JP2697233B2 - Neural network river flood forecasting device - Google Patents

Neural network river flood forecasting device

Info

Publication number
JP2697233B2
JP2697233B2 JP2075107A JP7510790A JP2697233B2 JP 2697233 B2 JP2697233 B2 JP 2697233B2 JP 2075107 A JP2075107 A JP 2075107A JP 7510790 A JP7510790 A JP 7510790A JP 2697233 B2 JP2697233 B2 JP 2697233B2
Authority
JP
Japan
Prior art keywords
layer
neural
river
neural network
rainfall
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.)
Expired - Fee Related
Application number
JP2075107A
Other languages
Japanese (ja)
Other versions
JPH03274031A (en
Inventor
良夫 泉井
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Mitsubishi Electric Corp
Original Assignee
Mitsubishi Electric Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Mitsubishi Electric Corp filed Critical Mitsubishi Electric Corp
Priority to JP2075107A priority Critical patent/JP2697233B2/en
Publication of JPH03274031A publication Critical patent/JPH03274031A/en
Application granted granted Critical
Publication of JP2697233B2 publication Critical patent/JP2697233B2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Landscapes

  • Measuring Volume Flow (AREA)

Description

【発明の詳細な説明】 [産業上の利用分野] この発明は、生体の神経細胞とその間の結合を模擬し
た神経回路網により河川の出水量の予測を行う神経回路
網河川出水量予測装置に関するものである。
Description: TECHNICAL FIELD The present invention relates to a neural network river water flow prediction device for predicting a water flow of a river by a neural network simulating a nerve cell of a living body and a connection therebetween. Things.

[従来の技術] 第6図は河川とその周辺を示す説明図で、(30)は河
川、(31)はダムを示す。河川出水量予測装置は、河川
(30)の周辺地点で降雨量x1,x2,x3,・・xNがあった
時、河川(30)に流出してくる量を示す河川出水量yが
どれだけあるかを予測するものであり、ダム(31)など
において、与えられた水資源をできるかぎり無駄にする
ことなく電気エネルギーに変換するためにも精度よく予
測することが要求される。第7図及び第8図は、例え
ば、「降雨パターンに対応した出水伝達関数による発電
用ダム流入量予測手法:一柳、小林、水野、松村、鬼
頭」,電気学会論文誌B,昭和63年1月号,第108巻,第
1号,第32頁〜第38頁に示された従来の河川出水量予測
装置における予測手法の手順の一部を示すフローチャー
トである。ステップ(21)では、各地点の各時刻におけ
る降雨量データX(t),X(t−1)...X(t−M)と
河川の出水量Y(t),Y(t−1)...Y(t−M)を学
習データとして多数収集する。降雨量データX(t)は
ある時刻tでの各地点における降雨量x1,x2,x3,・・xN
を表わしており、出水量Y(t)はある時刻tでの河川
出水量を表わしている。又、Mは収集したデータのうち
所定時刻より過去のデータの数であり、Nは収集したデ
ータの地点の数である。次にステップ(22)では、学習
データXi(X(t),X(t−1)...X(t−M);i=I
〜N)から有効雨量ri(r(t),r(t−1)...r(t
−M):i=1〜N)となる値を計算する。ステップ(2
3)では式(1),(2)によって有効雨量riと河川出
水量Yとの間の伝達関数G(Z)を推定する。
[Prior Art] FIG. 6 is an explanatory view showing a river and its surroundings, where (30) shows a river and (31) shows a dam. River flood amount prediction device, rainfall x 1 around a point of the river (30), x 2, x 3, when a · · x N, river flood amount indicating the amount flowing out into the river (30) It is to predict how much y is, and it is required to accurately predict given water resources to electric energy without wasting as much as possible in a dam (31) or the like. . FIGS. 7 and 8 show, for example, “Method of predicting dam inflow for power generation by water transfer function corresponding to rainfall patterns: Ichiyanagi, Kobayashi, Mizuno, Matsumura, Kito”, IEEJ Transactions on Journal B, 1988 It is a flowchart which shows a part of procedure of the prediction method in the conventional river flood forecasting apparatus shown in the monthly issue, the 108th volume, the 1st, and the 32nd page-the 38th page. In step (21), the rainfall data X (t), X (t-1)... X (t-M) at each time at each point, and the water discharge amounts Y (t), Y (t-1) of the river. ) ... A large number of Y (tM) are collected as learning data. Rainfall at each point of the rainfall amount data X (t) is the time t x 1, x 2, x 3, ·· x N
And the water output Y (t) represents the river water output at a certain time t. Further, M is the number of pieces of data collected before a predetermined time in the collected data, and N is the number of points of collected data. Next, in step (22), learning data X i (X (t), X (t−1)... X (t−M); i = I
~ N) to the effective rainfall r i (r (t), r (t-1) ... r (t
-M): Calculate the value of i = 1 to N). Step (2
3) In equation (1), we estimate a transfer function G (Z) between the effective rainfall r i and river flood amount Y by (2).

