CN112052978A - Water demand prediction system and method - Google Patents
Water demand prediction system and method Download PDFInfo
- Publication number
- CN112052978A CN112052978A CN202010410657.0A CN202010410657A CN112052978A CN 112052978 A CN112052978 A CN 112052978A CN 202010410657 A CN202010410657 A CN 202010410657A CN 112052978 A CN112052978 A CN 112052978A
- Authority
- CN
- China
- Prior art keywords
- water demand
- water
- prediction
- day
- demand
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 350
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000004364 calculation method Methods 0.000 claims description 29
- 230000008569 process Effects 0.000 claims description 24
- 238000012544 monitoring process Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims 2
- 238000013523 data management Methods 0.000 description 29
- 238000007726 management method Methods 0.000 description 17
- 230000008859 change Effects 0.000 description 11
- 239000013598 vector Substances 0.000 description 10
- 230000006870 function Effects 0.000 description 9
- 238000005259 measurement Methods 0.000 description 8
- 238000000746 purification Methods 0.000 description 6
- 230000001186 cumulative effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005406 washing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A20/00—Water conservation; Efficient water supply; Efficient water use
- Y02A20/152—Water filtration
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The conventional water demand predicting device cannot predict the water demand with high accuracy according to the variation of the uneven water demand. The invention provides a water demand prediction system and a method thereof. The computer obtains the water demand of the specified water distribution area, the weather information and the week information of the water distribution area including the forecast; setting weather information, week information, and a water demand for a predetermined period in the past as conditions relating to the water demand for the predetermined period; the similarity between the correlation information on the predicted day and the correlation condition on each day of the past predetermined period is calculated, and the water demand on the predicted day is predicted by preferentially using the actual water demand on the past day having the condition of high similarity.
Description
Technical Field
The present invention relates to a water demand prediction system and a method thereof for predicting a water demand per reference period for a predetermined water distribution area where water is distributed from a water distribution facility through a water distribution pipe network.
Background
Conventionally, water supply facilities employ control for supplying water to consumers including general households. The water demand varies not only by day, by predetermined time, by week and time, but also by external factors such as weather and temperature. Therefore, there is a conventional example in which the water demand is predicted accurately.
For example, japanese patent laid-open No. 2012-99049 discloses a water distribution plan prediction system as follows: the actual value of the water demand for the same week as the predicted day is acquired from the past predetermined period, and the water demand pattern value of 1 day and 24 hours calculated by taking the average value of the respective times is used to predict the water demand.
The water distribution plan prediction system obtains the latest water demand actual value, and corrects the pattern value so that the deviation between the pattern value and the actual value is reduced when the deviation between the water demand pattern value (predicted value) and the actual value is out of a predetermined allowable range, so that the water demand prediction accuracy is improved by adding or subtracting a certain amount to or from the pattern value.
Patent document 1: japanese laid-open patent publication No. 2012-99049
Disclosure of Invention
The water distribution plan prediction system described above can improve the prediction accuracy of the water demand when the water demand increases or decreases uniformly, but, for example, when the water demand changes over time, such as the washing time is changed according to the weather, and the water demand is advanced or retarded, the pattern value is corrected, and the water demand prediction accuracy is rather deteriorated. Therefore, an object of the present invention is to provide a system capable of predicting water demand with high accuracy even in a case where the change in water demand is not uniform.
In order to achieve the above object, the present invention is directed to a water distribution system for distributing water from a water distribution facility through a water distribution pipe network, wherein a water demand per reference period is predicted, and the water demand of the water distribution system, weather information and week information of the water distribution system are acquired; setting weather information and week information for the predetermined period and a water demand in a predetermined range before the reference period as conditions relating to the water demand for the reference period; the similarity between the relevant condition in each reference period of the past predetermined period and the relevant condition in the reference period for predicting the water demand is calculated, and it is preferable to predict the water demand in the reference period for predicting the water demand, for example, by using the actual water demand value in the reference period having a condition with a high similarity.
Effects of the invention
According to the present invention, it is possible to provide a system capable of predicting the water demand with high accuracy even when the change in the water demand is not uniform.
Drawings
Fig. 1 is an example of a block diagram of a water supply system including a computer system as a water demand prediction system.
Fig. 2 is an example of a water demand data management table.
Fig. 3 is an example of the week and weather data management table.
Fig. 4 shows an example of a screen for setting the water demand prediction condition.
Fig. 5 is an example of the prediction condition management table.
Fig. 6 is an explanatory diagram of a Kernel ridge regression (Kernel ridge regression) model.
Fig. 7 is a graph illustrating the relationship of the predicted day and the latest water demand data of the learning day.
Fig. 8 is a graph illustrating the relationship of the predicted day and the latest water demand data of the learning day.
Fig. 9 is a graph illustrating the relationship of the predicted day and the latest water demand data of the learning day.
Fig. 10 is an example of a water demand prediction data management table.
Fig. 11 is an example of a prediction result display data management table.
Fig. 12 shows an example of a prediction result display screen.
Fig. 13 is an example of a model parameter management table.
