JPH0131572B2 - - Google Patents

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a river flow rate prediction device capable of predicting river flow rate with higher accuracy for the purpose of operating a spillway valve installed in a river weir. .

[Technical background of the invention and its problems]

In recent years, in river basins, spillway valves installed at river weirs have been operated at river flood levels. In this case, since the operation of the spillway valve is performed according to the river flow rate, it is necessary to accurately grasp the river flow rate. Conventionally, there are various river runoff analysis methods, such as the unit diagram method and runoff function method that use a linear response function for short-term runoff, and the storage function method and tank model method that introduce a storage function, but all of these methods require engineers to It is used based on experience accumulated through manual calculations.

Therefore, with the recent spread of computers, it is being considered to mechanize analysis, but it is difficult to clarify the human judgment process, especially the selection part, and so only the four simple arithmetic operations have been mechanized. is the current situation.

[Purpose of the invention]

The present invention was made in view of the above circumstances, and its purpose is to be able to predict the river flow rate for adjusting the spillway valve installed in the river weir with high accuracy without requiring the experience of engineers. An object of the present invention is to provide a river flow rate prediction device.

[Summary of the invention]

In order to achieve the above object, the present invention includes a flow meter installed in a target river, a rain gauge installed in the target river basin, time series data of river flow measured by the flow meter, and a rainfall meter. The model coefficients of the continuity equation and the equation of motion are calculated using the storage function method using the continuity equation that holds between the river flow rate and the rainfall amount and the equation of motion that indicates the apparent basin storage amount. and a runoff model creation device that uses a goal programming method to calculate model coefficients that reduce the difference between the measured value and the calculated flow rate time curve obtained from these model coefficients, and a runoff model that inputs time series data of predicted rainfall. It is characterized by comprising a flow rate prediction device that calculates flow rate predicted value time series data for adjusting a spillway valve provided downstream of the flow meter using each model coefficient determined by the creation device. shall be.

[Embodiments of the invention]

An embodiment of the present invention shown in the drawings will be described below. FIG. 1 shows an example of the configuration of a symmetrical system to which the river flow rate prediction device of the present invention is applied. In the figure, 1 is the river in the target river basin 2,
A flood control dam 3 has been constructed on the upstream side.
Further, numeral 4 is a discharge flow meter for measuring the discharge flow rate from this flood control dam 3, and 5 is a spillway valve provided at the weir of the river 1, which can be operated by an operating motor 6. Furthermore, 7 is a rainfall meter that measures the amount of rainfall in the target river basin 2 at regular intervals, and 8 is a flow meter that also measures the flow rate of the river 1 at regular intervals.

On the other hand, 9 is a river flow rate prediction calculation device, which is composed of a runoff model creation device 10 and a flow rate prediction device 11. The runoff model creation device 10 receives the time-series data of the river flow rate measured by the flow meter 8 and the rainfall amount measured by the rainfall meter 7, and calculates the continuity equation and apparent relationship between the river flow rate and the rainfall amount. The model coefficients for the continuity equation and the equation of motion are calculated using the storage function method using the equation of motion that indicates the watershed storage amount, and the model coefficients are used to reduce the difference between the measured value and the calculated value for the flow rate time curve obtained from these model coefficients. is determined by the goal planning method. In addition, the flow rate prediction device 11 inputs time series data 12 of predicted rainfall based on weather forecasts, and uses each model coefficient obtained by the runoff model creation device 10 to predict the flow rate for the spillway installed downstream of the flow meter 8. It calculates flow rate predicted value time series data for adjusting the valve 5, and an operation command for the valve operation electric motor 6 is issued based on this output.

The time-series data measured by the rain gauge 7 and the flow meter 8 described above are recorded and stored as shown in FIG. 2.

Next, the operation of such a river flow rate prediction device will be explained in detail using FIG. 3.