Q(Z)=G(Z)R(Z) ……(1) G(Z)=(b0+b1z-1+b2z-2+・・・) /(1+a1z-1+a2z-2+・・・) ……(2) ここで、R(Z):有効雨量r1のZ変換 Q(Z):河川出水量のZ変換 である。過去の各データを用いて伝達関数Gの係数b0,b
1,・・a1,a2・・を求めている。これで降雨量Xと河川
出水量Yとの関係を求めた。
Q (Z) = G (Z ) R (Z) ...... (1) G (Z) = (b 0 + b 1 z -1 + b 2 z -2 + ···) / (1 + a 1 z -1 + a 2 z -2 + ···) ...... (2 ) where, R (Z): effective rainfall r 1 of Z transform Q (Z): a Z transform of river flood amount. Coefficients b 0 and b of the transfer function G using each past data
1 , ... a 1 , a 2 ... Thus, the relationship between the rainfall X and the river flood Y was obtained.

つぎに、第8図にあるように、予測希望の降雨量とし
て、未知の降雨量Xあるいは降雨パターンをステップ
(24)で入力してステップ(25)で有効雨量rを計算
し、予め求めておいた伝達関数G(Z)を使って、河川
出水量Yを予測する(ステップ(26))。
Next, as shown in FIG. 8, an unknown rainfall amount X or a rainfall pattern is input as a desired rainfall amount in step (24), and an effective rainfall amount r is calculated in step (25) and obtained in advance. Using the set transfer function G (Z), the river flooding amount Y is predicted (step (26)).

[発明が解決しようとする課題] 従来の河川出水量予測装置は、以上のように構成され
ているので、予測精度が低いという問題点があった。
[Problem to be Solved by the Invention] Since the conventional river water discharge prediction device is configured as described above, there is a problem that the prediction accuracy is low.

この発明は上記のような従来の問題点を解決するため
になされたもので、予測精度の良い神経回路網河川出水
量予測装置を得ることを目的とするものである。
The present invention has been made in order to solve the conventional problems as described above, and an object of the present invention is to provide a neural network flood water predicting apparatus with high prediction accuracy.

[課題を解決するための手段] この発明に係る神経回路網河川出水量予測装置は、入
力層,中間層,出力層を構成し、生体の神経細胞を模擬
した複数の神経素子,及び神経素子の出力の各々に結合
重みを乗じて次層の神経素子へ出力することにより、シ
ナブスを模擬する結合素子を備え、入力層の神経素子に
過去の複数場所,複数時点の降雨量、出力層の神経素子
に降雨量に対応して計測した過去の河川出水量をあらか
じめ設定し,結合重みを学習方程式に基いて演算し、入
力層の神経素子に予測希望の降雨量を入力して結合重み
を用いて演算し、出力層の神経素子より河川出水量の予
測値を得るようにしたものである。
[Means for Solving the Problems] A neural network river water flow prediction device according to the present invention comprises an input layer, an intermediate layer, and an output layer, a plurality of neural elements simulating nerve cells of a living body, and a neural element. By multiplying each of the outputs by the connection weight and outputting the result to the neural element of the next layer, a connection element for simulating a synapse is provided. The past river flow rate measured according to the rainfall is set in advance in the neural element, the connection weight is calculated based on the learning equation, and the predicted rainfall is input to the neural element in the input layer, and the connection weight is calculated. The calculated value is used to obtain a predicted value of the river water flow from the neural element in the output layer.

