CN117575873A - Flood warning method and system for comprehensive meteorological hydrologic sensitivity - Google Patents
Flood warning method and system for comprehensive meteorological hydrologic sensitivity Download PDFInfo
- Publication number
- CN117575873A CN117575873A CN202410050804.6A CN202410050804A CN117575873A CN 117575873 A CN117575873 A CN 117575873A CN 202410050804 A CN202410050804 A CN 202410050804A CN 117575873 A CN117575873 A CN 117575873A
- Authority
- CN
- China
- Prior art keywords
- water level
- flood
- data
- hydrologic
- meteorological
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 87
- 230000035945 sensitivity Effects 0.000 title claims abstract description 37
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 276
- 238000012544 monitoring process Methods 0.000 claims abstract description 98
- 238000004458 analytical method Methods 0.000 claims abstract description 52
- 238000010206 sensitivity analysis Methods 0.000 claims abstract description 31
- 238000010801 machine learning Methods 0.000 claims abstract description 27
- 230000002265 prevention Effects 0.000 claims abstract description 20
- 238000007781 pre-processing Methods 0.000 claims abstract description 19
- 230000008859 change Effects 0.000 claims abstract description 11
- 238000011156 evaluation Methods 0.000 claims abstract description 7
- 238000011160 research Methods 0.000 claims description 28
- 230000004044 response Effects 0.000 claims description 25
- 239000013598 vector Substances 0.000 claims description 21
- 238000012360 testing method Methods 0.000 claims description 18
- 238000000354 decomposition reaction Methods 0.000 claims description 16
- 238000013135 deep learning Methods 0.000 claims description 16
- 238000012417 linear regression Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000010586 diagram Methods 0.000 claims description 13
- 238000001556 precipitation Methods 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 9
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000012795 verification Methods 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 238000004891 communication Methods 0.000 claims description 6
- 238000007726 management method Methods 0.000 claims description 6
- 230000000873 masking effect Effects 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000003247 decreasing effect Effects 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000010972 statistical evaluation Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 2
- 238000009472 formulation Methods 0.000 claims 1
- 239000000203 mixture Substances 0.000 claims 1
- 238000010200 validation analysis Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 4
- 238000009825 accumulation Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000006378 damage Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 238000006424 Flood reaction Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000009514 concussion Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000005180 public health Effects 0.000 description 1
- 230000003014 reinforcing effect Effects 0.000 description 1
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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01F—MEASURING VOLUME, VOLUME FLOW, MASS FLOW OR LIQUID LEVEL; METERING BY VOLUME
- G01F1/00—Measuring the volume flow or mass flow of fluid or fluent solid material wherein the fluid passes through a meter in a continuous flow
- G01F1/002—Measuring the volume flow or mass flow of fluid or fluent solid material wherein the fluid passes through a meter in a continuous flow wherein the flow is in an open channel
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01F—MEASURING VOLUME, VOLUME FLOW, MASS FLOW OR LIQUID LEVEL; METERING BY VOLUME
- G01F23/00—Indicating or measuring liquid level or level of fluent solid material, e.g. indicating in terms of volume or indicating by means of an alarm
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/02—Instruments for indicating weather conditions by measuring two or more variables, e.g. humidity, pressure, temperature, cloud cover or wind speed
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- 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
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- 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
- G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/10—Alarms for ensuring the safety of persons responsive to calamitous events, e.g. tornados or earthquakes
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B31/00—Predictive alarm systems characterised by extrapolation or other computation using updated historic data
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Educational Administration (AREA)
- Environmental & Geological Engineering (AREA)
- Development Economics (AREA)
- Fluid Mechanics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Marketing (AREA)
- Emergency Management (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Primary Health Care (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geology (AREA)
Abstract
The invention discloses a flood warning method and a flood warning system for comprehensive meteorological hydrologic sensitivity, which relate to the technical field of hydraulic engineering and comprise a monitoring and data acquisition unit, a data driving water level prediction unit, a sensitivity analysis unit and a judgment unit; the data-driven water level prediction unit comprises a preprocessing module, an evaluation module and a water level prediction module. The invention adopts a transducer model to combine sparse attention and nonlinear output, and accurately predicts the water level change under multiple meteorological and hydrologic factors. Using the interpretable machine learning analysis of the sensitivity of water level to weather hydrology, water levels exceeding the warning water level are marked as risk points. And after flood warning, carrying out flood cause analysis on the high-sensitivity meteorological hydrologic factors, and establishing a historical flood prevention scheme database. The water level, rainfall and the like are used as characteristics, a preliminary flood prevention scheme is output, and a final scheme is determined by combining expert opinion, so that the water level is controlled in a reasonable range.
Description
Technical Field
The invention relates to the technical field of hydraulic engineering, in particular to a flood warning method and system for comprehensive meteorological hydrologic sensitivity.
Background
The hydrologic sensitive flood warning function is to provide timely and effective flood warning service to reduce flood risk and ensure life safety and property safety of people. The warning system is mainly based on hydrologic observation and analysis, and can more accurately estimate the possibility and the danger degree of flood occurrence through monitoring rainfall, river water level, soil water content and other data.
Firstly, the hydrologic sensitivity flood warning is helpful for early warning, and flood damage is reduced. By monitoring meteorological data and real-time hydrologic information, the system is able to quickly identify potential flood risk areas. Once the rainfall or river water level rise which possibly causes flood is detected, the system can give an alarm in time to remind residents and related institutions of taking emergency measures, such as evacuating personnel, reinforcing embankment inspection, closing traffic and the like, so that the loss caused by the flood is reduced to the greatest extent. Second, hydrologic sensitive flood alert is of great importance for water resource management and scheduling. Through the real-time monitoring of river flow, water level, soil water content and other data, the hydrologic sensitive flood warning system can help the water resource management mechanism to more accurately perform water resource distribution and water flow scheduling. Before flood comes, reservoir water level can be adjusted in advance, flood discharge operation can be carried out, stable water flow is ensured, and influence of flood on downstream areas is reduced. In addition, hydrologically sensitive flood warning also provides an important basis for city planning and flood control facility design. By analyzing historical flood events and real-time hydrologic data, flood sensitivity and potential risk of different areas can be more comprehensively assessed. This helps city planners to better choose construction land, design drainage systems, build flood protection facilities, and improve the flood protection capacity of the city. Finally, hydrologic sensitivity flood alert aids in scientific research and data accumulation. Through long-term hydrologic monitoring and data analysis, a large amount of hydrologic information can be accumulated, and abundant experimental data are provided for research of multiple subjects such as weather, geology, hydrology and the like. These data help to gain a better understanding of the flood formation mechanism, the improvement of predictive models, and the impact of floods on the ecosystem, thereby providing a scientific basis for future flood control and climate change adaptation.
In general, the hydrologic sensitivity flood warning provides timely flood early warning service by integrating multisource hydrologic data, and plays an important role in reducing flood risks and guaranteeing life safety and property safety of people. The system plays a key role in emergency situations and provides beneficial support for long-term water resource management, urban planning and scientific research.