First, in step 1, time series data of rainfall amount r and river flow rate q from the rain gauge 7 and flow meter 8 are input. Here, the data is N sets (number + sets) of rainfall over several years.

Next, in step 2, based on the data input in step 1, the following storage function calculation is performed. In other words, it is assumed that the following continuity equation holds between the value of the rainfall gauge 7 in the target watershed and the average rainfall intensity r, and the river flow rate q.

ds _l /dt=・r _l −(q−q _i )……(1) Furthermore, it is assumed that the following equation holds true as the equation of motion.

s _l = K・(q−q _i ) ^p ...(2) In equations (1) and (2), is the inflow coefficient, q _i is the initial flow rate, and r _l is the rainfall with T _l as the flow delay time. amount r
time series data delayed by T _l , and s _l indicates the apparent watershed storage amount according to the right-hand side of equation (1), K is a proportionality constant, and P is a power constant. Therefore, , T _l ,
It is sufficient to determine four coefficients, K and P.

In this case, equations (1) and (2) show that r is treated as r _l and q is treated as (q-q _i ), so first we calculate this (q-q _i ). do. next,
Perform the r _l calculation with T _l = T (0, 1, ..., 10), and calculate q = q _c = 0.1 x (q _nax - q _i ) (constant; q _nax is the maximum flow rate) and (q - q _i ), and let the intersection times be t ₁ and t ₂ (t ₁ <t ₂ ).

= _t2 〓 ^t1 (q−q _i )・Δt／ _t2 〓 ^t1 (r _l )・Δt……(3) When Δt is a constant period time, the coefficient can be obtained by equation (3) (>0). From equation (1), integrate both sides with respect to t s _l (t) = s _l (t ₁ ) = _t 〓 ^t1・r _l・Δt − _t 〓 ^t1 (q−q _i )・Δt Here, s _l Set (t ₁ )=0 and use the result,
Calculate the relationship between the storage function s _l and the flow rate (q−q _i ).
In this case, a curve in which the water increase period and the water decrease period are most similar is drawn, a case is selected that can be regarded as a single-valued function, and T _l =T ^* at that time is set as the delay time. Regarding the storage function curve s _l −(q−q _i ) when T _l , the coefficients K and P of equation (2) are determined using the least squares method. Therefore, as the above result, four coefficients, T _l , K, and P are obtained, so divide the storage function curve s _l - (q - q _i ) into 20 and calculate the flow rate using the rainfall amount r as input by polygonal line approximation. , perform flow tracking calculations (flow time curve).

Next, in step 3, the two coefficients K and P are polished. In other words, two coefficients are selected that will minimize the difference between the measured value and the calculated value for the flow rate time curve. First, regarding the setting of the objective function, it is assumed that it is composed of two types: maximum flow rate and water increase duration. The flow rate waveform is represented by the time on the horizontal axis and the flow rate on the vertical axis, and includes the actual measured value of the maximum flow rate (t _x , q _x ) and the calculated value (T _x ,
Q _x ), then the partial objective function I _q regarding the maximum flow rate
is defined by the following equation.

I _x = |q _x −Q _x | ...(4) Here, || is the absolute value symbol, and the time term is
It is abbreviated as t _x = T _x .

On the other hand, regarding the water increase duration, the maximum flow rate is j/
L times the flow rate q _j (=j/L・(q _x −q _i )+q _i ; j=1,
2, ..., (L-1)).
Let the function points of the measured values be (t ^s _j , q ^s _j ) and the end points (t ^E _j , q ^E _j ), and the respective points of the calculated values be (T ^s _j , Q ^s _j ), (T ^E _j , Q ^E _j )
As, the partial objective function I _j regarding the duration of water rise
is defined as I _j = |(t ^E _j −t ^s _j )−(T ^E _j −T ^s _j )| ……(5).

Since the physical dimensions of these two partial objective functions I _x and I _j are different [m ³ /h] and [h], respectively, the objective planning method assumes that the two quantities are dimensionless numbers and the actual measured values are the target values. We will treat it as such. Ultimately, the objective function I is defined by the following equation.