Description of the reference symbols
100 … water distribution equipment
101 … computer system
102 … Water supply System
103 … monitoring control system
104 … Water purification plant
105 … pump
106 … distributing pool
107 … water level gauge
108 … flowmeter
109 … water distribution network
110 … demand side
111 … network
121 … input unit
122 … display part
123 … data acquisition unit
124 … prediction condition setting unit
125 … prediction part of water demand
126 … prediction result display part
127 … model parameter determination section
131 … water demand data management table
132 … week and weather data management table
133 … prediction condition management table
134 … model parameter management table
135 … water demand forecast data management table
136 … prediction result display data management table
Detailed Description
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The present invention is not limited to the embodiments described below.
Fig. 1 shows a block diagram of the water supply system 102. The water supply system includes a water distribution facility 100, a supervisory control System (SCADA)103 for the water distribution facility, a computer system 101 for predicting water demand, and a network 111 for connecting the supervisory control system 103 and the computer system 101.
The water distribution facility includes a water purification plant 104, a pump 105, a distribution tank 106, a distribution pipe network 109, and meters 107 and 108, and a monitoring control system 103 monitors and controls the pump 105 and the meters 107 and 108.
The water purified in the water purification plant 104 is pumped to a distribution tank 106 by a pump 105, temporarily stored in the distribution tank 106, and then distributed to a customer 110 through a distribution network 109. The pump 105, the water level meter 107, and the flow meter 108 measure the operation state (ON/OFF state) of the pump 105, the water level of the distribution tank 106, and the amount of water distributed to the consumer 110 (water demand amount) at 1 minute intervals, and send the measured values to the monitoring control system 103.
The monitoring and control system 103 obtains the water demand measurement value (1 minute unit) of the target distribution area to be distributed from the water distribution facility via the distribution network 109, sums the water demand measurement values for 1 hour, counts the 1 hour unit demand for each hour, and transmits the counted amount to the computer system 101 via the network 111.
A weather information providing system (weather information source) 112 existing outside the water supply system 102 distributes weather information to the computer system 101 via the network 111 every time an actual value of weather information of an area including the subject water distribution area, and a latest value of a predicted value of weather information in the near future on the second day, the third day, and the like are obtained.
The computer system 101 acquires water demand data from the monitoring control system 103 via the network 111, acquires weather information data (including forecasts) from the weather information providing system 112, performs prediction calculation of an hourly demand from a predetermined time (for example, 0 hour) to 24 hours later every day (every reference period) using the acquired data, and transmits the prediction result to the monitoring control system 103 via the network 111.
The monitoring control system 103 obtains the water demand prediction result, makes a water operation plan (such as a water purification plan of water in the water purification plant 104, an operation plan of the pump 105, and a water storage plan of the distribution tank 106) for each hour from the predetermined time to 24 hours after the predetermined time, which is matched with the water demand prediction result, and performs operation control of the water purification plant 104 and the pump 105 based on the water operation plan. As described above, daily water demand prediction and planning of water use based on the prediction are performed, and operation control of the water supply system is performed based on the water use plan so that an appropriate amount of water is supplied to the demand side.
The computer system 101 includes general computer hardware such as a Controller (CPU), a storage device (RAM, hard disk, flash memory, etc.), an input unit 121 (keyboard, mouse, etc.), and a display unit 122 (display, printer, etc.).
The storage device is provided with a data acquisition unit 123, a prediction condition setting unit 124, a water demand prediction unit 125, a prediction result display unit 126, and a model parameter determination unit 127 as functional blocks realized by the execution of programs by the controller. Each functional module may be referred to as a mechanism, a unit, or the like.
The storage device stores a water demand data management table 131, a day and weather data management table 132, a prediction condition management table 133, a model parameter management table 134, a water demand prediction data management table 135, and a prediction result display data management table 136. The controller utilizes these tables in executing the modules described above.
The data acquisition unit 123 acquires the water demand actual data transmitted from the monitoring control system 103 and the weather information data transmitted from the weather information providing system 112, and registers them in a predetermined table.
The prediction condition setting unit 124 sets, in a predetermined table, the prediction start time, the conditions related to the water demand (weather, day of the week, water demand in the last predetermined period, etc.), the number of days of the last past day (learning day), etc., which are used for the prediction calculation, which are input by the administrator of the computer system 101 through the input unit 121.
The water demand predicting unit 125 predicts the water demand on the predicted day based on the kernel ridge regression model using the above-described related condition data on the predicted day and each day of the past predetermined period, and distributes the prediction result to the monitoring control system 103.
The prediction result display unit 126 creates a display screen of the water demand prediction result and displays the display screen on the display unit 122.
The model parameter determination unit 127 determines model parameters of a kernel ridge regression model used for the prediction calculation of the water demand.
The water demand data management table 131 is a data group for managing the hourly water demand meter values.
The day of the week and weather data management table 132 is a data group for managing day of the week information and weather information (including forecast).
The prediction condition management table 133 is a data group for managing conditions for prediction calculation, such as a prediction start time, conditions related to the water demand (weather, day of the week, water demand in the latest predetermined period, and the like), and the number of days of the latest past day (learning day) used for prediction calculation.
