CN113051833A - Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism - Google Patents

Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism Download PDF

Info

Publication number
CN113051833A
CN113051833A CN202110390930.2A CN202110390930A CN113051833A CN 113051833 A CN113051833 A CN 113051833A CN 202110390930 A CN202110390930 A CN 202110390930A CN 113051833 A CN113051833 A CN 113051833A
Authority
CN
China
Prior art keywords
flood
reservoir
deep learning
uncertainty
reserved
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.)
Granted
Application number
CN202110390930.2A
Other languages
Chinese (zh)
Other versions
CN113051833B (en
Inventor
张晓琦
刘攀
许继军
陈进
王永强
洪晓峰
袁喆
谢帅
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.)
Changjiang River Scientific Research Institute Changjiang Water Resources Commission
Original Assignee
Changjiang River Scientific Research Institute Changjiang Water Resources Commission
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 Changjiang River Scientific Research Institute Changjiang Water Resources Commission filed Critical Changjiang River Scientific Research Institute Changjiang Water Resources Commission
Priority to CN202110390930.2A priority Critical patent/CN113051833B/en
Publication of CN113051833A publication Critical patent/CN113051833A/en
Application granted granted Critical
Publication of CN113051833B publication Critical patent/CN113051833B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/04—Constraint-based CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/08—Fluids

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention provides a flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism, which comprises the following steps: step 1, selecting a deep learning model for constructing a mapping relation of flood uncertainty-reservoir flood control storage capacity; step 2, screening and describing parameters of the uncertainty characteristics of the flood and carrying out normalization processing; step 3, randomly sampling and screening 2/3 samples as a training set of the deep learning model, and setting an optimization objective function; step 4, the remaining 1/3 samples verify the simulation effect of the deep learning model. The invention can utilize the advantage of a deep learning theory method for processing large sample data, and reasonably describes the physical relation between 'flood uncertainty' and 'reservoir scheduling' in a data-driven mode, thereby better guiding the formulation of reservoir scheduling decision under the condition of considering runoff uncertainty, being widely applied to reservoir flood period forecast scheduling and providing basis and technical support for scientifically formulating scheduling decision.