[作用] この発明における神経回路網河川出水量予測装置は、
生体の神経細胞を模擬した神経素子とその間のシナプス
を模擬する結合重みから構成され、神経素子の出力が自
分も含めた他の神経素子の出力とその間の結合重みから
計算される神経回路網を利用している。結合重みが過去
の降雨量と河川出水量の具体例から学習アルゴリズムに
より計算され、現在より以前の降雨量等を入力すること
により河川の出水量を予測するようにし、神経回路網の
入力変数から出力変数への関数近似機能を用いることに
より、予測精度の向上を図るものである。
[Operation] The neural network river water discharge prediction device according to the present invention comprises:
A neural network consisting of neural elements that simulate biological nerve cells and connection weights that simulate synapses between them, and the output of a neural element is calculated from the output of other neural elements including itself and the connection weight between them We are using. The connection weight is calculated by a learning algorithm from a specific example of past rainfall and river flooding, and the river flooding is predicted by inputting the rainfall before the present, etc. from the input variables of the neural network. By using a function approximation function for an output variable, the prediction accuracy is improved.

[実施例] この発明の一実施例による神経回路網河川出水量予測
装置を第1図〜第3図により説明する。第1図,第2図
はこの発明の一実施例に係る処理を示すフローチャー
ト、第3図はこの実施例に係る神経回路網の構成を示す
説明図である。この実施例における神経回路網は第3図
に示すように、N*(M+1)個の入力層(10a)、L
個の中間層Z(10b)、1個の出力層(10c)の神経素子
で構成されており、かつ、入力層、中間層、出力層は各
々1層で構成されているものとする。この神経素子(1
0)は生体の神経細胞を模擬している。(11)は結合素
子で、神経素子(10)を模擬するものであり、神経素子
(10)の出力の各々に結合重みを乗じて次層の神経素子
(10)へ出力することによって、シナプスを模擬する。
ここでNは河川の周辺で、データを収集した地点の数、
Mは現在からさかのぼって収集した過去のデータの数で
ある。
[Embodiment] An apparatus for estimating a river flow rate of a neural network according to an embodiment of the present invention will be described with reference to FIGS. 1 and 2 are flowcharts showing processing according to one embodiment of the present invention, and FIG. 3 is an explanatory diagram showing the configuration of a neural network according to this embodiment. As shown in FIG. 3, the neural network in this embodiment has N * (M + 1) input layers (10a), L
It is assumed that each of the neural elements is composed of one intermediate layer Z (10b) and one output layer (10c), and each of the input layer, the intermediate layer, and the output layer is composed of one layer. This neural element (1
0) simulates a living nerve cell. (11) is a connection element which simulates the nerve element (10), and multiplies each of the outputs of the nerve element (10) by the connection weight and outputs the result to the next layer of the nerve element (10), thereby obtaining a synapse. To simulate
Where N is the number of points around the river where data was collected,
M is the number of past data collected from the present.

以下、この実施例における処理について説明する。第
1図に示す処理は、予め得られた多数場所の各時刻にお
ける降水量と河川出水量を学習データとして、神経回路
網の結合重みを学習する処理である。即ち、ステップ
(1)では、各地点の各時刻における降雨量データX
(t),X(t−1),,,X(t−M)と河川の出水量Y
(t),Y(t−1),,,Y(t−M)と学習データとして
多数収集する。この降雨量データX(t)はある時刻t
での各地点における降雨量x1,x2,x3,・・xNを表わして
おり、出水量Y(t)はある時刻tでの河川出水量を表
わしている。ステップ(2)では、学習データXi(X
(t),X(t−1),,,X(t−M);i=1〜N)を神経
回路網の入力層の神経素子(10a)に入力し、出力層(1
0c)から出力される河川出水量Yとして所望の出力が得
られるように、結合重みを演算する。出力層(10c)か
ら所望の河川出水量Yが得られれば、結合重みが決定さ
れ、学習は終了となる。
Hereinafter, processing in this embodiment will be described. The process shown in FIG. 1 is a process of learning connection weights of a neural network using, as learning data, the rainfall amount and river flooding amount at each time at a number of locations obtained in advance. That is, in step (1), the rainfall data X
(T), X (t-1) ,,, X (t-M) and river water output Y
(T), Y (t-1),..., Y (t−M) and a large number are collected as learning data. This rainfall data X (t) is at a certain time t
And rainfall x 1, x 2, x 3 , the · · x N represents at each point in, Izumi amount Y (t) represents the river flood amount at certain time t. In step (2), the learning data X i (X
(T), X (t-1) ,,, X (t-M); i = 1 to N) are input to the neural element (10a) of the input layer of the neural network, and the output layer (1
The connection weight is calculated so that a desired output can be obtained as the river flooding amount Y output from 0c). If the desired river flooding amount Y is obtained from the output layer (10c), the connection weight is determined, and the learning ends.