Flood warning methods often involve a variety of water level prediction models based on deep learning methods, whereby the physical processes present in the model and the natural phenomena of their reactions are revealed by using accurate water level prediction models. Along with the continuous progress of artificial intelligence and hydrologic prediction technology and the practical accumulation of hydraulic engineering construction, massive meteorological hydrologic data should be added to a water level prediction model, and the model performance is greatly improved. Meanwhile, with the development of an interpretable machine learning method, various interpretable artificial intelligence methods are gradually introduced into the water level prediction model, so that the transparency and the physical interpretability of the water level prediction model are improved. After entering summer, continuous storm water causes frequent flood disasters, the water levels of main rivers and sluice gates are greatly increased, huge life and property losses and public health infrastructure damages are caused, but the existing water level prediction model lacks a high-efficiency and interpretable flood warning method, and has great application risks. Therefore, the invention provides a flood warning method and a flood warning system integrating meteorological hydrologic sensitivity.
The above information disclosed in the background section is only for enhancement of understanding of the background of the disclosure and therefore it may include information that does not form the prior art that is already known to a person of ordinary skill in the art.
Disclosure of Invention
The invention aims to provide a flood warning method and a system for comprehensive meteorological hydrologic sensitivity, which accurately predicts water level under the condition that various meteorological and hydrologic elements exist, quantitatively estimates water level change of a research area and improves the performance of a transducer by using sparse attention and an additional nonlinear output layer; after a trained water level prediction model is obtained, the sensitivity of the water level to each meteorological and hydrologic element is analyzed by utilizing an interpretable machine learning method, if the water level prediction value exceeds the warning water level or guarantees the water level, the position of the water level monitoring point is marked as a risk point for warning and abstaining or a flood approaching point, and a flood warning judgment result is obtained; after obtaining the flood warning judgment result, carrying out flood cause analysis of the water level and the high-sensitivity meteorological hydrologic factors according to the judgment result, and judging the flood cause according to the sensitivity if a flood danger is generated; after the analysis result of the flood causes is obtained, relevant departments are informed in time to collect historical flood prevention schemes, a complete flood response scheme database is formulated, water levels, rainfall, weather and the like are used as flood characteristics, corresponding schemes are matched from the response scheme database, reasonable preliminary flood prevention schemes are output to flood related functional departments, expert opinions are collected to determine final flood response schemes later, and water levels are controlled in a reasonable range, so that the problems in the background technology are solved.
In order to achieve the above object, the present invention provides the following technical solutions: the flood warning system for comprehensive meteorological hydrologic sensitivity comprises a monitoring and data acquisition unit, a data driving water level prediction unit, a sensitivity analysis unit and a judgment unit;
the data-driven water level prediction unit comprises a preprocessing module, an evaluation module and a water level prediction module;
setting a plurality of hydrologic monitoring points at main rivers and floodgates of a research area, collecting flood related flow and water level at the monitoring points by using an open channel flowmeter and a water level ruler by using a monitoring and data acquisition unit, simultaneously acquiring average air temperature, precipitation, highest air temperature, lowest air temperature, average wind speed, relative humidity and sunshine time of a meteorological site, and storing historical hydrologic and meteorological data into an initial data set;
the method comprises the steps that a preprocessing module in a data-driven water level prediction unit is used for preprocessing hydrologic and meteorological data of an initial data set, outlier rejection and missing value filling are adopted for hydrologic data, boundary vectors of a research area are firstly determined for the meteorological data, the meteorological data of the research area are obtained by cutting and splicing, then the meteorological data consistent with the time resolution of the hydrologic data are obtained by a statistical downscaling method based on multiple linear regression and singular value decomposition, and the meteorological data are stored into a meteorological hydrologic data set;
The water level prediction module in the data-driven water level prediction unit is used for constructing a water level prediction model based on a transform deep learning method for the stored initial data set, and an accurate water level prediction model is obtained by adjusting model super-parameters;
evaluating the performance of the water level prediction model by an evaluation module;
the sensitivity analysis unit displays the water level prediction result, a water level time evolution relation diagram is obtained by using a trend analysis and significance analysis method, and then a weather and hydrologic sensitivity analysis diagram is obtained by using an interpretable machine learning method;
and comparing the predicted water level with the guard and guaranteed water level of the area by a judging unit, marking the position of the water level monitoring point as a risk point for vigilance and abstainment if the predicted water level value exceeds the guard water level of the area, and marking the position of the water level monitoring point as a neighboring flood point if the predicted water level value exceeds the guaranteed water level of the area to obtain a flood guard judging result.
Preferably, the statistical downscaling method based on multiple linear regression and singular value decomposition comprises the following specific steps:
the weather data and the forecasting factors are expressed by linear relation, namely:wherein, the parameter meaning is: / >For the predicted quantity, A is regression coefficient matrix, < ->Is a predictor;
mapping the original high-dimensional meteorological data into a low-dimensional subspace to realize data dimension reduction to obtain high-resolution meteorological data, wherein the definition of singular value decomposition is as follows:wherein, the parameter meaning is: matrix->Middle column vector->Is a left singular vector, matrix->Is a singular value matrix, a diagonal matrix, +.>Is->Singular value, matrix->Middle column vector->Is right singular vector, matrix->The unfolding is as follows: />Then intercept front +.>Singular values, in->Searching the original matrix in the dimensional space>Low-dimensional matrix with shortest distance +.>Then: />。
Preferably, the monitoring and data acquisition unit comprises a precipitation monitoring module, a water level monitoring module, a temperature monitoring module, an average wind speed monitoring module, a sunshine time monitoring module and a relative humidity monitoring module, and the meteorological hydrology data set is constructed as follows:
the rainfall data is monitored and collected in the monitoring point by the rainfall monitoring module, the water level data is monitored and collected in the monitoring point by the water level monitoring module, the temperature data is monitored and collected in the monitoring point by the temperature monitoring module, the average air temperature, the highest air temperature and the lowest air temperature data are obtained through mathematical calculation respectively, the wind speed data is monitored and collected in the monitoring point by the average wind speed monitoring module, the sunshine time data is monitored and collected in the monitoring point by the sunshine time monitoring module, and the relative humidity data is monitored and collected in the monitoring area by the relative humidity monitoring module;
The preprocessing module screens out abnormal values and missing values existing in the hydrological data, eliminates the abnormal values and complements the missing values by adopting an interpolation method, wherein the interpolation method is one of an inverse distance weight method, a Kriging interpolation method, a natural neighbor method, a spline function interpolation method, a trend method and a depth interpolation method;
and then the preprocessing module cuts and splices the meteorological data according to the boundary vector data of the research area, obtains the meteorological data with the same time resolution as the hydrologic data by using a statistical downscaling method based on multiple linear regression and singular value decomposition, splices the preprocessed meteorological hydrologic data according to a time sequence, and constructs a meteorological hydrologic data set.