Description

Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism
Technical Field
The invention relates to the technical field of reservoir scheduling, in particular to a flood-reservoir mapping relation simulation method for deep learning guided by a physical mechanism.
Background
The runoff process has both deterministic and stochastic components, and reservoir water level characteristic parameter design, scheduling rule formulation and scheduling decision in the actual scheduling operation process are all based on the analysis of the characteristics of the warehousing runoff, so that the establishment of the response relationship between the warehousing runoff and the reservoir scheduling is the basis of water resource management and configuration.
At present, the existing reservoir optimization scheduling research considering runoff uncertainty mainly takes a mode of constructing a scheduling model based on a water balance principle. On one hand, the research idea is that a large number of runoff scenes are generated through random simulation to depict uncertainty characteristics of the runoff scenes, and then the runoff scenes are taken as input conditions of a scheduling model to carry out reservoir optimal scheduling research, namely that the 'runoff uncertainty depiction' and the 'reservoir optimal scheduling' are carried out in steps; on the other hand, if the dimension of the optimized variable is high, the solution of the reservoir scheduling model has certain difficulty. In recent years, with the rapid development of deep learning theoretical methods, a lot of students have applied the deep learning theoretical methods to the water resource field. However, the deep learning method based on big data does not have description on the physical mechanism among variables, so the theoretical method cannot completely analyze and establish the causal response relationship between 'warehousing runoff' and 'reservoir dispatching'.
The flood control dispatching of the reservoir in the flood season needs to pay attention to the relationship between the flood process forecast and the flood control reservoir capacity reserved by the reservoir in the soakage, but the existing method for constructing a dispatching model based on the water balance principle has the following problems: (1) the mapping aiming at the response relation between 'flood uncertainty' and 'reservoir flood control storage capacity' is realized by building a scheduling model, a large amount of numerical calculation exists in the runoff simulation and model solving process, namely, one reservoir flood regulation research calculation needs to be correspondingly carried out when any given warehousing runoff input scene is given; (2) when the response relation between the process of storing flood and the flood control storage capacity of the reservoir is calculated by adopting a traditional reservoir flood diversion calculation method, the larger the magnitude of the stored flood is, the larger the space of the flood control storage capacity which needs to be reserved is, but if the simulation between 'flood uncertainty' and 'reservoir flood control storage capacity' is carried out by only utilizing deep learning, objective monotonicity cannot be ensured.
Disclosure of Invention
The invention is carried out to solve the problems, and aims to provide a flood-reservoir mapping relation simulation method for deep learning guided by a physical mechanism, which can reasonably describe the physical relation between 'flood uncertainty' and 'reservoir scheduling' in a data-driven manner by utilizing the advantage of a deep learning theoretical method in processing large sample data, thereby better guiding the formulation of reservoir scheduling decisions under the condition of considering runoff uncertainty.
The invention provides a flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism, which is characterized by comprising the following steps:
step 1, selecting a deep learning model for depicting a mapping relation according to inherent characteristics of the mapping relation of flood uncertainty-reservoir flood control storage capacity;
step 2, according to the characteristic analysis of the flood in storage, screening parameters describing the uncertainty characteristics of the flood, inputting the parameters as a model after normalization and arrangement, and simulating a flood control storage capacity value V to be reserved by a reservoir according to a deep learning model;
step 3, randomly sampling 2/3 samples in the sample capacity N to serve as a training set of a deep learning model, selecting target function training model parameters, inputting parameter values for describing flood uncertainty characteristics screened in the step 2, outputting simulated reservoir capacity values V which are reserved for a flood control reservoir, and adding penalty terms in the model target function for considering a physical mechanism in a reservoir flood regulating calculation process;
and 4, taking the remaining 1/3 samples as a verification set of the deep learning model, and verifying the fitting effect of the model to obtain the reserved flood control reservoir capacity value V and the feasible interval of the reservoir corresponding to different flood magnitude and different flood forecasting errors.