The model parameter management table 134 is a data group for managing model parameters of a kernel ridge regression model used for the prediction calculation of the water demand.
The water demand prediction data management table 135 is a data group for managing a water demand prediction value for each hour.
The prediction result display data management table 136 is a data group for managing display information of prediction results such as a predicted value of water demand, a measured value (actual value), a cumulative error of water demand, and a converted value of the water level of the distribution tank for the cumulative error every hour from the prediction start time.
The computer system 101 executes the following processes (1) to (5) to predict the water demand with reference to the water demand of the past day having a condition close to the condition associated with the current water demand, thereby achieving highly accurate water demand prediction that matches the change in the water demand.
(1) Water demand and weather information acquisition
(2) Setting of prediction conditions
(3) Water demand prediction
(4) Display of water demand prediction results
(5) Determination of model parameters
Hereinafter, (1) to (5) will be described with reference to fig. 2 to 13. In the process (1) of acquiring the water demand and the weather information, the monitoring and control system 103 transmits the latest water demand measurement value and the measurement time thereof to the computer system 101 via the network 111 every time the water demand of 1 hour unit in the target distribution area is counted.
The data acquisition unit 123 of the computer system 101 sequentially acquires the transmitted water demand measurement value (actual value) and the measurement time thereof, and registers the acquired value and time in the water demand data management table 131. Fig. 2 shows an example of the water demand data management table 131 registered by the data acquisition unit 123.
The weather information providing system 112 distributes the actual value of the weather information of the region including the water distribution area of the subject and the predicted value of the weather information of the next day 1 time a day to the computer system via the network 111.
The data acquisition unit 123 acquires the distributed weather information, classifies the weather information into information of the highest temperature, the presence or absence of sunny weather, the presence or absence of rain weather, and the presence or absence of snow weather, and registers (updates) the weather information in the day and day management table 132 together with the date, the day (monday to sunday, holidays), and the actual/forecast. Fig. 3 shows an example of the day and weather data management table 132 registered by the data acquisition unit 123.
The setting process (2) of the prediction conditions will be described. In general, the water demand for 1 day (time series of 24 hours) varies depending on the weather, the highest temperature, the week, the latest water demand (time series) of the day, and the like, and days similar to the conditions associated with these water demands are in a tendency to become the same water demand as each other.
Therefore, the computer system 101 predicts the water demand of the predicted day using the past water demand actual data (may be multi-day data) having the above-described association condition similar to the predicted day. The prediction condition setting process (2) sets in advance the preconditions (prediction conditions) required for the calculation of the water demand prediction, such as the prediction start time required for the calculation of the water demand prediction, the conditions relating to the water demand amount, and the number of days of the latest past day referred to in the prediction calculation.
The manager of the computer system 101 inputs the above-described prediction start time, the conditions related to the water demand (selected from the above-described weather, maximum temperature, day of the week, recent water demand, and the like), and the number of days of the latest past day referred to in the prediction calculation, from the input unit 121. Then, the prediction condition setting unit 124 of the computer system registers the set prediction conditions in the prediction condition management table 133.
Fig. 4 shows an example of a display screen for setting the prediction conditions displayed on the display unit 122 by the prediction condition setting unit 124. The display screen is configured such that, as a representative condition of the conditions relating to the water demand, the manager can select an appropriate condition for the target distribution area from the weather, the highest temperature, the week, the latest water demand (time series), and the like.
In the case where the latest water demand (time series) is selected, the manager also inputs the latest water demand used for several hours in the prediction calculation, and the period thereof. For example, when the prediction start time is 0 and the period of the latest water demand is 8 hours, the water demand time-series data of 16 hours to 0 hours (24 hours) are used to predict the water demand for 0 to 24 hours on the second day.
When the latest water demand (time series) is selected, the manager selects whether the water demand is used in the prediction calculation as "water demand time-series data alone", as "water demand time-series data + differential data", or as "water demand time-series data + total data".
As will be described later, as the above-described correlation condition for the water demand, the prediction calculation using not only the latest water demand time-series data but also the difference data or the total amount data thereof can perform prediction calculations having different characteristics (prediction in which the waveform is similar to the trend of the actual value, prediction in which the accumulated error from the actual value is small, and the like).
When the week information is selected, the administrator selects whether to use the week information as "weekday and holiday categories (divided into two categories)" for prediction calculation, as "each week and holiday category (divided into 8 categories)", or as "each week and holiday category (divided into 8 categories)" but change the week information to a predetermined week of continuous rest of 3 days or more ".
"the predetermined day of continuous rest which is changed by 3 or more days at each day of the week and each holiday (divided into 8 types)" is to change the day before the continuous rest of 3 or more days to friday, the first day to saturday, and the last day to sunday. This is because, depending on the water distribution area, performing the week change as described above improves the accuracy of predicting the water demand. For example, in the case of 3 consecutive holidays of "friday (holiday), saturday, sunday", although the day before that is thursday, since it is a weekday before the holiday, there is a case where a pattern similar to the water demand of friday is presented. In addition, for example, in the case of 3 consecutive holidays of "saturday, sunday, and monday", the last monday (holiday) is a holiday, but since it is a holiday before weekday, a pattern similar to the water demand of the sunday may be presented. Therefore, the change of the week described above leads to an improvement in the prediction accuracy.