Preferably, the mode of constructing the water level prediction model based on the transform deep learning method is as follows:
the water level prediction module divides a meteorological hydrologic data set into a training set, a verification set and a test set according to the ratio of 7:2:1, a water level prediction model based on a transducer deep learning method is obtained on the training set, super-parameters of the water level prediction model are adjusted on the verification set, performance indexes of the model are obtained on the test set, and the performance indexes comprise four different statistical evaluation indexes, a standard root mean square error, an average absolute error, an average deviation error and a decision coefficient;
The water level prediction model based on the transform deep learning method consists of a two-layer encoder and a one-layer decoder, wherein:
each layer of encoder consists of two sub-layers, wherein each sub-layer comprises a multi-head sparse attention and position full-connection feedforward network module, and the sub-layers are combined through residual connection and layer normalization;
the decoder comprises three sub-layers, namely a multi-head mask sparse attention, a multi-head sparse attention and a position full-connection feedforward network module, and is combined through residual connection and layer normalization;
the multi-head sparse attention module adopts masking operation to determine the sparsity of vectors, the probability is specially allocated to the most important elements in the input sequence, then the attention score is calculated for the selected positions, and the output of each head is calculated concurrently by utilizing a multi-head mechanism;
the multi-head mask sparse attention module is used for masking information of all positions after the current moment;
the position full-connection feedforward network module is a two-layer full-connection layer, and the corresponding expression is:;
wherein the residual connection is used for solving the problem of multi-layer network training;
wherein layer normalization is used to accelerate network convergence.
Preferably, the trend analysis, significance analysis and interpretable machine learning method are as follows:
The trend change of the water level is researched by using a unitary linear regression through a trend analysis method, and the specific equation is as follows:wherein n is the number of days, < >>Is the firstiDay water level value, ">For multiple days->Value linear fit slope, when +.>When indicating +.>Gradually increasing day by day, and gradually decreasing otherwise;
the significance analysis method comprises calculating a test statistic related to the length of water level sequence by Mann-Kendall (M-K) method, and performing significance test to determine whether there is trend in water level sequence, for a water level sequence with length of nFirst calculate +.>WhereinThen statistics->Calculate +.about.in its normal distribution>The value of the sum of the values,,/>wherein, the parameter meaning is: z is a significance value, S is a statistic, < ->Is a water level sequence variable, if->The water level sequence is considered to have a significant change trend through the significance test, wherein the significance test shows that the confidence coefficient is 90%, 95% and 99% respectively;
the sensitivity analysis unit comprises a hydrologic sensitivity analysis module, a meteorological sensitivity analysis module and a water level result display module, wherein the water level result display module displays a water level prediction result, a water level time evolution relation diagram is obtained by using a trend analysis and significance analysis method, and the hydrologic sensitivity analysis module and the meteorological sensitivity analysis module then obtain a meteorological and hydrologic sensitivity analysis diagram by using an interpretable machine learning method;
The contribution of each variable to the overall model is determined by an interpretable machine learning method, the equation being as follows:
wherein, the parameter meaning is:Erepresenting an interpretation model of the model,irepresenting the number of input variables to be processed,tsimplification of representative feature vectors, ++>Representing the contribution of each variable to the machine learning model, < ->The expression of (2) is as follows:wherein, the method comprises the steps of, wherein,Ma machine learning model representative of the input,xis the input variable, is the differential set symbol of the set operation.
Preferably, the judging unit comprises a guaranteed water level judging module and a warning water level judging module, the predicted water level is compared with the warning water level and the guaranteed water level of the area, if the predicted water level value exceeds the warning water level of the area, the position of the water level monitoring point is marked as a risk point for vigilance and withdrawal by the warning water level judging module, and if the predicted water level value exceeds the guaranteed water level of the area, the position of the water level monitoring point is marked as a nearby flood point by the guaranteed water level judging module, so that a flood warning judging result is obtained.
Preferably, the system further comprises an analysis unit and a communication unit, wherein:
the analysis unit is used for carrying out flood cause analysis of water level and high-sensitivity meteorological hydrologic factors according to flood warning judgment results, and judging the flood cause according to sensitivity if a flood dangerous situation occurs;
The communication unit sends the time evolution relation graph of the water level, the flood warning judgment result and the flood cause analysis result to related departments to timely develop flood prevention work.
Preferably, the flood countermeasure planning unit further comprises:
the flood response scheme making unit widely collects historical flood prevention schemes, makes a complete flood response scheme database, takes water levels, rainfall, weather and the like as flood characteristics, matches corresponding schemes from the response scheme database and outputs reasonable preliminary flood prevention schemes to flood related functional departments.
Preferably, the system further comprises a flood countermeasure pushing and opinion levying unit, wherein:
the flood response scheme pushing and opinion gathering unit pushes the preliminary flood prevention scheme to the mobile end of the flood management department from the back-end cloud server, and gathers expert opinions to determine a final flood response scheme.
The flood warning method for comprehensive meteorological hydrologic sensitivity comprises the following steps:
setting a plurality of hydrologic monitoring points at main rivers and floodgates of a research area, collecting relevant flood flow and water level by using an open channel flowmeter and a water level ruler, simultaneously obtaining average air temperature, precipitation, highest air temperature, lowest air temperature, average wind speed, relative humidity and sunshine time of a meteorological site, and storing historical hydrologic and meteorological data into an initial data set;
Performing hydrologic and meteorological data preprocessing on an initial data set, adopting outlier rejection and missing value filling for hydrologic data, firstly determining boundary vectors of a research area for the meteorological data, cutting and splicing to obtain the meteorological data of the research area, obtaining the meteorological data consistent with the time resolution of the hydrologic data by using a statistical downscaling method based on multiple linear regression and singular value decomposition, and storing the meteorological data into a meteorological hydrologic data set;
constructing a water level prediction model based on a transform deep learning method for the stored initial data set, obtaining an accurate water level prediction model by adjusting model super-parameters, and evaluating the performance of the water level prediction model;
the water level prediction result is displayed, a water level time evolution relation graph is obtained by using a trend analysis and significance analysis method, and then a weather and hydrologic sensitivity analysis graph is obtained by using an interpretable machine learning method;
and comparing the predicted water level with the guard and guaranteed water level of the area, marking the position of the water level monitoring point as a risk point for vigilance and withdrawal if the predicted water level value exceeds the guard water level of the area, marking the position of the water level monitoring point as a neighboring flood point if the predicted water level value exceeds the guaranteed water level of the area, and obtaining a flood guard judgment result.
In the technical scheme, the invention has the technical effects and advantages that:
according to the invention, the water level is accurately predicted under the condition that various meteorological and hydrologic elements exist, meanwhile, the water level change of a research area is quantitatively estimated, and the performance of a transducer is improved by using sparse attention and an additional nonlinear output layer; after a trained water level prediction model is obtained, the sensitivity of the water level to each meteorological and hydrologic element is analyzed by utilizing an interpretable machine learning method, if the water level prediction value exceeds the warning water level or guarantees the water level, the position of the water level monitoring point is marked as a risk point for warning and abstaining or a flood approaching point, and a flood warning judgment result is obtained;
after the flood warning judgment result is obtained, carrying out flood cause analysis of the water level and the high-sensitivity meteorological hydrologic factors according to the judgment result, and judging the flood cause according to the sensitivity if a flood danger is generated; after the analysis result of the flood causes is obtained, relevant departments are informed in time to collect historical flood prevention schemes, a complete flood response scheme database is formulated, water levels, rainfall, weather and the like are used as flood characteristics, corresponding schemes are matched from the response scheme database, reasonable preliminary flood prevention schemes are output to flood related functional departments, expert opinions are collected to determine final flood response schemes later, and the water levels are controlled in a reasonable range.