Further, the parameters describing the uncertainty characteristics of the flood in step 2 include two parts: one is the design frequency P and the flood peak flow value Q related to the randomness of the flood processpMaximum 3,7,15 days flood value WiD(i-3, 7,15), flood sample Y for multi-field different types of typical yearskWhere k is 1,2, …, m is the typical annual sample number; secondly, the uncertainty of flood forecast is described, the flood forecast error sigma is used for representing the level of flood forecast precision, and the simulated reservoir should reserve the flood control reservoir capacity value V and the feasible interval [ V [ [ V ]D,VU]The following formula:
[V,VD,VU]＝f(X,σ) (1)
in the formula, X is input data after normalization processing, and is obtained by calculation according to the parameters describing the uncertainty characteristics of the flood process, including design frequency P and flood peak flow value QpMaximum 3,7,15 days flood value WiD(i-3, 7,15), flood sample Y for multi-field different types of typical yearsk(ii) a Sigma is flood forecast error; vUAnd VDRespectively representing the upper and lower threshold values of the capacity value of the flood control reservoir to be reserved; f (-) is a simulation function of the deep learning method, and if only the influence of the uncertainty of the flood process on the reserved flood-control capacity of the reservoir is considered, the simulation function can be simplified into V ═ f (x).
Further, the objective function calculation formula in step 3 is as follows:
min G＝α1·MSE1+α2·MSE2+α3·MSE3 (2)
Figure BDA0003016725590000031
in the formula, G1And G2Is the objective function value; MSE1Reserving a mean square error of the capacity value of the flood control reservoir for the reservoir simulated by the deep learning model; alpha is alpha1Is MSE1The weight coefficient of (a); n is the number of calculation samples; MSE2A mean square error penalty term for dealing with the monotonicity constraint of the flood magnitude for the deep learning model; alpha is alpha2Is MSE2The weight coefficient of (a); MSE3A mean square error punishment item for the deep learning model corresponding to the monotonicity constraint of the flood forecasting error; alpha is alpha3Is MSE3The weight coefficient of (2).
Further, the deep learning model in step 3 adopts the following formula to calculate the monotonicity constraint of flood magnitude:
Figure BDA0003016725590000032
Figure BDA0003016725590000033
in the formula:
Figure BDA0003016725590000034
used for screening out error items which do not meet the monotonicity of flood magnitude corresponding to the capacity value of the reserved flood control reservoir of the reservoir under the condition of not considering the forecast error,
Figure BDA0003016725590000035
is used for screening out error items which do not meet the monotonicity of flood magnitude corresponding to the lower limit value of the reserved flood control reservoir of the reservoir,
Figure BDA0003016725590000036
the method is used for screening out error items which do not meet the monotonicity of the corresponding flood magnitude of the flood protection upper limit value reserved by the reservoir;
the monotonicity constraint of the deep learning model corresponding to the flood forecasting error is calculated by adopting the following formula:
Figure BDA0003016725590000041
Figure BDA0003016725590000042
in the formula:
Figure BDA0003016725590000043
is used for screening out error items which do not satisfy the monotonicity of flood forecast errors corresponding to the logical relationship between the upper limit and the lower limit of the reserved flood control capacity of the reservoir,
Figure BDA0003016725590000044
and the method is used for screening out error items which do not meet the requirement of monotonicity of flood magnitude corresponding to the width of the feasible interval of the reserved flood control capacity of the reservoir.
Compared with the prior art, the invention has the beneficial effects that:
(1) the invention provides a flood-reservoir mapping relation simulation method for deep learning guided by a physical mechanism, which can reasonably describe the physical relation between flood uncertainty and reservoir scheduling in a data-driven manner by utilizing the advantages of a deep learning theoretical method in processing large sample data;
(2) the flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism disclosed by the invention punishs the calculation results of flood magnitude and flood forecast error monotonicity which do not meet the requirement of reserved flood-control storage capacity of a reservoir, and solves the problem that the traditional deep learning method is lack of the physical mechanism involved in the reservoir scheduling process.
Drawings
Fig. 1 is a schematic flow chart of a flood-reservoir mapping relationship simulation method for deep learning guided by a physical mechanism according to an embodiment of the present invention;
fig. 2 is a comparison graph of a conventional point set of flood-reservoir mapping relation simulation of deep learning guided by a physical mechanism proposed in the embodiment of the present invention and an abnormal point set generated only based on the deep learning simulation.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
As shown in fig. 1, a flood-reservoir mapping relationship simulation method for deep learning guided by physical mechanisms according to a first embodiment includes the following steps:
step 1, selecting a deep learning model for constructing a mapping relation of 'flood uncertainty-reservoir flood control storage capacity', such as a long-term and short-term neural network model;