Fig. 5 shows an example of the prediction condition management table 133 registered by the prediction condition setting unit 124. The setting contents of the prediction condition management table 133 in fig. 5 correspond to the setting contents when the prediction conditions are set as shown in fig. 4.
As described above, the prediction condition setting unit 124 performs the setting process of the prediction condition.
Next, the water demand amount prediction process (3) will be described. The computer system 101 predicts the water demand on the predicted day by preferentially using the actual water demand value for at least one day of the last past predetermined period having the relevant condition data similar to the predicted day using the relevant condition data (weather, maximum temperature, week, time series of the last water demand, etc.) corresponding to the water demand on the predicted day, the relevant condition data corresponding to the water demand on each day of the last past predetermined period, and the actual water demand data corresponding to the predicted target time zone on each day.
For example, the similarity between the related condition data on the current day of the prediction day and the related condition data on each day of the last predetermined period is calculated, and prediction calculation is performed using a model formula (for example, a weighted average of the water demand actual values in accordance with the similarity on each past day) that gives priority to the water demand actual value in the prediction target time zone on the past day having the related condition data with a high similarity. The computer system 101 uses, for example, a kernel ridge regression model as a model for the above-described similarity-based prediction calculation.
The kernel ridge regression model will be described with reference to fig. 6. When actual data (learning data) indicated by ○ is given on the XY plane, it may be considered to predict the value Ynew of the Y coordinate corresponding to the input of the value Xnew of the X coordinate. In this case, yenew is expected to take a value close to the Y coordinate of real data in the vicinity of Xnew, and needs not much consideration for real data far from Xnew. Therefore, when N actual data (learning data) are given, the value Ynew of Y corresponding to the input Xnew can be expressed as [ equation 1]
The kernel ridge regression model is a model in which a function indicating the proximity between input data and actual data in (equation 1) is expressed by a Gaussian (Gaussian) kernel function and a multiplier corresponding to the function is expressed not by a Y coordinate value of the actual data but by a parameter, based on the above-described consideration, and is formulated as follows.
When setting:
yt: water demand at time t (time scale is 1 hour unit),
z1: binary variables (1: present, 0: absent) indicating the presence or absence of a sunny day of the forecast,
z2: a binary variable indicating the presence or absence of rain for the predicted day,
z3: a binary variable indicating the presence or absence of snow on the predicted day,
z4: the maximum air temperature at the day of the forecast,
and Wi: a binary variable representing the week of the predicted day (one variable of W1 only when distinguished by weekday and holiday, and 8 variables of W1 to W8 corresponding to monday, sunday, and holiday when distinguished by monday, sunday, and holiday), and input data (vector) composed of condition data (weather, maximum temperature, week, and latest water demand time series) relating to the water demand are set to Xt ═ y (Yt, Yt-1, …, Yt-m + 1, Z1, Z2, Z3, Z4, Wi)
Then, the water demand Yt + n at time t + n (n is 1, …, 24) is predicted using the following expression,
[ numerical formula 2]
Here, N: number of days of study data
α i: regression parameters (i ═ 1, …, N)
(Xt (i), Yt + n (i)): the model learning data on the ith day (i.e., the input/output pair data corresponding to the past day at the same time as the input/output data on the prediction day) of the last N days. Here, k (Xt, Xt (i)) is a gaussian kernel function representing the degree of closeness between input data Xt for a prediction day and input data Xt (i) for learning on the ith day in the past, and is defined by the following expression,
[ numerical formula 3]
k (Xt, Xt (i)) exp (- β × | < Xt-Xt (i) | | < 2 > · (formula 3))
Here, | Xt-Xt (i) |: distance (L2 norm, square root of sum of squared residuals of each component of each vector) between input data (vector) Xt and Xt (i)
Beta: hyperparameters (. beta. > 0).
The Gaussian kernel function takes continuous values of 0 to 1, and takes values closer to 1 as the distance between the input data Xt on the prediction day and the input data Xt (i) for learning on the ith day in the past is closer, and takes values closer to 0 as the distance is farther. Therefore, the more the data of the past day having input data similar to (close to) the input data of the predicted day (the relevant condition data for the water demand amount: weather, maximum temperature, day of the week, and latest water demand time series), the more the influence on the predicted value is, the more the data of the past day having input data similar to (close to) the input data of the predicted day is used for prediction with priority.
In the above description, the input data vectors corresponding to the predicted day and the past day (learning day) are made up of the weather, the highest temperature, the week, and the latest water demand time series, but here, focusing only on the water demand time series, as shown in the example of fig. 7, it is assumed that there are water demand data corresponding to the predicted day and water demand data corresponding to the two learning days 1 and 2.