次に、第2図は河川出水量Yを予測する処理における
フローチャートである。第1図に示したようにして学習
した神経回路網に、予測希望の未知の降水量データXを
設定し(ステップ(3))、神経素子の入力層(10a)
から入力する(ステップ(4))。神経回路網を実行す
ると、出力層(10c)より河川出水量Yを出力として得
る。この出力値Yが未知の降雨量Xの時に予測した河川
出水量Yとなる(ステップ(5))。
Next, FIG. 2 is a flowchart in the process of estimating the river flooding amount Y. The unknown precipitation data X which is desired to be predicted is set in the neural network learned as shown in FIG. 1 (step (3)), and the input layer of the neural element (10a).
(Step (4)). When the neural network is executed, the output Y of the river is obtained from the output layer (10c). This output value Y becomes the river flooding amount Y predicted at the time of the unknown rainfall amount X (step (5)).

なお、第3図に示した神経回路網での学習過程におけ
る動作は以下のようである。入力層(10a)から入力し
たデータは中層層(10b)を介して、出力層(10c)に伝
搬されていく。定量的には次のようになる。dk pを出力
層(10c)における、第p番目の学習データの第k番目
の値、uh j,vh jを第h層のj番目の神経素子の内部状態
と出力値、wh jiを第h層の第i番目の神経素子と第h+
1層における第j番目の神経素子との間の結合重みとす
る。第3図の構成では出力層(10c)は1層1個である
ので、河川出水量Yはd1 1に対応し、 d1 1=v3 1 である。この時、各変数の関係は式(1),(2)のよ
うになる。
The operation in the learning process in the neural network shown in FIG. 3 is as follows. Data input from the input layer (10a) is propagated to the output layer (10c) via the middle layer (10b). It is quantitatively as follows. d k p is the k-th value of the p-th learning data in the output layer (10c), u h j and v h j are the internal state and output value of the j-th neural element in the h-th layer, w h ji is the i-th neural element of the h-th layer and the h +
The weight is the connection weight with the j-th neural element in one layer. Since in the third diagram of the output layer (10c) is one single layer, river flood amount Y corresponds to d 1 1, a d 1 1 = v 3 1. At this time, the relationship between the variables is as shown in equations (1) and (2).

uh j=Σwh-1 ji vh-1 i ……(1) vh j=g(uh j) ……(2) ここで、関数g(*)は微分可能で非減少な関数であ
ればよく、一例として式(3)とする。
u h j = Σw h-1 ji v h-1 i (1) v h j = g (u h j ) (2) where the function g (*) is a differentiable and non-decreasing function It suffices to use Equation (3) as an example.

g(x)=1/(1+exp(−x)) ……(3) さらに、結合重みwh jiは式(4)の学習則で決定され
る。出力層における学習データ、即ち希望値と、神経回
路網によって実際に得られた値で定義される2乗誤差に
関する最急降下法で逐次的に決定される。神経回路網の
層の数をHとすると、2乗誤差E(エネルギーと言う事
もある)は、 E=(1/2)ΣΣ(vH k−dk p ……(4) となる。
g (x) = 1 / ( 1 + exp (-x)) ...... (3) In addition, link weight w h ji is determined by learning rule of equation (4). The learning data in the output layer, that is, the desired value, is sequentially determined by a steepest descent method with respect to a square error defined by a value actually obtained by the neural network. Assuming that the number of layers in the neural network is H, the square error E (sometimes called energy) is: E = (1/2) ΣΣ (v H k −d k p ) 2 (4) Become.

