predetermined period. Is corrected and added to the predicted value of the daily water distribution amount.

[0009] Further, a calculation is performed to obtain an average error value between a plurality of past actual values of the determined special day and the predicted value of the daily water distribution, and the output of the arithmetic unit is corrected and added for each special day. is there.

Secondly, the amount of water in a predetermined time zone in the early morning or the previous night of the day to be predicted is measured, and the difference from the amount of water in the same time zone on the reference day is obtained by an equation incorporated in the arithmetic unit. The daily conversion amount based on the water amount difference is subtracted as statistical data, and for the prediction target day, the water amount is added to the previously calculated daily water distribution prediction value.

[0011]

The operation has been already described as means for solving the above-mentioned problems, and the description thereof will be omitted.

[0012]

Next, the present invention will be described in more detail with reference to the drawings. FIG. 1 is a trend graph showing the relationship between the maximum temperature and the daily water distribution for each weather, where A is a fine trend curve, B is a cloudy trend curve, and C is a rainy trend curve. It shows the relationship between the daily water distribution and the maximum temperature for each weather, based on the actual data on water flow.

Here, an approximate expression for predicting the daily water distribution is created based on the trend curves A, B, and C, and this is incorporated in an arithmetic unit for predicting the water distribution. By inputting the predicted value of the weather and the maximum temperature of the day to this arithmetic unit, the water distribution amount of the day can be calculated and transmitted.

Further, by incorporating the actual data further, the prediction error is calculated, and the function of automatically or manually correcting the trend curve to gradually improve the prediction function is also provided for the water distribution amount prediction. It is provided in the arithmetic unit.

[0015] The daily water distribution needs to be corrected for natural increase factors such as population increase, and of course, it is automatically corrected from past data. Manual corrections are to be made.

In such an example of the prediction method, when actual data is verified, a prediction accuracy of about 5% is obtained.
Therefore, it can be used for normal operation planning. However, when considering the above-mentioned trend curve to improve the prediction accuracy, it is noted that the trend curve that seems to show a clear correlation also has a subtle difference depending on the season and month.

Next, a description will be given of an all-weather system in which the prediction accuracy is further improved as compared with the first example.

FIG. 2 is a trend graph showing the relationship between the maximum temperature and the daily water distribution shown in FIG. The trend curve D captures the relationship between the daily water distribution amount and the maximum temperature of all weather at once based on the actual data of the daily water distribution amount in the latest period N1.

Here, the period N1 can be set arbitrarily. However, if the period N1 is too large, the influence of the season appears. If the period N1 is too small, the accuracy of the trend curve itself deteriorates.
As a result of the experiment, about 120 days were optimal.

Based on the trend curve D, an approximate expression for estimating the daily water distribution is created, and is incorporated in an arithmetic unit for predicting the water distribution.

Here, the trend curve D is automatically updated every day from that day. That is, a trend curve is always created with the data closest to the present, and the parameters of the prediction formula are also updated.

Further, the trend curve D is made by taking in all the weather data collectively. For example, the data in the case of rain is calculated from the predicted value obtained from the trend curve like the rain position CR1 shown in the figure. Also shows a low value, and the data in the case of fine shows a value higher than the predicted value like the fine position CR2.

Therefore, a period N2 shorter than the period N1 is selected, and for each weather, the data of the latest period N2 is predicted by the prediction formula generated from the trend curve D, and the average error from the actual water distribution amount is calculated. Ask for.

That is, as the addition / subtraction amount (%), fine is +1.0,
+1.2,…, +1.5 gives +1.3, rain is -2.0,
−1.5 is obtained from −1.2,…, +1.3.

Since this error indicates how the data for each weather varies with respect to the trend curve D, a highly accurate prediction can be realized based on the calculation result of correcting this variation to the previous predicted value. ing.

In the case of the second example, a long-term trend is accurately grasped, and an error caused by the weather is always corrected by reflecting the latest data with respect to all the weather to derive a predicted value. It becomes.

Therefore, it is possible to make a sufficiently accurate prediction even for abnormal weather such as record cold summer or warm winter.

It is apparent that the trend curve D may be changed to a seasonal or monthly one.

Further, instead of the above-described method of calculating the average of the errors, it is also possible to obtain the influence of the change of the weather and the like up to the previous day as a correction by the degree of ambiguity using fuzzy inference.

Further, a function for correcting an abnormal amount of water for preventing freezing of a water pipe on a winter day can be provided as follows.

On a winter day in winter, measures such as running water from a faucet at night to prevent freezing of the water pipe are taken. This amount of water is so large that it deviates greatly from the trend curve obtained simply from the weather conditions, and a phenomenon that the amount of water abnormally increases is seen.

Therefore, not only is the problem that the prediction error on that day is large, but also some measures are required to reduce the accuracy of the trend curve itself.