Drawings
For a clearer description of embodiments of the present application or of the solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments described in the present invention, and that other drawings may be obtained according to these drawings for a person skilled in the art.
FIG. 1 is a schematic block diagram of the integrated weather hydrographic sensitivity flood alert method and system of the present invention.
FIG. 2 is a flow chart of a method and system for providing flood alert in combination with weather hydrologic sensitivity according to the present invention.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments may be embodied in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.
The invention provides a flood warning system for comprehensive meteorological hydrologic sensitivity, which is shown in figure 1, and comprises a monitoring and data acquisition unit, a data driving water level prediction unit, a sensitivity analysis unit and a judgment unit;
The data-driven water level prediction unit comprises a preprocessing module, an evaluation module and a water level prediction module;
setting a plurality of hydrologic monitoring points at main rivers and floodgates of a research area, collecting flood related flow and water level at the monitoring points by using an open channel flowmeter and a water level ruler by using a monitoring and data acquisition unit, simultaneously acquiring average air temperature, precipitation, highest air temperature, lowest air temperature, average wind speed, relative humidity and sunshine time of a meteorological site, and storing historical hydrologic and meteorological data into an initial data set;
the method comprises the steps that a preprocessing module in a data-driven water level prediction unit is used for preprocessing hydrologic and meteorological data of an initial data set, outlier rejection and missing value filling are adopted for hydrologic data, boundary vectors of a research area are firstly determined for the meteorological data, arcGIS geographic information system series software is utilized for cutting and splicing so as to obtain the meteorological data of the research area, then a statistical downscaling method based on multiple linear regression and singular value decomposition is utilized to obtain the meteorological data with the same time resolution as the hydrologic data, and the meteorological data are stored into a meteorological hydrologic data set;
the water level prediction module in the data-driven water level prediction unit is used for constructing a water level prediction model based on a transform deep learning method for the stored initial data set, and an accurate water level prediction model is obtained by adjusting model super-parameters;
Evaluating the performance of the water level prediction model by an evaluation module;
the sensitivity analysis unit displays the water level prediction result, a water level time evolution relation diagram is obtained by using a trend analysis and significance analysis method, and then a weather and hydrologic sensitivity analysis diagram is obtained by using an interpretable machine learning method;
comparing the predicted water level with the guard and guaranteed water level of the area by a judging unit, marking the position of the water level monitoring point as a risk point for vigilance and abstinence if the predicted water level value exceeds the guard water level of the area, and marking the position of the water level monitoring point as a neighboring flood point if the predicted water level value exceeds the guaranteed water level of the area to obtain a flood guard judging result;
the statistical downscaling method comprises the following steps:
the statistical downscaling method comprises a multiple linear regression and singular value decomposition method;
the statistical downscaling method based on multiple linear regression is one of the simplest methods in the statistical downscaling method, and the linear relationship between meteorological data and a forecasting factor is assumed to exist, namely:wherein, the parameter meaning is: />For predicting quantity (average air temperature, precipitation, highest air temperature, lowest air temperature, average wind speed, relative humidity and sunshine time), A is regression coefficient matrix >Is a predictor;
the statistical downscaling method based on singular value decomposition is to search the main dimension of the meteorological data distribution, map the original high-dimensional meteorological data into a low-dimensional subspace to realize the data dimensionality reduction to obtain high-resolution meteorological data, and the definition of the singular value decomposition is as follows:wherein, the parameter meaning is: matrix->Middle column vector->Is a left singular vector, matrix->Is a singular value matrix, a diagonal matrix, +.>Is->Singular value, matrix->Middle column vector->Is right singular vector, matrix->Can be unfolded as follows:then intercept front +.>Singular values, in->Searching the nearest primary matrix in the dimensional space>Is>I.e. +.>。
The monitoring and data acquisition unit comprises a precipitation monitoring module, a water level monitoring module, a temperature monitoring module, an average wind speed monitoring module, a sunshine time monitoring module and a relative humidity monitoring module, and the meteorological hydrologic data set is constructed as follows:
the rainfall data is monitored and collected in the monitoring point by the rainfall monitoring module, the water level data is monitored and collected in the monitoring point by the water level monitoring module, the temperature data is monitored and collected in the monitoring point by the temperature monitoring module, the average air temperature, the highest air temperature and the lowest air temperature data are obtained through mathematical calculation respectively, the wind speed data is monitored and collected in the monitoring point by the average wind speed monitoring module, the sunshine time data is monitored and collected in the monitoring point by the sunshine time monitoring module, and the relative humidity data is monitored and collected in the monitoring area by the relative humidity monitoring module;
The preprocessing module firstly solves the problems of abnormal values and missing values existing in hydrologic data, eliminates the abnormal values and complements the missing values by adopting an interpolation method, wherein the interpolation method comprises an inverse distance weight method, a Kriging interpolation method, a natural neighbor method, a spline function interpolation method, a trend method and a depth interpolation method;
and then, the preprocessing module cuts and splices the meteorological data according to the boundary vector data of the research area by using ArcGIS geographic information system serial software, obtains the meteorological data with the same time resolution as the meteorological data by using a statistical downscaling method based on multiple linear regression and singular value decomposition, and splices the preprocessed meteorological data according to a time sequence, thereby constructing a meteorological hydrologic data set.
The mode of constructing the water level prediction model based on the transform deep learning method is as follows:
the water level prediction module divides a meteorological hydrologic data set into a training set, a verification set and a test set according to the ratio of 7:2:1, a water level prediction model based on a transducer deep learning method is obtained on the training set, super-parameters of the water level prediction model are adjusted on the verification set, performance indexes of the model are obtained on the test set, the performance indexes comprise four different statistical evaluation indexes, namely standard Root Mean Square Error (RMSE), mean Absolute Error (MAE), mean deviation error (MBE) and decision coefficient (R) 2 );
In the deep learning model, the super parameters refer to parameters which need to be set before model training, are not obtained through training data learning, but are manually set by researchers before training, in the training process of the water level prediction model, the selection of the super parameters has an important influence on the model performance, and the process of adjusting the super parameters on the verification set is used for finding the optimal super parameter combination so as to improve the model performance.
The following are some of the super-parameters that may be involved in adjusting the super-parameters of the water level prediction model on the validation set and their significance:
learning rate: the learning rate is the step size of the control model parameter update. Setting a suitable learning rate helps to avoid gradient explosion or gradient disappearance problems during model training. The learning rate is adjusted on the validation set to find an optimal value that is neither too large to cause concussion nor too small to cause slow convergence.