また、結合重みの逐次変更はα,βを適当なパラメー
タとし、モーメント法を使用した場合には次の学習方程
式(5)で実行できる。
The sequential change of the connection weight can be executed by the following learning equation (5) when α and β are used as appropriate parameters and the moment method is used.

d2 wh ji/dt2+(1−α)dwh ji/dt =−β ∂E/∂wh ji ……(5) 結合重みwh jiが決定すれば、学習の過程は終了であ
り、この結合重みwh jiと予測希望の降雨量を用いて式
(1),(2)を計算して、河川出水量の予測値
(v3 1)を得る。
If d 2 w h ji / dt 2 + (1-α) dw h ji / dt = -β ∂E / ∂w h ji ...... (5) combining weights w h ji is determined, the process of learning the end Yes, equation (1) using the rainfall prediction hope this link weight w h ji, by calculating the (2) to obtain the predicted value of river flood amount (v 3 1).

このように、神経回路網の入力変数から出力変数への
関数近似機能を用いることにより、予測精度が向上でき
る。
Thus, the prediction accuracy can be improved by using the function approximation function from the input variables to the output variables of the neural network.

又、第3図に示す神経回路網は一例であってこれに限
るものではなく、例えば中間層(10b)は1層でなくて
もよい。又、各層を構成する個数は上記実施例に限るも
のではないが、この個数が多いと計算時間がかかり、少
ないと収束しにくくなることがある。
Further, the neural network shown in FIG. 3 is an example and is not limited to this. For example, the intermediate layer (10b) may not be a single layer. Further, the number of layers constituting each layer is not limited to that in the above embodiment, but if the number is large, calculation time is required, and if the number is small, convergence may be difficult.

第4図はこの発明の他の実施例に係る神経回路網の構
成を示す説明図である。1つの四角が神経素子(10)を
表わし、入力層(10a)では縦方向に時間の変化に応じ
て現在tから過去へさかのぼってt−Mまで並んでお
り、横方向にはデータの収集場所に応じて並んでいる。
また中間層(10b)は横方向に1層の個数分、縦方向に
層の数だけ並んでいる。この例の中間層(10b)は3層
で構成している。又、図に示すように、中間層の各神経
素子(10b)が、過去の時刻の降水量データにしか結合
しないように構成になっている。例えば、第2層目の中
間層Aは矢印A1(t−1)からA2(t−M)の範囲の入
力層(10a)と結合している。この構成は全ての神経素
子を接続する必要がない。
FIG. 4 is an explanatory diagram showing a configuration of a neural network according to another embodiment of the present invention. One square represents a neural element (10), and in the input layer (10a), it is arranged from the current t to the past in the vertical direction from time t to time tM according to a change in time, and a data collection location in the horizontal direction. Lined up according to.
The intermediate layers (10b) are arranged in the horizontal direction by the number of one layer and in the vertical direction by the number of layers. The intermediate layer (10b) in this example is composed of three layers. Also, as shown in the figure, each neural element (10b) in the intermediate layer is configured so as to be coupled only to precipitation data at a past time. For example, the intermediate layer A of the second layer are attached from an arrow A 1 (t-1) input layer ranging A 2 (t-M) and (10a). This configuration does not require connection of all neural elements.

又、第5図に示すように、内部状態を表す神経素子
(10d)を導入して、過去の時点における降水量を神経
回路網の内部に織り込んでもよい。
Further, as shown in FIG. 5, a neural element (10d) representing an internal state may be introduced to incorporate the precipitation amount at a past time into the neural network.