step 2, according to the characteristic analysis of the flood in storage, screening parameters describing the uncertainty characteristics of the flood, inputting the parameters as a model after normalization and arrangement, and simulating a flood control storage capacity value V to be reserved by a reservoir according to a deep learning model;
and 3, randomly sampling and screening 2/3 samples in the sample capacity N to serve as a training set of the deep learning model, selecting target function training model parameters, taking the parameter values for describing the uncertainty characteristics of flood water screened in the step 2 as input, taking the simulated reservoir capacity value V which is required to reserve a flood control reservoir as output, and adding a penalty term into the model target function for considering a physical mechanism in the reservoir flood control calculation process. As shown in fig. 2, if only the deep learning model is used to simulate the mapping relationship between the "flood uncertainty-reservoir flood control capacity", it is possible to output an abnormal point set that does not satisfy the monotonicity characteristic, and the simulation model using the physical mechanism to guide the deep learning can ensure that the output results are all conventional point sets.
Step 4, taking the remaining 1/3 samples as a verification set of the deep learning model, and using the verification set to verify the fitting effect of the model, and obtaining the reserved flood control reservoir capacity value V and the feasible interval of the reservoir corresponding to different flood magnitude and different flood forecasting errors;
according to the scheme, the parameters for describing the uncertainty characteristics of the flood in the step 2 comprise two parts: one is about the randomness of the flood process, and the screening parameters are the design frequency P and the peak flow value QpMaximum 3,7,15 days flood value WiD(i-3, 7,15), flood sample Y for multi-field different types of typical yearsk(m is the typical number of annual samples, k is 1,2, …, m); the second is uncertainty of flood forecast, which adopts flood forecast error sigma to characterize the level of flood forecast precision, and the simulated reservoir should reserve flood control reservoir capacity value V and feasible interval [ V [ ]D,VU]The following formula:
[V,VD,VU]＝f(X,σ) (1)
in the formula, X is input data after normalization processing, and is obtained by calculation according to the parameter describing the uncertainty characteristic of the flood process; sigma is flood forecast error; vUAnd VDRespectively representing the upper and lower threshold values of the capacity value of the flood control reservoir to be reserved; f (·) is a simulation function of the deep learning method, and if only the influence of uncertainty of a flood process on the reserved flood control capacity of the reservoir is considered, the simulation function can be simplified into V ═ f (x);
according to the scheme, the objective function calculation formula in the step 3 is as follows:
min G＝α1·MSE1+α2·MSE2+α3·MSE3 (2)
Figure BDA0003016725590000061
in the formula, G1And G2Is a value of an objective function；MSE1Reserving a mean square error of the capacity value of the flood control reservoir for the reservoir simulated by the deep learning model; alpha is alpha1Is MSE1The weight coefficient of (a); n is the number of calculation samples; MSE2A mean square error penalty term for dealing with the monotonicity constraint of the flood magnitude for the deep learning model; alpha is alpha2Is MSE2The weight coefficient of (a); MSE3A mean square error punishment item for the deep learning model corresponding to the monotonicity constraint of the flood forecasting error; alpha is alpha3Is MSE3The weight coefficient of (a);
according to the scheme, the penalty item in the step 3 comprises the following contents:
1) the deep learning model is used for solving the monotonicity constraint of flood magnitude by adopting the following formula:
Figure BDA0003016725590000062
Figure BDA0003016725590000063
in the formula:
Figure BDA0003016725590000064
used for screening out error items which do not meet the monotonicity of flood magnitude corresponding to the capacity value of the reserved flood control reservoir of the reservoir under the condition of not considering the forecast error,
Figure BDA0003016725590000065
is used for screening out error items which do not meet the monotonicity of flood magnitude corresponding to the lower limit value of the reserved flood control reservoir of the reservoir,
Figure BDA0003016725590000066
the method is used for screening out error items which do not meet the monotonicity of the corresponding flood magnitude of the flood protection upper limit value reserved by the reservoir;
2) the monotonicity constraint of the deep learning model corresponding to the flood forecasting error is calculated by adopting the following formula:
Figure BDA0003016725590000071
Figure BDA0003016725590000072
in the formula:
Figure BDA0003016725590000073
is used for screening out error items which do not satisfy the monotonicity of flood forecast errors corresponding to the logical relationship between the upper limit and the lower limit of the reserved flood control capacity of the reservoir,
Figure BDA0003016725590000074
and the method is used for screening out error items which do not meet the requirement of monotonicity of flood magnitude corresponding to the width of the feasible interval of the reserved flood control capacity of the reservoir.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (4)