I=α・(Q _x /q _x −1) ² + _L-1 〓 ^j=1 β _j・{(T ^E / _j − T ^s / _j ) / (t ^E / _j − t ^s / _j )− 1} ²
...(6) Here, α and β _j (j=1, 2, ..., (L-1))
is the weighting coefficient, α+ _L-1 〓 ^j=1 β _j =1. And this (6)
A combination of two coefficients K and P is found so as to minimize the objective function of the equation. This search is performed by constructing a lattice point using two coefficients K and P, and targeting that lattice point. The initial values of the second coefficients K and P are the values already obtained in step 2, but the first step targets the entire area that satisfies the conditions attached to the second coefficients (K>0;1>P>0). , find the grid center point (K _c , P _c ) of the subarea to be searched.

First, a sensitivity analysis is performed on the coefficient P with respect to the objective function, and a quadratic curve is fitted on the P-I plane. In this case, obtain the value P _c of P that is the minimum value I _po .

I (P) = a (P - P _c ) ² + I _po , (a > 0) ... (7) Also, regarding the coefficient K, the value of K that similarly becomes the minimum value I _ko as P = _{P c} get. As a result, the search grid center point (K _c , P _c ) is obtained, so finer grid points are created using the following equation.

K=K _c {1+ΔK(m−m ₀ )} P=P _c {1+ΔP(n−n ₀ )} …(8) Here, m=1, 2, …, (2m ₀ −1); n=
1, 2, ..., (2n ₀ -1) [m ₀ and n ₀ are positive integers], and ΔK and ΔP are constants (for example, 0.01). However, if the condition in equation (7) is not satisfied (for example, a≦0 or
When P _c ≧1), the values obtained by the conventional method are considered as K _c and P _c . Also, for all the lattice points in equation (8), the objective function I
is calculated, and the combination of two coefficients (K ^* , P ^* ) that gives the minimum value I ^* is obtained as a result of polishing. Steps 2 and 3 are completed by performing calculations a number of times equal to the number of input data sets N, so 4 of the N sets are completed.
Coefficients ( _i , T _li , K ^* _i , P ^* _i ; = 1, 2, ... N) are obtained, but the ability to exclude bad data (t ₂ indefinite,
_t2 〓 ^t1 r _l ·Δt=0, etc.), so the number of sets of four coefficients that satisfy the conditions is N' sets (N'≦N).

Next, in step 4, the upper limit is unlimited due to the conditions attached to 2 coefficients K ^* (>0, K ^* >0), so there is a possibility that it may contain too large a value, so we first calculate the coefficient _i . _{1 2} ≦... _N ′, arrange them in descending order, exclude the largest [0.1×N′] ([ ] is a Gauss symbol), and find the average value _ta .

K _ta is obtained in the same manner for the coefficient K ^* .

Next, in step 5, an operation for excluding abnormal values of two coefficients, K ^*, is performed using _ta and _Kta . _i is
When it is larger than M times _ta (for example, M=10), the set i of four coefficients is removed. The same applies when K ^* _i is larger than M times _Kta . moreover,
Similarly, when _i is larger than M' times _ta (M'<M; for example, M'=3) and K ^* _i is larger than M' times K _ta , it is also excluded. In this way, the N′ groups become N″ groups (N″N′).

Next, in step 6, the average value of each of the N'' sets of 4 coefficients is calculated.This is the 4 coefficients ( _a , _Tla , K ^* _a , P ^* _a ) that can be used throughout the year.
It is.

Finally, in step 7, using the four coefficients obtained in step 6, predicted flow rate time series data is calculated based on the predicted rainfall time series data of the weather forecast. In this case, since it is necessary to handle values beyond the range of the input data in step 1, the storage function curve is expanded. Then, the flow rate section up to the maximum flow rate of the input data is divided into 20 equal parts and approximated by a broken line, but by extending the straight line of the last 20th section, it becomes possible to target the entire flow rate area.