In this case, other conditions (weather, maximum temperature, and week) are assumed to be the same for each day. Assuming that the learning day 1 data (the recent water demand) always exceeds the predicted day data by a predetermined amount, the learning day 2 data is shifted up or down by the predetermined amount with respect to the predicted day data. In this case, it is generally expected that if the other conditions are the same, the above-described relationship tends to continue to some extent also in the water demand data after the prediction start time on the prediction day and the learning days 1 and 2. As described above, since the distance (similarity) between the two input data vectors is calculated by the L2 norm, that is, the sum of the squares of the errors of the respective components, the distance between the predicted day data and the two learning day 1 and 2 data is the same, and as shown in fig. 7, the water demand in the prediction time zone of the learning day data 1 and 2 is likely to be reflected in the prediction result of the water demand prediction on the predicted day of the prediction day to the same extent (fig. 7 is a pictogram, and this is not necessarily true).
Here, if the difference value from the previous time of each time of the latest water demand time series is added as a component of the input data vector, the difference values of the latest water demand time series of the predicted day data and the learning day 1 data are of the same degree, so that the learning day 1 data is closer to the predicted day data than the learning day 2 data with respect to the distance (degree of similarity) to the predicted day data, and as shown in fig. 8, it is easy to be a prediction result in which the water demand in the prediction period of the learning day 1 data is actually reflected largely in the water demand prediction on the predicted day, that is, a prediction result of a waveform approximating an actual value more than a predetermined amount than the actual value on the predicted day. However, in this case, since the predicted value always exceeds the actual value, the accumulated error of the water demand prediction increases. Since the error in the prediction of the water demand is expressed as an actual error in the water distribution tank level and an error in the prediction, an increase in the accumulated error may cause the water distribution tank level to deviate from the upper and lower limit levels. Therefore, for water use, a predicted value that fluctuates around an actual value so as not to increase an accumulated error may be more desirable than a predicted result of a water demand that always exceeds the actual value.
Next, if the total value of the latest water demand time series is added as a component of the input data vector, since the total value of the latest water demand time series of the prediction day data and the learning day 2 data takes the same value, the learning day 2 data is closer to the learning day 1 data with respect to the distance (similarity) to the prediction day data, and as shown in fig. 9, it is easy for the water demand in the prediction period of the learning day 2 data to be reflected substantially in the prediction result of the water demand prediction on the prediction day, that is, the prediction result of the water demand prediction on the prediction day that has shifted up and down by a predetermined amount with respect to the actual value on the prediction day. In this case, since the actual value is likely to be close to the total water demand amount, it is possible to expect a reduction in the cumulative error of the water demand prediction.
Therefore, by newly adding the difference value or the total value of the latest water demand time series to the input data vector corresponding to the prediction day and the past day (learning day), it is possible to select a prediction result in which the waveform is actually approximated to the water demand, a prediction result in which the accumulated error is small, or the like. The addition of the difference value or the total value to the input data vector corresponds to the addition of the difference data or the addition of the total amount when the latest water demand is selected as the relevant condition data for the water demand described in the setting process (2) of the prediction condition, and the administrator of the computer system 101 may switch the selection in accordance with the purpose of the water demand prediction.
Here, the regression parameter α i (i is 1, …, N) is determined so as to minimize an evaluation function R composed of a sum of squares of errors with respect to the prediction result when all the learning data of the past day are input to (expression 2) and a regularization term for preventing overfitting to the learning data, each time prediction calculation is performed.
[ numerical formula 4]
The evaluation function R is a function of the regression parameter α i, and the regression parameter α i that minimizes R can be analytically calculated (numerical expression is omitted). Therefore, each time the prediction calculation is performed, the regression parameter α i can be calculated using the input data Xt for the prediction day obtained up to the time t and the input/output data Xt (i) and Yt + n (i) for the past day (learning data day), and the predicted water demand value Yt + n for the prediction target time t + n can be obtained by the formula 2. By repeating the above calculation (the parameter α i is also calculated for each n) while changing n to n 1, …, 24, it is possible to calculate the predicted water demand value for the prediction target time zone (24 hours from time t +1 to time t + 24).
The regularization term includes a parameter λ (not shown) that defines the degree of prevention of overfitting to the learning data. β included in the kernel function and λ included in the regularization term are hyper-parameters (parameters for controlling the dynamics of the algorithm), and are determined in advance by an exhaustive method (japanese text: Gross — たり method) or the like using data in a predetermined past period so that the sum of squares of errors between the predicted value and the actual value of the water demand amount calculated by the above calculation is minimized.
Therefore, the prediction calculation of the water demand on the day of the prediction day can be performed using the relevant condition data (weather, maximum temperature, week, latest water demand time series, etc.) corresponding to the water demand on the prediction day, the relevant condition data corresponding to the water demand on each day of the latest past predetermined period, and the water demand actual data corresponding to the prediction target time period on each day, on the basis of the kernel-based regression model, with priority given to the water demand actual value on the day (which may be multiple days) of the latest past predetermined period having the relevant condition data similar to the day of the prediction day.