1. A flood-reservoir mapping relation simulation method for deep learning guided by a physical mechanism is characterized by comprising the following steps:
step 1, selecting a deep learning model for depicting a mapping relation according to inherent characteristics of the mapping relation of flood uncertainty-reservoir flood control storage capacity;
step 2, according to the characteristic analysis of the flood in storage, screening parameters describing the uncertainty characteristics of the flood, inputting the parameters as a model after normalization and arrangement, and simulating a flood control storage capacity value V to be reserved by a reservoir according to a deep learning model;
step 3, randomly sampling 2/3 samples in the sample capacity N to serve as a training set of a deep learning model, selecting target function training model parameters, inputting parameter values for describing flood uncertainty characteristics screened in the step 2, outputting simulated reservoir capacity values V which are reserved for a flood control reservoir, and adding penalty terms in the model target function for considering a physical mechanism in a reservoir flood regulating calculation process;
and 4, taking the remaining 1/3 samples as a verification set of the deep learning model, and verifying the fitting effect of the model to obtain the reserved flood control reservoir capacity value V and the feasible interval of the reservoir corresponding to different flood magnitude and different flood forecasting errors.
2. The flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism according to claim 1, characterized in that:
the parameters describing the uncertainty characteristics of the flood in the step 2 comprise two parts: one is the design frequency P and the flood peak flow value Q related to the randomness of the flood processpMaximum 3,7,15 days flood value WiD(i-3, 7,15), flood sample Y for multi-field different types of typical yearskWhere k is 1,2, …, m is the typical annual sample number; secondly, the uncertainty of flood forecast is described, the flood forecast error sigma is used for representing the level of flood forecast precision, and the simulated reservoir should reserve the flood control reservoir capacity value V and the feasible interval [ V [ [ V ]D,VU]The following formula:
[V,VD,VU]＝f(X,σ) (1)
in the formula, X is input data after normalization processing, and is obtained by calculation according to the parameters describing the uncertainty characteristics of the flood process, including design frequency P and flood peak flow value QpMaximum 3,7,15 days flood value WiD(i-3, 7,15), flood sample Y for multi-field different types of typical yearsk(ii) a Sigma is flood forecast error; vUAnd VDRespectively representing the upper and lower threshold values of the capacity value of the flood control reservoir to be reserved; f (-) is a simulation function of the deep learning method, and if only the influence of the uncertainty of the flood process on the reserved flood-control capacity of the reservoir is considered, the simulation function can be simplified into V ═ f (x).
3. The flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism according to claim 1, characterized in that:
the objective function calculation formula in step 3 is as follows:
min G＝α1·MSE1+α2·MSE2+α3·MSE3 (2)
Figure FDA0003016725580000021
in the formula, G1And G2Is the objective function value; MSE1Reserving a mean square error of the capacity value of the flood control reservoir for the reservoir simulated by the deep learning model; alpha is alpha1Is MSE1The weight coefficient of (a); n is the number of calculation samples; MSE2A mean square error penalty term for dealing with the monotonicity constraint of the flood magnitude for the deep learning model; alpha is alpha2Is MSE2The weight coefficient of (a); MSE3A mean square error punishment item for the deep learning model corresponding to the monotonicity constraint of the flood forecasting error; alpha is alpha3Is MSE3The weight coefficient of (2).
4. The flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism according to claim 3, characterized in that:
in the step 3, the deep learning model is used for solving the monotonicity constraint of the flood magnitude by adopting the following formula:
Figure FDA0003016725580000022
Figure FDA0003016725580000023
in the formula:
Figure FDA0003016725580000024
used for screening out error items which do not meet the monotonicity of flood magnitude corresponding to the capacity value of the reserved flood control reservoir of the reservoir under the condition of not considering the forecast error,
Figure FDA0003016725580000031
is used for screening out error items which do not meet the monotonicity of flood magnitude corresponding to the lower limit value of the reserved flood control reservoir of the reservoir,
Figure FDA0003016725580000032
the method is used for screening out error items which do not meet the monotonicity of the corresponding flood magnitude of the flood protection upper limit value reserved by the reservoir;
the monotonicity constraint of the deep learning model corresponding to the flood forecasting error is calculated by adopting the following formula:
Figure FDA0003016725580000033
Figure FDA0003016725580000034
in the formula:
Figure FDA0003016725580000035
is used for screening out error items which do not satisfy the monotonicity of flood forecast errors corresponding to the logical relationship between the upper limit and the lower limit of the reserved flood control capacity of the reservoir,
Figure FDA0003016725580000036
and the method is used for screening out error items which do not meet the requirement of monotonicity of flood magnitude corresponding to the width of the feasible interval of the reserved flood control capacity of the reservoir.
CN202110390930.2A 2021-04-12 2021-04-12 Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism Active CN113051833B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110390930.2A CN113051833B (en) 2021-04-12 2021-04-12 Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110390930.2A CN113051833B (en) 2021-04-12 2021-04-12 Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism

Publications (2)

Publication Number Publication Date
CN113051833A true CN113051833A (en) 2021-06-29
CN113051833B CN113051833B (en) 2022-04-05

Family

ID=76519109

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110390930.2A Active CN113051833B (en) 2021-04-12 2021-04-12 Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism

Country Status (1)

Country Link
CN (1) CN113051833B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110675579A (en) * 2019-09-24 2020-01-10 安徽沃特水务科技有限公司 Reservoir early warning monitoring method and system
CN111428421A (en) * 2020-03-10 2020-07-17 武汉大学 Rainfall runoff simulation method for deep learning guided by physical mechanism
CN111859787A (en) * 2020-07-02 2020-10-30 北京工业大学 Daily runoff forecasting method based on coupling of meteorological hydrological forecasting factor classification and deep learning model
US20210007695A1 (en) * 2019-07-12 2021-01-14 Canon Medical Systems Corporation Apparatus and method using physical model based deep learning (dl) to improve image quality in images that are reconstructed using computed tomography (ct)
CN112330005A (en) * 2020-10-27 2021-02-05 武汉祁联生态科技有限公司 Water quality prediction method based on sequence-to-sequence deep learning mechanism
CN112396152A (en) * 2020-11-17 2021-02-23 郑州大学 Flood forecasting method based on CS-LSTM
CN112418491A (en) * 2020-11-06 2021-02-26 黄河勘测规划设计研究院有限公司 Dynamic configuration method for residual sediment storage capacity of reservoir

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20210007695A1 (en) * 2019-07-12 2021-01-14 Canon Medical Systems Corporation Apparatus and method using physical model based deep learning (dl) to improve image quality in images that are reconstructed using computed tomography (ct)
CN110675579A (en) * 2019-09-24 2020-01-10 安徽沃特水务科技有限公司 Reservoir early warning monitoring method and system
CN111428421A (en) * 2020-03-10 2020-07-17 武汉大学 Rainfall runoff simulation method for deep learning guided by physical mechanism
CN111859787A (en) * 2020-07-02 2020-10-30 北京工业大学 Daily runoff forecasting method based on coupling of meteorological hydrological forecasting factor classification and deep learning model
CN112330005A (en) * 2020-10-27 2021-02-05 武汉祁联生态科技有限公司 Water quality prediction method based on sequence-to-sequence deep learning mechanism
CN112418491A (en) * 2020-11-06 2021-02-26 黄河勘测规划设计研究院有限公司 Dynamic configuration method for residual sediment storage capacity of reservoir
CN112396152A (en) * 2020-11-17 2021-02-23 郑州大学 Flood forecasting method based on CS-LSTM

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
曹枝俏等: "基于随机模拟信息的神经网络洪水预报模型", 《水力发电学报》 *

Also Published As

Publication number Publication date
CN113051833B (en) 2022-04-05

Similar Documents

Publication Publication Date Title
CN109657881A (en) A kind of neural network photovoltaic power generation prediction technique and system suitable for small sample
CN104899431B (en) Based on ant colony and swarm of particles into LSSVM fluctuating wind speed Forecasting Methodologies
CN110910004A (en) Reservoir dispatching rule extraction method and system with multiple uncertainties
CN105809349B (en) Dispatching method for step hydropower station group considering incoming water correlation
CN113554466A (en) Short-term power consumption prediction model construction method, prediction method and device
CN107392446B (en) A kind of step power station scheduling scheme evaluation method based on sensitivity analysis
Sørensen et al. Recent developments in multivariate wind and solar power forecasting
CN112990500A (en) Transformer area line loss analysis method and system based on improved weighted gray correlation analysis
CN112329997A (en) Power demand load prediction method and system, electronic device, and storage medium
CN110428270A (en) The potential preference client recognition methods of the channel of logic-based regression algorithm
CN109472403A (en) A kind of set empirical mode decomposition and distant relevant Medium-and Long-Term Runoff Forecasting method
Müller et al. An econometric model of international long-run growth dynamics
CN112653198A (en) Wind power output scene generation method and system based on prediction box
CN111008790A (en) Hydropower station group power generation electric scheduling rule extraction method
Dong et al. Uncertainty-based multi-objective decision making with hierarchical reliability analysis under water resources and environmental constraints
CN116050595A (en) Attention mechanism and decomposition mechanism coupled runoff amount prediction method
Liu et al. Dynamic economic dispatch with wind power penetration based on non-parametric kernel density estimation
CN110648248B (en) Control method, device and equipment for power station
Zhijun RBF neural networks optimization algorithm and application on tax forecasting
CN113051833B (en) Flood-reservoir mapping relation simulation method for deep learning guided by physical mechanism
CN113762591A (en) Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy
Guo et al. Predicting the Population Growth and Structure of China Based on Grey Fractional‐Order Models
CN108110756A (en) Consider the industrial park distribution network planning method of uncertain factor
CN117291069A (en) LSTM sewage water quality prediction method based on improved DE and attention mechanism
CN116646917A (en) Campus multi-step probability power load prediction method and system

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
GR01 Patent grant
GR01 Patent grant