In particular, these tendencies involve the human system, and the following behaviors are observed in human behavior. That is, (1) Even if the minimum temperature falls and the anti-freezing is required, the water for the anti-freezing is not necessarily flown from that day. (2) The minimum temperature drops and the freezing damage actually begins to occur, and water is flushed before going to bed the next night. Generally, the amount of water increases the night after the lowest temperature drops. (3) Once the water starts to flow at night, it is difficult to stop even if it is predicted that the minimum temperature will rise. [Water volume does not decrease even if the minimum temperature forecast on the second day is high. ]

Based on these tendencies, the correction of a winter day will be described with reference to FIG. In FIG. 3, (a) shows an example of a water distribution pattern for 24 hours, (b) shows a transition of the water distribution pattern for several days including a midwinter day, and crosses indicate the same time on the same day. .

Here, the mode is switched to the midwinter day prediction mode on condition that the predicted value of the minimum temperature has become a certain value or less.

As described above, since the increase in the amount of water in the midwinter day appears remarkably early in the morning, the amount of water in a stable time zone in the early morning is detected and compared with the amount of water in the same time period in the previous day. That is, a winter day is detected at the water level shown in FIG.

Further, the difference is converted into the amount of water for one day,
This is a correction amount. Therefore, the effect of the correction is exerted after the nighttime water amount actually increases.

Specifically, for the statistical data of the latest N1 and N2 periods, a trend curve is first created by excluding this correction, and a predicted value of daily water distribution is obtained by a prediction formula based on the trend curve. This correction is added to the predicted value.

Thereafter, the water amount at the same time is sequentially compared with the data of the day before the start of the winter day, and the same correction is made.

Then, the amount of water increased until the value of the amount of water increased slightly to 110%, for example, 110% of the data on the day before the winter day.
The mode is returned to the normal period mode because the temperature has decreased to the position indicated by the mark (level P2) and the minimum temperature condition has reached the predetermined value.

Further, correction of special days such as the year-end and New Year holidays is performed in the same manner. This will be described with reference to FIG. That is, first, an anomalous day of the year is determined, an error for each anomalous day is totaled every year from the trend curve E based on data excluding the anomalous day, and an average value is obtained.

As in the case of the midwinter day, the peculiar day is, of course, excluded from the statistical data of the N1 period and the N2 period. The predicted value of the daily water distribution is obtained by a prediction formula based on the trend curve, and the correction value is added to the predicted value.

The result of the second example is that, by trial in an actual water purification plant, almost all data including an anomalous day and a midwinter day have actually obtained an accuracy of within ± 1%.

[0044]

As described above, according to the present invention, it is possible to provide a special method which can realize a water distribution amount prediction device with extremely high accuracy.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of a trend graph representing the relationship between the maximum temperature and the daily water distribution for each weather according to the present invention.

FIG. 2 is a diagram showing an example of a trend graph representing a relationship between a maximum temperature and a daily water distribution according to the present invention.

FIG. 3 is a diagram showing an example of a water distribution pattern shown for explaining a correction function of a midwinter day according to the present invention.

FIG. 4 is a diagram showing an example of a trend graph shown for explaining a correction function of a unique day according to the present invention.

[Explanation of symbols]

 A Trend curve for fine weather B Trend curve for cloudy C Trend curve for rain D Trend curve for all weather collectively (for the latest period N1) E Trend curve based on data excluding special days CR1 Rain position CR2 Fine position P1 Water level P2 Water level

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-4-73332 (JP, A) JP-A-3-202515 (JP, A) JP-A-5-17976 (JP, A) (58) Field (Int.Cl. ⁷ , DB name) E03B 1/00

Claims (2)

(57) [Claims] 1. A water distribution amount prediction method for predicting a daily water distribution amount of a demand network, based on a trend curve of an actual daily water distribution amount with respect to a maximum temperature in a first predetermined period from a past to a present excluding a specific day. , Water distribution amount incorporated in the arithmetic unit
The estimated value of the daily water distribution is obtained by inputting the expected maximum temperature to the approximate expression for calculating the predicted value of the daily water distribution, and the predicted value of the daily water distribution is calculated based on each of the predicted value of the daily water distribution and the latest second predetermined period within the first predetermined period. Starring for obtaining an average error value between the predicted value of each weather
Calculation, the output of the arithmetic unit is corrected and added to the predicted value of the daily water distribution amount, and further, an operation of calculating an average error value between a plurality of past actual values and the predicted value of the daily water distribution amount determined on a specific day. line
A water distribution amount prediction method , wherein the output of the arithmetic unit is corrected and added for each unique day. 2. A method for measuring the amount of water in a predetermined time zone in the early morning or the night of the previous day on the day to be predicted and calculating the difference from the amount of water in the same time zone on the reference day in an arithmetic unit. 2. The method according to claim 1, wherein the daily conversion amount based on the water amount difference is subtracted as statistical data.