Batch size: the batch size defines the number of samples used by the model each time the parameters are updated. Different batch sizes may lead to differences in convergence speed and model performance of the training process. The batch size is adjusted on the validation set to find a suitable value that improves training efficiency without causing model over-fitting or under-fitting.
Number of layers and number of heads: for the transducer model, the number of layers and the number of heads are important super parameters. The number of layers represents the number of transducer layers in the model, while the number of heads represents the number of multi-headed attentions in each attentiveness mechanism. These superparameters are adjusted on the validation set to find a model that both adequately expresses the data and is not overly complex.
Regularization parameters: regularization parameters include L1 regularization, L2 regularization, etc., to prevent model overfitting. Regularization parameters are adjusted on the validation set to find an optimal value that suppresses overfitting without affecting the model's learning efficient features.
Parameters of the attention mechanism: for the transducer model, parameters of the attention mechanism such as the scaling factor are also super parameters that need to be adjusted. These parameters affect the degree of attention of the model to the input sequence while learning. These parameters are adjusted on the validation set in order to find the optimal way of attention.
The water level prediction model based on the transform deep learning method consists of a two-layer encoder and a one-layer decoder, wherein:
each layer of encoder consists of two sub-layers, wherein each sub-layer comprises a multi-head sparse attention and position full-connection feedforward network module, and the sub-layers are combined through residual connection and layer normalization;
The decoder comprises three sub-layers, namely a multi-head mask sparse attention, a multi-head sparse attention and a position full-connection feedforward network module, and is combined through residual connection and layer normalization;
the multi-head sparse attention module adopts masking operation to determine the sparsity of vectors, the probability is exclusively allocated to the most important elements in the input sequence, then the attention score is exclusively calculated for the selected positions, and the output of each head is concurrently calculated by utilizing a multi-head mechanism;
the multi-head mask sparse attention module is used for masking information of all positions after the current moment;
the position full-connection feedforward network module is a two-layer full-connection layer, and the corresponding expression is:;
wherein the residual connection is used for solving the problem of multi-layer network training;
wherein layer normalization is used to accelerate network convergence.
The trend analysis, significance analysis and interpretable machine learning method is specifically as follows:
the trend analysis method is to research the trend change of the water level by using a unitary linear regression analysis method, and the specific equation is as follows:wherein, the parameter meaning is: n is the number of days, & lt + & gt>Is the firstiDay water level value, ">For multiple days->Value linear fit slope, when +. >When indicating +.>Gradually increasing day by day, and gradually decreasing otherwise;
the significance analysis method comprises calculating a test statistic related to the length of water level sequence by Mann-Kendall (M-K) method, and performing significance test to determine whether there is trend in water level sequence, for a water level sequence with length of nFirst calculate +.>WhereinThen statistics->Calculate +.about.in its normal distribution>The value of the sum of the values,,/>wherein, the parameter meaning is: z is a significance value, S is a statistic, < ->Is a water level sequence variable, if->The water level sequence is considered to have a significant change trend through the significance test, wherein the significance test shows that the confidence coefficient is 90%, 95% and 99% respectively;
the sensitivity analysis unit comprises a hydrologic sensitivity analysis module, a meteorological sensitivity analysis module and a water level result display module, wherein the water level result display module displays a water level prediction result, a water level time evolution relation diagram is obtained by using a trend analysis and significance analysis method, and the hydrologic sensitivity analysis module and the meteorological sensitivity analysis module then obtain a meteorological and hydrologic sensitivity analysis diagram by using an interpretable machine learning method;
an interpretable machine learning method is to effectively determine the contribution of each variable to the overall model by using an interpretation model with a trained machine learning model and input variables, as follows:
Wherein, the parameter meaning is:Erepresenting an interpretation model of the model,irepresenting the number of input variables to be processed,tsimplification of representative feature vectors, ++>Representing the contribution of each variable to the machine learning model, < ->The expression of (2) is as follows:wherein, the parameter meaning is:Ma machine learning model representative of the input,xis the input variable, is the differential set symbol of the set operation.
The judging unit comprises a guaranteed water level judging module and a warning water level judging module, the predicted water level is compared with the warning water level of the area and the guaranteed water level, if the predicted water level value exceeds the warning water level of the area, the position of the water level monitoring point is marked as a risk point for warning and abstaining through the warning water level judging module, and if the predicted water level value exceeds the guaranteed water level of the area, the position of the water level monitoring point is marked as a nearby flood point through the guaranteed water level judging module, so that a flood warning judging result is obtained.
Further comprising an analysis unit and a communication unit, wherein:
the analysis unit is used for carrying out flood cause analysis of water level and high-sensitivity meteorological hydrologic factors according to flood warning judgment results, and judging the flood cause according to sensitivity if a flood dangerous situation occurs;
the communication unit sends the time evolution relation graph of the water level, the flood warning judgment result and the flood cause analysis result to related departments to timely develop flood prevention work.
Further comprising a flood countermeasure planning unit, wherein:
the flood response scheme making unit widely collects historical flood prevention schemes, makes a complete flood response scheme database, takes water levels, rainfall, weather and the like as flood characteristics, matches corresponding schemes from the response scheme database and outputs reasonable preliminary flood prevention schemes to flood related functional departments.
The system further comprises a flood countermeasure pushing and opinion gathering unit, wherein:
the flood response scheme pushing and opinion gathering unit pushes the preliminary flood prevention scheme to the mobile end of the flood management department from the back-end cloud server, and gathers expert opinions to determine a final flood response scheme.
The invention provides a flood warning method for comprehensive meteorological hydrologic sensitivity, which is shown in figure 2 and comprises the following steps:
setting a plurality of hydrologic monitoring points at main rivers and floodgates of a research area, collecting relevant flood flow and water level by using an open channel flowmeter and a water level ruler, simultaneously obtaining average air temperature, precipitation, highest air temperature, lowest air temperature, average wind speed, relative humidity and sunshine time of a meteorological site, and storing historical hydrologic and meteorological data into an initial data set;
performing hydrologic and meteorological data preprocessing on an initial data set, adopting outlier rejection and missing value filling for hydrologic data, firstly determining boundary vectors of a research area for the meteorological data, cutting and splicing to obtain the meteorological data of the research area, obtaining the meteorological data consistent with the time resolution of the hydrologic data by using a statistical downscaling method based on multiple linear regression and singular value decomposition, and storing the meteorological data into a meteorological hydrologic data set;
Constructing a water level prediction model based on a transform deep learning method for the stored initial data set, obtaining an accurate water level prediction model by adjusting model super-parameters, and evaluating the performance of the water level prediction model;
the water level prediction result is displayed, a water level time evolution relation graph is obtained by using a trend analysis and significance analysis method, and then a weather and hydrologic sensitivity analysis graph is obtained by using an interpretable machine learning method;
comparing the predicted water level with the guard and guaranteed water level of the area, if the predicted water level value exceeds the guard water level of the area, marking the position of the water level monitoring point as a risk point for vigilance and withdrawal, and if the predicted water level value exceeds the guaranteed water level of the area, marking the position of the water level monitoring point as a neighboring flood point to obtain a flood guard judgment result;
the flood warning method for comprehensive weather hydrographic sensitivity provided by the embodiment of the invention is realized through the flood warning system for comprehensive weather hydrographic sensitivity, and the specific method and flow of the flood warning system for comprehensive weather hydrographic sensitivity are detailed in the embodiment of the flood warning method for comprehensive weather hydrographic sensitivity, and are not repeated here.