[発明の効果] 以上説明したように、この発明によれば、入力層,中
間層,出力層を構成し、生体の神経細胞を模擬した複数
の神経素子、及び神経素子の出力の各々に結合重みを乗
じて次層の神経素子へ出力することによって、シナプス
を模擬する結合素子を備え、入力層の神経素子に過去の
複数場所,複数時点の降雨量、出力層の神経素子に降雨
量に対応して計測した過去の河川出水量をあらかじめ設
定し、結合重みを学習方程式に基いて演算し、入力層の
神経素子に予測希望の降雨量を入力して結合重みを用い
て演算し、出力層の神経素子より河川出水量の予測値を
得るようにしたので、神経回路網の関数近似機能によ
り、予測精度が向上するという効果が得られた。
[Effects of the Invention] As described above, according to the present invention, an input layer, an intermediate layer, and an output layer are configured and connected to a plurality of neural elements simulating nerve cells of a living body, and to outputs of the neural elements. By multiplying the weight and outputting to the neural element of the next layer, a coupling element that simulates a synapse is provided, and the neural element of the input layer is used for rainfall at multiple locations in the past, at multiple times, and the neural element of the output layer is used for rainfall. Correspondingly measured past river flooding is set in advance, the connection weight is calculated based on the learning equation, the desired rainfall is input to the neural element in the input layer, the calculation is performed using the connection weight, and the output is calculated. Since the predicted value of the river water flow was obtained from the neural elements in the layer, the effect of improving the prediction accuracy was obtained by the function approximation function of the neural network.

【図面の簡単な説明】[Brief description of the drawings]

第1図はこの発明の一実施例による神経回路網河川出水
量予測装置に係る学習の処理過程を示すフローチャー
ト、第2図は一実施例に係り、学習した神経回路網によ
る河川出水量の予測処理を示すフローチャート、第3図
は一実施例に係る神経回路網の構成を示す説明図、第4
図,第5図は各々この発明の他の実施例による神経回路
網河川出水量予測装置に係る神経回路網の構成を示す説
明図、第6図は河川出水量予測装置の概念を示す説明
図、第7図は従来の河川出水量予測装置に係る伝達関数
を求める処理のフローチャート、第8図は従来の河川出
水量予測装置の予測方法を示すフローチャートである。 (10a)……入力層神経素子、(10b)……中間層神経素
子、(10c)……出力層神経素子、(11)……結合素
子、(30)……河川、(31)……ダム。 なお、図中、同一符号は同一、又は相当部分を示す。
FIG. 1 is a flowchart showing a learning process according to an embodiment of a neural network river water flow predicting apparatus according to one embodiment of the present invention. FIG. 2 is a flow chart showing prediction of river water flow by a learned neural network according to one embodiment. FIG. 3 is a flowchart showing processing, FIG. 3 is an explanatory diagram showing the configuration of a neural network according to one embodiment, FIG.
FIG. 5 and FIG. 5 are explanatory diagrams each showing a configuration of a neural network according to a neural network river flood prediction device according to another embodiment of the present invention, and FIG. 6 is an explanatory diagram showing a concept of a river flood forecast device. FIG. 7 is a flowchart of a process for obtaining a transfer function according to a conventional river flood prediction device, and FIG. 8 is a flowchart showing a prediction method of the conventional river flood prediction device. (10a) ... input layer neural element, (10b) ... hidden layer neural element, (10c) ... output layer neural element, (11) ... coupling element, (30) ... river, (31) ... dam. In the drawings, the same reference numerals indicate the same or corresponding parts.

Claims (1)

(57)【特許請求の範囲】(57) [Claims] 【請求項1】入力層,中間層,出力層を構成し、生体の
神経細胞を模擬した複数の神経素子,及び上記神経素子
の出力の各々に結合重みを乗じて次層の神経素子へ出力
することにより、シナプスを模擬する結合素子を備え、
上記入力層の神経素子に過去の複数場所,複数時点の降
雨量、上記出力層の神経素子に上記降雨量に対応して計
測した過去の河川出水量をあらかじめ設定し、上記結合
重みを学習方程式に基いて演算し、上記入力層の神経素
子に予測希望の降雨量を入力して上記結合重みを用いて
演算し、上記出力層の神経素子より河川出水量の予測値
を得るようにした神経回路網河川出水量予測装置。
1. An input layer, an intermediate layer, and an output layer, wherein a plurality of neural elements simulating biological nerve cells and each of the outputs of the neural elements are multiplied by a connection weight and output to a neural element of a next layer. By providing a coupling element that simulates a synapse,
The neural elements in the input layer are set in advance to past rainfall amounts at a plurality of locations and at multiple points in time, and the neural elements in the output layer are set in advance to past river floodwaters measured in accordance with the rainfall amounts. A neural network in which the predicted rainfall amount is input to the neural element of the input layer and is calculated using the connection weight, and a predicted value of river water flow is obtained from the neural element of the output layer. Network water flow forecasting device.
JP2075107A 1990-03-23 1990-03-23 Neural network river flood forecasting device Expired - Fee Related JP2697233B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2075107A JP2697233B2 (en) 1990-03-23 1990-03-23 Neural network river flood forecasting device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2075107A JP2697233B2 (en) 1990-03-23 1990-03-23 Neural network river flood forecasting device

