KR20140147456A

KR20140147456A - Method for predicting daily water demand in water distribution network

Info

Publication number: KR20140147456A
Application number: KR1020130070751A
Authority: KR
Inventors: 이봉국; 강민구; 김병섭
Original assignee: 엘에스산전 주식회사
Priority date: 2013-06-20
Filing date: 2013-06-20
Publication date: 2014-12-30

Abstract

The present invention relates to a method for predicting the daily demand of water supplied through a water distribution network. The method includes, by a control unit, collecting daily water demand history data from a database system of a host server, searching and compensating error data and missing data from the collected daily water demand history data, dividing the period of the compensated daily water demand history data and predicting the water demand by periods, and predicting the daily water demand by integrating the water demand predicted by periods.

Description

[0001] The present invention relates to a method for predicting daily demand of a water network,

The present invention relates to a method for predicting a daily demand amount of a water supply network for predicting a daily demand amount of water supplied through a water supply network.

In general, the facilities of the water supply network are essential for maintaining the activities of the city, and are influenced by socioeconomic factors such as urban population concentration, improvement of living standard, economic growth, .

This requires a comprehensive maintenance plan for the water supply network, and it is necessary to predict the demand for long-term waterworks in each region considering the factors such as population development trends, changes in living standards and economic growth in the future.

Demand quantities for water demand forecasting through water distribution networks are nonlinear and have complex time series characteristics. In the past, the unit time series, such as the artificial intelligence method of the neural network or the statistical method of the auto regressive moving average model, And the demand for water is predicted by the use of water.

The problem to be solved by the present invention is to solve the time series by characteristic periods in consideration of the nonlinearity and complexity of the water distribution network and to predict the demand amount by the decomposed period and integrate them to accurately predict the daily demand amount. Provides a daily demand forecasting method.

It is to be understood that both the foregoing general description and the following detailed description of the present invention are exemplary and explanatory and are intended to provide further explanation of the invention as claimed. There will be.

The method of predicting the daily demand of a water network according to the present invention comprises the steps of: collecting daily demand history data from a database system of a host server; searching and correcting error data and missing data from the collected daily demand history data; Estimating a demand amount for each cycle, and estimating a daily demand amount by integrating a demand amount predicted for each cycle.

The step of retrieving and correcting the error data and the missing data from the collected daily demand amount history data may include correcting the error data and the missing data by performing an average value reuse of the previous daily demand amount history data.

The step of decomposing the period of the corrected daily demand amount history data may include decomposing the period through the wavelet transform process.

The step of predicting the demand amount for each cycle may include a step of predicting a demand amount for each cycle using a SVM (Support Vector Machine) model, which is a nonlinear time series prediction model.

According to the method of predicting the daily demand of the water distribution network of the present invention, the daily demand amount data of the water supplied through the water pipe network is decomposed and simplified according to the characteristic period, the demand amount is predicted by the decomposed period, .

Therefore, it is possible to minimize the prediction error of the daily demand of water supplied through the water network.

In addition, by minimizing the forecast error of the daily demand of water, it is possible to improve the prediction error of the subsequent demand, that is, the demand amount by time, and to optimally perform the optimal operation process by utilizing it.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, in which like reference numerals refer to like elements throughout.
1 is a signal flow diagram for explaining a conventional daily demand forecasting method,
2 is a block diagram showing the configuration of a system to which the daily demand forecasting method of the present invention is applied, and FIG.
3 is a signal flow diagram for explaining a daily demand forecasting method of the present invention.

The following detailed description is merely illustrative, and is merely an example of the present invention. Further, the principles and concepts of the present invention are provided for the purpose of being most useful and readily explaining.

Accordingly, it is not intended to provide a more detailed structure than is necessary for a basic understanding of the present invention, but it should be understood by those skilled in the art that various forms that can be practiced in the present invention are illustrated in the drawings.

1 is a signal flow diagram for explaining a conventional daily demand forecasting method. Referring to FIG. 1, the forecasting system first accesses a database system of a host server and collects daily demand quantity history data for predicting a daily demand quantity of the waterworks (S100).