According to the invention, the water level is accurately predicted under the condition that various meteorological and hydrologic elements exist, meanwhile, the water level change of a research area is quantitatively estimated, and the performance of a transducer is improved by using sparse attention and an additional nonlinear output layer; after a trained water level prediction model is obtained, the sensitivity of the water level to each meteorological and hydrologic element is analyzed by utilizing an interpretable machine learning method, if the water level prediction value exceeds the warning water level or guarantees the water level, the position of the water level monitoring point is marked as a risk point for warning and abstaining or a flood approaching point, and a flood warning judgment result is obtained;
after the flood warning judgment result is obtained, carrying out flood cause analysis of the water level and the high-sensitivity meteorological hydrologic factors according to the judgment result, and judging the flood cause according to the sensitivity if a flood danger is generated; after the analysis result of the flood causes is obtained, relevant departments are informed in time to collect historical flood prevention schemes, a complete flood response scheme database is formulated, water levels, rainfall, weather and the like are used as flood characteristics, corresponding schemes are matched from the response scheme database, reasonable preliminary flood prevention schemes are output to flood related functional departments, expert opinions are collected to determine final flood response schemes later, and the water levels are controlled in a reasonable range.
While certain exemplary embodiments of the present invention have been described above by way of illustration only, it will be apparent to those of ordinary skill in the art that modifications may be made to the described embodiments in various different ways without departing from the spirit and scope of the invention. Accordingly, the drawings and description are to be regarded as illustrative in nature and not as restrictive of the scope of the invention, which is defined by the appended claims.
Claims (10)
1. The flood warning system for comprehensive meteorological hydrologic sensitivity is characterized by comprising a monitoring and data acquisition unit, a data driving water level prediction unit, a sensitivity analysis unit and a judgment unit;
the data-driven water level prediction unit comprises a preprocessing module, an evaluation module and a water level prediction module;
setting a plurality of hydrologic monitoring points at main rivers and floodgates of a research area, collecting flood related flow and water level at the monitoring points by using an open channel flowmeter and a water level ruler by using a monitoring and data acquisition unit, simultaneously acquiring average air temperature, precipitation, highest air temperature, lowest air temperature, average wind speed, relative humidity and sunshine time of a meteorological site, and storing historical hydrologic and meteorological data into an initial data set;
The method comprises the steps that a preprocessing module in a data-driven water level prediction unit is used for preprocessing hydrologic and meteorological data of an initial data set, outlier rejection and missing value filling are adopted for hydrologic data, boundary vectors of a research area are firstly determined for the meteorological data, the meteorological data of the research area are obtained by cutting and splicing, then the meteorological data consistent with the time resolution of the hydrologic data are obtained by a statistical downscaling method based on multiple linear regression and singular value decomposition, and the meteorological data are stored into a meteorological hydrologic data set;
the water level prediction module in the data-driven water level prediction unit is used for constructing a water level prediction model based on a transform deep learning method for the stored initial data set, and an accurate water level prediction model is obtained by adjusting model super-parameters;
evaluating the performance of the water level prediction model by an evaluation module;
the sensitivity analysis unit displays the water level prediction result, a water level time evolution relation diagram is obtained by using a trend analysis and significance analysis method, and then a weather and hydrologic sensitivity analysis diagram is obtained by using an interpretable machine learning method;
and comparing the predicted water level with the guard and guaranteed water level of the area by a judging unit, marking the position of the water level monitoring point as a risk point for vigilance and abstainment if the predicted water level value exceeds the guard water level of the area, and marking the position of the water level monitoring point as a neighboring flood point if the predicted water level value exceeds the guaranteed water level of the area to obtain a flood guard judging result.
2. The comprehensive meteorological hydrologic sensitivity flood warning system according to claim 1, wherein the statistical downscaling method based on multiple linear regression and singular value decomposition comprises the following specific steps:
the weather data and the forecasting factors are expressed by linear relation, namely:wherein, the parameter meaning is: />For the predicted quantity, A is regression coefficient matrix, < ->Is a predictor;
mapping original high-dimensional meteorological data into a low-dimensional subspace to realize data dimension reduction to obtain high-resolution meteorological data, and obtaining the high-resolution meteorological dataThe definition of outlier decomposition is:wherein, the parameter meaning is: matrix->Middle column vector->As left singular vectors, matrixIs a singular value matrix, a diagonal matrix, +.>Is->Singular value, matrix->Middle column vector->Is right singular vector, matrix->The unfolding is as follows: />Then intercept front +.>Singular values, in->Searching the original matrix in the dimensional space>Low-dimensional matrix with shortest distance/>Then: />。
3. The flood warning system for comprehensive meteorological hydrologic sensitivity according to claim 1, wherein the monitoring and data acquisition unit comprises a precipitation monitoring module, a water level monitoring module, a temperature monitoring module, an average wind speed monitoring module, a sunshine time monitoring module and a relative humidity monitoring module, and the meteorological hydrologic data set is constructed in the following manner:
The rainfall data is monitored and collected in the monitoring point by the rainfall monitoring module, the water level data is monitored and collected in the monitoring point by the water level monitoring module, the temperature data is monitored and collected in the monitoring point by the temperature monitoring module, the average air temperature, the highest air temperature and the lowest air temperature data are obtained through mathematical calculation respectively, the wind speed data is monitored and collected in the monitoring point by the average wind speed monitoring module, the sunshine time data is monitored and collected in the monitoring point by the sunshine time monitoring module, and the relative humidity data is monitored and collected in the monitoring area by the relative humidity monitoring module;
the preprocessing module screens out abnormal values and missing values existing in the hydrological data, eliminates the abnormal values and complements the missing values by adopting an interpolation method, wherein the interpolation method is one of an inverse distance weight method, a Kriging interpolation method, a natural neighbor method, a spline function interpolation method, a trend method and a depth interpolation method;
and then the preprocessing module cuts and splices the meteorological data according to the boundary vector data of the research area, obtains the meteorological data with the same time resolution as the hydrologic data by using a statistical downscaling method based on multiple linear regression and singular value decomposition, splices the preprocessed meteorological hydrologic data according to a time sequence, and constructs a meteorological hydrologic data set.