Publications (2)

Publication Number Publication Date
JPH03274031A JPH03274031A (en) 1991-12-05
JP2697233B2 true JP2697233B2 (en) 1998-01-14

Family

ID=13566619

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2075107A Expired - Fee Related JP2697233B2 (en) 1990-03-23 1990-03-23 Neural network river flood forecasting device

Country Status (1)

Country Link
JP (1) JP2697233B2 (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4590794B2 (en) * 2001-07-13 2010-12-01 富士電機システムズ株式会社 Hydroelectric power generation prediction device, hydroelectric power generation prediction method
CN102032935B (en) * 2010-12-07 2012-01-11 杭州电子科技大学 Soft measurement method for sewage pumping station flow of urban drainage converged network
JP7354213B2 (en) * 2021-12-09 2023-10-02 八千代エンジニヤリング株式会社 Model generation method and inflow prediction system

Also Published As

Publication number Publication date
JPH03274031A (en) 1991-12-05

Similar Documents

Publication Publication Date Title
Vaziri Predicting Caspian Sea surface water level by ANN and ARIMA models
Huang et al. Neural network modeling of salinity variation in Apalachicola River
Kisi et al. River suspended sediment estimation by climatic variables implication: Comparative study among soft computing techniques
Aziz et al. Application of artificial neural networks in regional flood frequency analysis: a case study for Australia
Zounemat-Kermani et al. Estimation of current-induced scour depth around pile groups using neural network and adaptive neuro-fuzzy inference system
Yitian et al. Modeling flow and sediment transport in a river system using an artificial neural network
More et al. Forecasting wind with neural networks
Thirumalaiah et al. River stage forecasting using artificial neural networks
Zounemat-Kermani et al. Using adaptive neuro-fuzzy inference system for hydrological time series prediction
Talebizadeh et al. Uncertainty analysis for the forecast of lake level fluctuations using ensembles of ANN and ANFIS models
Feng et al. The practical research on flood forecasting based on artificial neural networks
CN109711617B (en) Medium-and-long-term runoff prediction method based on BLSTM deep learning
KR102236678B1 (en) Method and device for forecasting flood based on data analyzing
Pan et al. State space neural networks for short term rainfall-runoff forecasting
Vafakhah Comparison of cokriging and adaptive neuro-fuzzy inference system models for suspended sediment load forecasting
Bouzeria et al. Using artificial neural network (ANN) for prediction of sediment loads, application to the Mellah catchment, northeast Algeria
Ashraf et al. Artificial neural network based models for forecasting electricity generation of grid connected solar PV power plant
Lai et al. Intelligent weather forecast
JP2697233B2 (en) Neural network river flood forecasting device
Demirkaya Deformation analysis of an arch dam using ANFIS
Mohamed et al. Suspended sediment concentration modeling using conventional and machine learning approaches in the Thames River, London Ontario
Bellamine et al. Modeling of complex dynamic systems using differential neural networks with the incorporation of a priori knowledge
Deng et al. Hybrid metaheuristic machine learning approach for water level prediction: A case study in Dongting Lake
Tsai et al. Prediction of storm-built beach profile parameters using neural network
Mynett Hydroinformatics and its applications at Delft Hydraulics

Legal Events

Date Code Title Description
LAPS Cancellation because of no payment of annual fees