Then, the collected daily demand amount history data is initially processed (S102). That is, the prediction system searches for the error data, the missing data, and the like from the collected daily demand history data, corrects the retrieved error data and the missing data so that the error data and the missing data can be reused with the previous average value, Into data that can be read.

The initial demanded daily demand history data is applied to a predetermined model, for example, a neural network model or a time series model (Auto Regressive Moving Average Model) to perform a learning data process (S104) And real-time daily demand is predicted (S106).

However, the above-mentioned conventional daily demand forecasting method has a nonlinear and complex time series characteristic for the demand quantity forecast for water demand forecast supplied through a water supply network, and has a time series characteristic such as the artificial intelligence method of the neural network or the statistical method of the ARIMA model There is a problem that the predictive power of the daily demand is deteriorated because the time series is used as it is.

2 is a block diagram showing a configuration of a forecasting system to which a daily demand forecasting method of the present invention is applied. Reference numeral 200 denotes a communication unit. The communication unit 200 communicates with a host server (not shown in the figure), and receives daily demand amount history data from a database provided in the host server.

Reference numeral 202 denotes an input unit. The input unit 202 inputs a predetermined operation command according to the operation of the operator.

Reference numeral 204 denotes a control unit. The control unit 204 controls the communication unit 200 to receive the daily demand amount history data from the database provided in the host server according to a user's operation command inputted through the input unit 202 The daily demand amount is predicted by the history data of the received daily demand amount, and the predicted daily demand amount is displayed and provided to the operator.

Reference numeral 206 denotes a memory. The operation program of the control unit 204 is stored in advance in the memory 206, and the operation data and the like of the control unit 204 are stored.

Reference numeral 208 denotes a display section. Under the control of the control unit 204, the display unit 208 displays a predicted operation state of the daily demand amount, a predicted daily demand amount, and the like.

3 is a signal flow diagram for explaining a daily demand forecasting method of the present invention. Referring to FIG. 3, the control unit 204 communicates with the host server through the communication unit 200 to collect daily demand history data stored in the database system (S300).

Then, the controller 204 performs initial processing of the collected daily demand quantity history data (S302). In the initial process, the error data and the missing data are searched in the collected daily demand history data, and the retrieved error data and the missing data are converted into data that can predict the daily demand amount by using the average value reuse of the previous value do.

When the initial processing is completed, the controller 204 performs a period decomposition process (S304). The period decomposition process decomposes the nonlinear and complex daily demand history data by, for example, wavelet transformation.

When the period decomposition process is completed, the controller 204 performs a demand amount prediction operation for each cycle (S306). That is, the control unit 204 predicts a demand amount for each cycle using a SVM (Support Vector Machine) model, which is a nonlinear time series prediction model.

If the demand for each cycle is predicted, the controller 204 integrates the predicted demand signals for each cycle to predict the daily demand (S308).

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, I will understand.

Therefore, the scope of the present invention should not be limited to the above-described embodiments, but should be determined by equivalents to the appended claims, as well as the appended claims.

200: communication unit 202: input unit
204: control unit 206: memory
208:

Claims

The control unit collecting daily demand amount history data from the database system of the host server;
Retrieving and correcting the error data and the missing data from the collected daily demand history data;
Decomposing the period of the corrected daily demand quantity history data;
Estimating a demand amount for each cycle; And
And estimating a daily demand amount by integrating forecasted demand amounts for each cycle.

The method according to claim 1,
Searching and correcting the error data and the missing data in the collected daily demand history data,
And correcting the error data and the missing data by performing an average value reuse of the previous daily demand quantity history data to correct the daily demand quantity of the water network.

The method according to claim 1,
The step of decomposing the cycle of the corrected daily demand amount history data includes:
And decomposing the cycle through a wavelet transform process.

The method according to claim 1,
The step of predicting the demand for each cycle comprises:
And predicting demand quantities for each cycle using a SVM (Support Vector Machine) model, which is a nonlinear time series prediction model.