In addition, Figure 4 a and b show examples of flow rate curves when using the device of the present invention. The cases in which trimmed average coefficients are used are shown, with the solid line being the measured value and the broken line being the calculated value.

In this way, a runoff model that can be used throughout the year is created using time-series data of rainfall r and river flow rate q measured by rainfall gauge 7 and flowmeter 8 installed in the target river basin 2. A river flow rate prediction device 9 is constructed from a model creation device 10 and a flow rate prediction device 11 that receives time series data 12 of predicted rainfall as input, predicts flow rate using the runoff model, and outputs time series data of predicted flow values. It is composed of

Therefore, it is possible to predict the flow rate of a river with extremely high accuracy without requiring the experience of engineers as in the past. In order to increase the use of river surface water, it is necessary to reduce the amount released from spillways as much as possible, but this device predicts river flow from rainfall data, which is a leading variable of river flow. Therefore, it is possible to respond to momentary changes in rainfall to prevent flooding and to reduce the amount of discharge as much as possible. In addition, it will be possible to utilize past accumulated data, utilize weather forecasts that have become more accurate using the latest technology, and improve the utilization of installed computers.

Note that the present invention is not limited to the above embodiments.

The apparatus of the present invention can operate as long as there is one rainfall gauge in the target river basin, but of course, the more rainfall gauges there are, the higher the accuracy of the rainfall amount, which is preferable. For example, in the case of the number R, the Thiessen polygon (a polygon created around each observation station by marking the rainfall observation position on the map of the target basin and drawing a perpendicular bisector on the line connecting each observation time) ), and calculate the average rainfall amount r from the rainfall amount r _i at the rainfall observatory i and the polygon area s _i .
^is calculated using the following formula (i = ₁ ^, _{2 , ...} , _R ) It can be used for rainfall conditions (e.g. typhoons, thunderstorms, long rains, etc.), but when it is possible to increase the accuracy for special forms, it is also possible to create a dedicated runoff model and estimate it. . in this case,
Rainfall type division (categorization) requires decades of past data, but the feature of categorization is improved accuracy, and it is still a device that can be used throughout the year.

In addition, the present invention can be modified and implemented in various ways without changing the gist thereof.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to provide a river flow rate prediction device that can predict a river flow rate for adjusting a spillway valve provided in a river weir with high accuracy without requiring the experience of an engineer.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram showing one embodiment of the present invention, and FIG.
Figure 3 shows the time change curve of rainfall and flow rate.
This figure is a flowchart showing the operation of the river flow rate prediction device in FIG. 2, and FIGS. 4a and 4b are diagrams showing flow rate curves when the device of the present invention is used. 1... River, 2... Target river basin, 3... Flood control dam, 4... Outflow flow meter, 5... Spillway valve,
6... Operating motor, 7... Rain gauge, 8... Flow meter, 9... River flow rate prediction calculation device, 10... Runoff model creation device, 11... Flow rate prediction device.

Claims (1)

[Claims] 1. A flow meter installed in the target river, a rainfall gauge installed in the target river basin, time series data of the river flow rate measured by the flow meter, and rainfall amount measured by the rain gauge. Then, each model coefficient of the continuity equation and the equation of motion is determined by the storage function method using the continuity equation that holds true between the river flow rate and the amount of rainfall and the equation of motion that indicates the apparent basin storage amount, and these model coefficients are calculated using the storage function method. A runoff model creation device that calculates model coefficients that reduce the difference between actual measured values and calculated values for the flow rate time curve obtained from the coefficients using a goal planning method, and a runoff model creation device that inputs time series data of predicted rainfall amount and calculates them using the runoff model creation device. and a flow rate prediction device that calculates flow rate predicted value time series data for adjusting a spillway valve provided downstream of the flow meter using each of the calculated model coefficients. Flow rate prediction device.