4. The integrated weather hydrographic sensitivity flood warning system according to claim 1, wherein the water level prediction model based on the transform deep learning method is constructed as follows:
the water level prediction module divides a meteorological hydrologic data set into a training set, a verification set and a test set according to the ratio of 7:2:1, a water level prediction model based on a transducer deep learning method is obtained on the training set, super-parameters of the water level prediction model are adjusted on the verification set, performance indexes of the model are obtained on the test set, and the performance indexes comprise four different statistical evaluation indexes, a standard root mean square error, an average absolute error, an average deviation error and a decision coefficient;
the water level prediction model based on the transform deep learning method consists of a two-layer encoder and a one-layer decoder, wherein:
each layer of encoder consists of two sub-layers, wherein each sub-layer comprises a multi-head sparse attention and position full-connection feedforward network module, and the sub-layers are combined through residual connection and layer normalization;
the decoder comprises three sub-layers, namely a multi-head mask sparse attention, a multi-head sparse attention and a position full-connection feedforward network module, and is combined through residual connection and layer normalization;
The multi-head sparse attention module adopts masking operation to determine the sparsity of vectors, the probability is specially allocated to the most important elements in the input sequence, then the attention score is calculated for the selected positions, and the output of each head is calculated concurrently by utilizing a multi-head mechanism;
the multi-head mask sparse attention module is used for masking information of all positions after the current moment;
the position full-connection feedforward network module is a two-layer full-connection layer, and the corresponding expression is:;
wherein the residual connection is used for solving the problem of multi-layer network training;
wherein layer normalization is used to accelerate network convergence.
5. The integrated weather hydrographic sensitive flood warning system according to claim 1, wherein the trend analysis, saliency analysis and interpretable machine learning method is as follows:
the trend change of the water level is researched by using a unitary linear regression through a trend analysis method, and the specific equation is as follows:wherein n is the number of days, < >>Is the firstiDay water level value, ">For multiple days->Value linear fit slope, when +.>When indicating +.>Gradually increasing day by day, and gradually decreasing otherwise;
the significance analysis method comprises calculating a test statistic related to the length of water level sequence by Mann-Kendall (M-K) method, and performing significance test to determine whether there is trend in water level sequence, for a water level sequence with length of n First calculate +.>WhereinThen statistics->Calculate +.about.in its normal distribution>The value of the sum of the values,,/>wherein, the parameter meaning is: z is a significance value, S is a statistic, < ->Is a water level sequence variable, if->The water level sequence is considered to have a significant change trend through the significance test, wherein the significance test shows that the confidence coefficient is 90%, 95% and 99% respectively;
the sensitivity analysis unit comprises a hydrologic sensitivity analysis module, a meteorological sensitivity analysis module and a water level result display module, wherein the water level result display module displays a water level prediction result, a water level time evolution relation diagram is obtained by using a trend analysis and significance analysis method, and the hydrologic sensitivity analysis module and the meteorological sensitivity analysis module then obtain a meteorological and hydrologic sensitivity analysis diagram by using an interpretable machine learning method;
the contribution of each variable to the overall model is determined by an interpretable machine learning method, the equation being as follows:
wherein, the parameter meaning is:Erepresenting an interpretation model of the model,irepresenting the number of input variables to be processed,tsimplification of representative feature vectors, ++>Representing each variable for machine learning modelContribution of type->The expression of (2) is as follows:wherein, the method comprises the steps of, wherein, MA machine learning model representative of the input,xis the input variable, is the differential set symbol of the set operation.
6. The flood warning system for comprehensive meteorological hydrologic sensitivity according to claim 1, wherein the judging unit comprises a guaranteed water level judging module and a warning water level judging module, the predicted water level is compared with the warning and guaranteed water levels of the area, if the predicted water level value exceeds the warning water level of the area, the position of the water level monitoring point is marked as a risk point for vigilance and withdrawal by the warning water level judging module, and if the predicted water level value exceeds the guaranteed water level of the area, the position of the water level monitoring point is marked as a nearby flood point by the guaranteed water level judging module, and a flood warning judging result is obtained.
7. The integrated weather hydrographic sensitive flood warning system according to claim 1, further comprising an analysis unit and a communication unit, wherein:
the analysis unit is used for carrying out flood cause analysis of water level and high-sensitivity meteorological hydrologic factors according to flood warning judgment results, and judging the flood cause according to sensitivity if a flood dangerous situation occurs;
the communication unit sends the time evolution relation graph of the water level, the flood warning judgment result and the flood cause analysis result to related departments to timely develop flood prevention work.
8. The integrated weather hydrographic-sensitive flood alert system according to claim 1, further comprising a flood response scheme formulation unit, wherein:
the flood response scheme making unit widely collects historical flood prevention schemes, makes a complete flood response scheme database, takes water levels, rainfall, weather and the like as flood characteristics, matches corresponding schemes from the response scheme database and outputs reasonable preliminary flood prevention schemes to flood related functional departments.
9. The integrated weather hydrographic sensitive flood alert system according to claim 1, further comprising a flood countermeasure proposal pushing and opinion solicitation unit, wherein:
the flood response scheme pushing and opinion gathering unit pushes the preliminary flood prevention scheme to the mobile end of the flood management department from the back-end cloud server, and gathers expert opinions to determine a final flood response scheme.
10. A method of flood alert for integrated weather hydrographic sensitivity based on the integrated weather hydrographic sensitivity flood alert system implementation of any one of claims 1-9, comprising the steps of:
setting a plurality of hydrologic monitoring points at main rivers and floodgates of a research area, collecting relevant flood flow and water level by using an open channel flowmeter and a water level ruler, simultaneously obtaining average air temperature, precipitation, highest air temperature, lowest air temperature, average wind speed, relative humidity and sunshine time of a meteorological site, and storing historical hydrologic and meteorological data into an initial data set;
Performing hydrologic and meteorological data preprocessing on an initial data set, adopting outlier rejection and missing value filling for hydrologic data, firstly determining boundary vectors of a research area for the meteorological data, cutting and splicing to obtain the meteorological data of the research area, obtaining the meteorological data consistent with the time resolution of the hydrologic data by using a statistical downscaling method based on multiple linear regression and singular value decomposition, and storing the meteorological data into a meteorological hydrologic data set;