The water demand prediction process (3) is basically performed 1 time a day at the time when the latest water demand measurement value is obtained up to the prediction start time, but when the manager of the computer system 101 determines that the water demand prediction error is large as described later, the water demand prediction process (re-prediction) can be performed by the instruction of the manager.
The water demand predicting unit 125 of the computer system 101 acquires, from the prediction condition management table 133, a prediction start time required for water demand prediction, conditions (weather, maximum temperature, day of the week, latest water demand time series, etc.) relating to the water demand, and the number of days of the latest past day (the number of days of learning data).
Then, the water demand predicting unit 125 acquires the predicted day, the latest water demand actual value up to the prediction start time of the specified past day (learning day), the data of the weather, the maximum temperature, the day of the week, and the water demand actual value data of the predicted time zone of the specified past day (learning day) from the water demand data management table 131 and the week and weather data management table 132.
At this time, the water demand predicting unit 125 performs a change of day information (a predetermined change of day of continuous rest for +3 or more days on weekday and holiday/week and holiday +3 days or more) and an addition of the water demand time-series data related information to the input data vector (water demand time-series data alone/water demand time-series data + difference data/water demand time-series data + total data) in accordance with the setting contents performed in the process (2). Then, the water demand predicting unit 125 obtains the hyper-parameters β and λ of the kernel ridge regression model from the model parameter management table 134.
The water demand predicting unit 125 calculates a predicted water demand value for a predicted time zone of a predicted day by using the obtained data and prediction calculation based on the kernel ridge regression model, registers the prediction result in the water demand prediction data management table 135, and distributes the result to the monitoring control system 103 via the network 111. As a result, the supervisory control system 103 can make an operation plan of the pump 105 and a planned water level of the distribution tank 106 based on the result of the demand prediction of the water to be distributed. Fig. 10 shows an example of the water demand prediction data management table 135 registered by the water demand prediction unit 125.
As described above, the water demand predicting unit 125 performs the water demand predicting process.
Next, the display process (4) of the water demand prediction result will be described. This process (4) is executed each time the latest water demand prediction result of the process (3) is obtained and each time the latest water demand measurement value (actual value) of the process (1) is obtained. The prediction result display unit 126 of the computer system 101 acquires the latest prediction result from the water demand prediction data management table 135, acquires the latest actual water demand value from the water demand data management table 131, and calculates the cumulative error of the water demand by taking the sum of the prediction errors (predicted value-actual value) at the respective times from the prediction start time (1 hour).
The prediction result display unit 126 calculates a converted value of the water level of the distribution tank 106 by dividing the calculated integrated error by the cross-sectional area of the distribution tank 106. The prediction result display unit 126 registers the prediction date, the actual value, the accumulated error, and the converted value of the sump water level for the accumulated water demand amount error, which are obtained and calculated as described above, in the prediction result display data management table 136, creates a prediction result display screen on which the data are displayed as icons, and displays the prediction result display screen on the display unit 122.
Fig. 11 shows an example of the prediction result display data management table 136 registered in the prediction result display unit 126. Fig. 12 shows an example of a prediction result display screen created by the prediction result display unit 126.
As described above, the prediction result display unit 126 performs the display processing of the water demand prediction result.
The manager of the computer system 101 can read the prediction result display screen to determine whether or not the water demand prediction error is increased, and can predict the water demand again using the latest data when it is determined that the water demand prediction error is increased. When performing the re-prediction, the administrator inputs a new prediction start time (current time) from the input unit 121 as in the above-described process (2), and instructs the re-prediction from the prediction start time.
Then, the water demand predicting unit 125 performs the prediction calculation of the water demand 24 hours after the new prediction start time using the latest water demand actual value, weather data, and the like as in the above-described process (3), registers the prediction result in the water demand prediction data management table 135, and distributes the result to the monitoring control system 103. Thus, even when the prediction error is large, the water demand can be re-predicted and calculated using the latest data.
Next, the determination process (5) of the model parameters will be described. This treatment (5) may be performed periodically, for example, 1 time per year or 1 time per half year. The model parameter determination unit 127 sets initial values (sufficiently large values) of the parameters β and λ of the kernel ridge regression model, sets each day of a predetermined period (for example, 3 months) in the past from the current time as a prediction day, calls up the water demand prediction unit 125, and executes the prediction process (3) of the water demand for a prediction time zone on each prediction day. The conditions used in the prediction at this time use the conditions registered in the prediction condition management table 133.
Then, the model parameter determination unit 127 acquires the water demand actual value for each prediction day from the water demand data management table 131, and calculates the square sum of the prediction errors of the prediction values and the actual values for all the prediction time periods and the prediction days.
The model parameter determination unit 127 repeats the prediction process while decreasing the values of the parameters β and λ by predetermined values, and registers the values of the parameters β and λ that minimize the sum of squares of the prediction errors calculated as described above as optimal parameter values in the model parameter management table 134. Fig. 13 shows an example of the model parameter management table 134.
As described above, the model parameter determination unit 127 performs the process of determining the model parameters.