constructing a water level prediction model based on a transform deep learning method for the stored initial data set, obtaining an accurate water level prediction model by adjusting model super-parameters, and evaluating the performance of the water level prediction model;
the water level prediction result is displayed, a water level time evolution relation graph is obtained by using a trend analysis and significance analysis method, and then a weather and hydrologic sensitivity analysis graph is obtained by using an interpretable machine learning method;
and comparing the predicted water level with the guard and guaranteed water level of the area, marking the position of the water level monitoring point as a risk point for vigilance and withdrawal if the predicted water level value exceeds the guard water level of the area, marking the position of the water level monitoring point as a neighboring flood point if the predicted water level value exceeds the guaranteed water level of the area, and obtaining a flood guard judgment result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410050804.6A CN117575873B (en) | 2024-01-15 | 2024-01-15 | Flood warning method and system for comprehensive meteorological hydrologic sensitivity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410050804.6A CN117575873B (en) | 2024-01-15 | 2024-01-15 | Flood warning method and system for comprehensive meteorological hydrologic sensitivity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117575873A true CN117575873A (en) | 2024-02-20 |
CN117575873B CN117575873B (en) | 2024-04-05 |
Family
ID=89890346
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410050804.6A Active CN117575873B (en) | 2024-01-15 | 2024-01-15 | Flood warning method and system for comprehensive meteorological hydrologic sensitivity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117575873B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117827990A (en) * | 2024-03-04 | 2024-04-05 | 山东齐鸿工程建设有限公司 | Data processing method and system suitable for hydraulic engineering |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108986413A (en) * | 2018-08-16 | 2018-12-11 | 长江水利委员会长江科学院 | A kind of glade disaster monitoring and method for early warning based on multi-source information data |
CN112561178A (en) * | 2020-12-21 | 2021-03-26 | 国网甘肃省电力公司电力科学研究院 | Distributed photovoltaic output prediction method considering microclimate factors |
CN113219555A (en) * | 2020-09-30 | 2021-08-06 | 国家气候中心 | Short-term climate prediction method based on statistical downscaling technology |
CN114240119A (en) * | 2021-12-10 | 2022-03-25 | 浙江水利水电学院 | Digital twin-based flood control and waterlogging prevention system and early warning method for whole elements of territorial universe |
CN114819386A (en) * | 2022-05-19 | 2022-07-29 | 河海大学 | Conv-Transformer-based flood forecasting method |
CN115186858A (en) * | 2022-03-29 | 2022-10-14 | 南京南瑞水利水电科技有限公司 | Transformer substation flood inundation risk early warning method and system based on different influence types |
US20230281493A1 (en) * | 2022-03-02 | 2023-09-07 | University Of Iowa Research Foundation | Forecasting method for flood crest magnitude and arrival time |
-
2024
- 2024-01-15 CN CN202410050804.6A patent/CN117575873B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108986413A (en) * | 2018-08-16 | 2018-12-11 | 长江水利委员会长江科学院 | A kind of glade disaster monitoring and method for early warning based on multi-source information data |
CN113219555A (en) * | 2020-09-30 | 2021-08-06 | 国家气候中心 | Short-term climate prediction method based on statistical downscaling technology |
CN112561178A (en) * | 2020-12-21 | 2021-03-26 | 国网甘肃省电力公司电力科学研究院 | Distributed photovoltaic output prediction method considering microclimate factors |
CN114240119A (en) * | 2021-12-10 | 2022-03-25 | 浙江水利水电学院 | Digital twin-based flood control and waterlogging prevention system and early warning method for whole elements of territorial universe |
US20230281493A1 (en) * | 2022-03-02 | 2023-09-07 | University Of Iowa Research Foundation | Forecasting method for flood crest magnitude and arrival time |
CN115186858A (en) * | 2022-03-29 | 2022-10-14 | 南京南瑞水利水电科技有限公司 | Transformer substation flood inundation risk early warning method and system based on different influence types |
CN114819386A (en) * | 2022-05-19 | 2022-07-29 | 河海大学 | Conv-Transformer-based flood forecasting method |
Non-Patent Citations (7)
Title |
---|
MARCO CASTANGIA等: "Transformer neural networks for interpretable flood forecasting", 《ENVIRONMENTAL MODELLING & SOFTWARE》, 9 November 2022 (2022-11-09) * |
ZHANG, Z.等: "Implementing a new fully stepwise decomposition-based sampling technique for the hybrid water level forecasting model in real-world application", 《ARXIV》, 12 October 2023 (2023-10-12) * |
刘永和;郭维栋;冯锦明;张可欣;: "气象资料的统计降尺度方法综述", 地球科学进展, no. 08, 10 August 2011 (2011-08-10) * |
张子谦等: "基于SSA-LSTM模型的日水位预测—以涡河流域涡阳闸为例", 《计算机***应用》, 18 November 2022 (2022-11-18) * |
李欣宇等: "融合Transformer与残差通道注意力的恶劣场景水位智能检测方法", 《中国优秀硕士论文 工程科技Ⅱ辑》, 21 December 2022 (2022-12-21) * |
杨继;邵利民;郭超;易海祁;曹祥村;: "混沌理论在水文***中的应用分析及海浪预报探讨", 海洋技术, no. 01, 15 March 2008 (2008-03-15) * |
潘小多;李新;钞振华;: "区域尺度近地表气候要素驱动数据研制的研究综述", 地球科学进展, no. 12, 31 December 2010 (2010-12-31) * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117827990A (en) * | 2024-03-04 | 2024-04-05 | 山东齐鸿工程建设有限公司 | Data processing method and system suitable for hydraulic engineering |
CN117827990B (en) * | 2024-03-04 | 2024-05-10 | 山东齐鸿工程建设有限公司 | Data processing method and system suitable for hydraulic engineering |
Also Published As
Publication number | Publication date |
---|---|
CN117575873B (en) | 2024-04-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111798051B (en) | Air quality space-time prediction method based on long-term and short-term memory neural network | |
Chang et al. | An LSTM-based aggregated model for air pollution forecasting | |
Bass et al. | Surrogate modeling of joint flood risk across coastal watersheds | |
CN111815037B (en) | Interpretable short-critical extreme rainfall prediction method based on attention mechanism | |
CN117575873B (en) | Flood warning method and system for comprehensive meteorological hydrologic sensitivity | |
JP4807565B2 (en) | Flow prediction device | |
CN109448361B (en) | Resident traffic travel flow prediction system and prediction method thereof | |
CN104408900A (en) | Dynamic optimization based neural network flood warning device and method | |
CN101354757A (en) | Method for predicting dynamic risk and vulnerability under fine dimension | |
CN113343563B (en) | Landslide susceptibility evaluation method based on automatic sample selection and surface deformation rate | |
CN111325947B (en) | Storm surge flood early warning method and system based on long-term and short-term memory network | |
CN114611778A (en) | Reservoir water level early warning method and system based on warehousing flow | |
Karyotis et al. | Deep learning for flood forecasting and monitoring in urban environments | |
CN117332909B (en) | Multi-scale urban waterlogging road traffic exposure prediction method based on intelligent agent | |
CN117238126A (en) | Traffic accident risk assessment method under continuous flow road scene | |
CN117556197A (en) | Typhoon vortex initialization method based on artificial intelligence | |
CN114385611A (en) | Precipitation prediction method and system based on artificial intelligence algorithm and knowledge graph | |
Nguyen et al. | A framework for flood depth using hydrodynamic modeling and machine learning in the coastal province of Vietnam | |
CN116703004B (en) | Water system river basin intelligent patrol method and device based on pre-training model | |
CN117332291A (en) | Distributed photovoltaic-oriented regional resource monitoring layout method and system | |
CN117233869A (en) | Site short-term wind speed prediction method based on GRU-BiTCN | |
Velasco et al. | Days-ahead water level forecasting using artificial neural networks for watersheds | |
Mandal et al. | Prediction of Wind Speed using Machine Learning | |
CN114662894A (en) | Urban waterlogging early warning method, device, equipment and storage medium | |
Jonnalagadda et al. | Spatial-Temporal Forecast of the probability distribution of Oceanic Nino Index for various lead times. |
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 |