In the embodiment described above, if the correction of the model parameters is continued to perform the learning of the kernel ridge regression model, the computer system 101 can predict the water demand amount based only on the correlation condition of the time point of predicting the water demand amount.
As described above, according to the embodiment of the present invention, the water demand prediction on the day of the predicted day is performed using the relevant condition data (weather, maximum temperature, day of the week, recent water demand time series, etc.) for the water demand on the predicted day, using the relevant condition data for the water demand on each day of the recent past predetermined period and the water demand actual data corresponding to the prediction target time period on each day, preferentially using the water demand actual value on the day (which may be multiple days) of the recent past predetermined period having the relevant condition data similar to the day of the predicted day. As a result, the water demand amount of the past day close to the change in the water demand amount is predicted with reference to the water demand amount, so that the water demand prediction with high accuracy matching the change in the water demand amount can be performed.
Claims (11)
1. A water demand prediction system for predicting a water demand amount per reference period for a predetermined water distribution area where water is distributed from a water distribution facility through a water distribution pipe network,
the device is provided with a memory and a controller;
the process executed by the controller based on the water demand prediction program includes:
acquiring an actual value of the water demand of the predetermined distribution area continuously measured by the water distribution facility from a monitoring control system of the water distribution facility, and recording the actual value, weather information and day information in the memory;
setting a correlation condition for use in the prediction of the water demand, the correlation condition including weather information, week information, and an actual value of the water demand for a predetermined period before the reference period;
calculating the correlation condition for each of the reference periods with respect to the actual value of the water demand recorded in the memory for the past predetermined period, and calculating a similarity between the correlation condition and the correlation condition for the reference period for predicting the water demand;
performing a prediction calculation of the water demand amount with reference to the similarity degree based on actual values of the water demand amount in at least 1 reference period of the plurality of reference periods of the past predetermined period; and
and transmitting a result of the prediction calculation to the monitoring control system.
2. The water demand forecasting system of claim 1,
the process of performing the prediction calculation of the water demand includes: the actual value of the water demand in the reference period having the higher similarity among the reference periods of the past predetermined period is used with higher priority for the prediction calculation of the water demand.
3. The water demand forecasting system of claim 2,
the process of performing the prediction calculation of the water demand uses a kernel ridge regression model to predict the water demand based on the similarity, the kernel ridge regression model using the correlation condition of the reference period for predicting the water demand as input data, and using the correlation condition and the water demand actual value of each reference period of the past predetermined period as learning data.
4. The water demand forecasting system of claim 1,
the processing for setting the relevant conditions to be used for the prediction of the water demand includes: the difference value at each time of the water demand for the predetermined period before the reference period is used as the correlation condition.
5. The water demand forecasting system of claim 1,
the processing for setting the relevant conditions to be used for the prediction of the water demand includes: the total value of the water demand for the predetermined period before the reference period is added to the correlation condition.
6. The water demand forecasting system of claim 1,
the processing for setting the relevant condition to be used for the prediction of the water demand includes at least one of: for the above-described day information, the previous day of a continuous rest of three or more days is regarded as friday, the first day of a continuous rest of three or more days is regarded as saturday, and the last day of a continuous rest of three or more days is regarded as sunday.
7. The water demand forecasting system of claim 1,
the controller displays at least one of: the water level conversion value is calculated by using the predicted water demand and the actual value of the water demand, an accumulated error between the predicted water demand and the actual value of the water demand from a predetermined time, and a water level conversion value of the distribution tank of the water distribution facility for the accumulated error.
8. The water demand forecasting system of claim 1,
the water demand prediction system is configured to allow a manager of the water demand prediction system to input the predetermined period before the reference period and the past predetermined period to the system via an input device.
9. The water demand forecasting system of claim 1,
the water demand prediction system is configured such that a manager of the water demand prediction system can input week information as the correlation condition to the system via an input device, the week information having a plurality of forms, and the manager can select a predetermined form from the plurality of forms.
10. The water demand forecasting system of claim 1,
the water demand prediction system is configured such that a manager of the water demand prediction system can input, via an input device, a plurality of actual values of the water demand as the relevant condition, and the manager can select a predetermined form from the plurality of forms.
11. A method for predicting, by a computer, a water demand amount per reference period for a prescribed water distribution area where water is distributed from a water distribution facility via a water distribution pipe network,
the computer performs the following processing:
acquiring the water demand of the specified water distribution area, the weather information and the week information of the water distribution area;
setting weather information and week information for the predetermined period and a water demand in a predetermined range before the reference period as conditions relating to the water demand for the reference period;
preferably, the similarity between the relevant condition in each reference period of the past predetermined period and the relevant condition in the reference period for predicting the water demand is calculated, and the water demand in the reference period for predicting the water demand is predicted by using the actual water demand value in the reference period having the condition with the high similarity.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2019-106590 | 2019-06-06 | ||
| JP2019106590A JP7399631B2 (en) | 2019-06-06 | 2019-06-06 | Water demand forecasting system and its method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112052978A true CN112052978A (en) | 2020-12-08 |
Family
ID=73609332
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010410657.0A Withdrawn CN112052978A (en) | 2019-06-06 | 2020-05-15 | Water demand prediction system and method |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP7399631B2 (en) |
| CN (1) | CN112052978A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10219758A (en) * | 1997-01-31 | 1998-08-18 | Toshiba Corp | Demand forecasting device |
| JP2000163394A (en) * | 1998-11-27 | 2000-06-16 | Toyo Electric Mfg Co Ltd | Method for predicting water distribution demand amount |
| JP2001055763A (en) * | 1999-08-18 | 2001-02-27 | Toshiba Corp | Wide-area optimum water operating device for waterwork plant |
| JP2001245432A (en) * | 2000-02-29 | 2001-09-07 | Hitachi Ltd | Device for demand forecast |
| JP2012079229A (en) * | 2010-10-05 | 2012-04-19 | Hitachi Ltd | Demand prediction device and water operation monitoring system |
| WO2016136323A1 (en) * | 2015-02-27 | 2016-09-01 | 株式会社Ｅ．Ｉ．エンジニアリング | Energy predict system, energy predict method, computer program for causing execution thereof, recording medium whereupon said program is recorded, and operation assistance system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2971276B2 (en) * | 1992-12-25 | 1999-11-02 | 株式会社東芝 | Water distribution pattern prediction device |
-
2019
- 2019-06-06 JP JP2019106590A patent/JP7399631B2/en active Active
-
2020
- 2020-05-15 CN CN202010410657.0A patent/CN112052978A/en not_active Withdrawn
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10219758A (en) * | 1997-01-31 | 1998-08-18 | Toshiba Corp | Demand forecasting device |
| JP2000163394A (en) * | 1998-11-27 | 2000-06-16 | Toyo Electric Mfg Co Ltd | Method for predicting water distribution demand amount |
| JP2001055763A (en) * | 1999-08-18 | 2001-02-27 | Toshiba Corp | Wide-area optimum water operating device for waterwork plant |
| JP2001245432A (en) * | 2000-02-29 | 2001-09-07 | Hitachi Ltd | Device for demand forecast |
| JP2012079229A (en) * | 2010-10-05 | 2012-04-19 | Hitachi Ltd | Demand prediction device and water operation monitoring system |
| WO2016136323A1 (en) * | 2015-02-27 | 2016-09-01 | 株式会社Ｅ．Ｉ．エンジニアリング | Energy predict system, energy predict method, computer program for causing execution thereof, recording medium whereupon said program is recorded, and operation assistance system |
Non-Patent Citations (1)
| Title |
|---|
| 刘世昌 等: "多尺度分析与数据互迁移相结合的短期电力负荷预测方法", 《计算机科学》, 31 July 2019 (2019-07-31), pages 318 * |
Also Published As
| Publication number | Publication date |
|---|---|
| JP7399631B2 (en) | 2023-12-18 |
| JP2020201609A (en) | 2020-12-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11093893B2 (en) | Apparatus and method of maintaining accurate perpetual inventory information | |
| Romeijnders et al. | A two-step method for forecasting spare parts demand using information on component repairs | |
| Yip et al. | Predicting the maintenance cost of construction equipment: Comparison between general regression neural network and Box–Jenkins time series models | |
| JP5287478B2 (en) | Power energy monitoring system | |
| EP3300205A1 (en) | Energy demand predicting system and energy demand predicting method | |
| US20210399575A1 (en) | Planning apparatus, controlling apparatus, method, and rcording medium | |
| US10846715B2 (en) | Energy operation apparatus, method, and system | |
| JP6134253B2 (en) | Energy consumption prediction system and energy consumption prediction method | |
| JP7099805B2 (en) | Predictors, prediction systems, prediction methods and programs | |
| JP2017034797A (en) | Baseline load estimation device and baseline load estimation program | |
| JP6520228B2 (en) | Power demand forecasting system, power demand forecasting method and program | |
| JP4511435B2 (en) | Water supply operation planning device | |
| JP2005012912A (en) | Operation planning method for power generation facility, and operation planning system for power generation facility | |
| JP4261471B2 (en) | Distribution system load assumption method and apparatus | |
| Matos et al. | Operating reserve adequacy evaluation using uncertainties of wind power forecast | |
| JP5476275B2 (en) | Water distribution plan prediction system, prediction method thereof, and program thereof | |
| CN112052978A (en) | Water demand prediction system and method | |
| CN117196250A (en) | Water plant control method and control equipment | |
| CN117217804A (en) | Intelligent pricing and inventory management method and system | |
| EP1270827B1 (en) | Water distribution amount predicting system | |
| EP4138008A1 (en) | Energy supply/demand operation guidance device and method for energy supply/demand operation in ironworks | |
| Heitmann et al. | Framework to decide for a volume flexible chemical plant during early phases of plant design | |
| JP2018041222A (en) | Power use amount prediction device and power use amount prediction method | |
| JP5561643B2 (en) | Demand forecasting device and water operation monitoring system | |
| KR101139161B1 (en) | Simulator for forecasting short-term demand of water |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20201208 |