CN111199298A - Flood forecasting method and system based on neural network - Google Patents
Flood forecasting method and system based on neural network Download PDFInfo
- Publication number
- CN111199298A CN111199298A CN201811454308.8A CN201811454308A CN111199298A CN 111199298 A CN111199298 A CN 111199298A CN 201811454308 A CN201811454308 A CN 201811454308A CN 111199298 A CN111199298 A CN 111199298A
- Authority
- CN
- China
- Prior art keywords
- flood
- rainfall
- data
- neural network
- forecasting
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/02—Reservations, e.g. for tickets, services or events
-
- 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
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Abstract
The invention discloses a real-time flood forecasting method based on continuous multi-prophase period data recursion of a neural network, which is characterized in that on the basis of an established data warehouse taking a research object as a theme, the average rainfall of an area calculated by a Thiessen polygon method is added, the original data warehouse is adjusted and expanded, the rainfall and water conservancy project scheduling data of a plurality of continuous prophase periods of a flood field which are transformed according to the idea of a flood forecasting synthesis flow method in the data warehouse are innovatively adopted as a data analysis unit to carry out neural network analysis modeling on at least one of the flood peak flow and the water level of the forecasting object in the forecasting period, and the flood peak flow/water level of the research object in the forecasting period is recurred in real time.
Description
Technical Field
The invention relates to a flood forecasting method and system based on a neural network.
Background
The modern flood forecasting technology is established on the technical basis of the existing weather forecasting and hydrologic forecasting theory, and has three main aspects: firstly, researching a quantitative precipitation forecasting technology which can meet special requirements of flood forecasting; secondly, establishing a mode of organically combining quantitative precipitation forecast and flood forecast; and thirdly, a real-time flood forecasting method, which comprises river basin product converging and river channel flood forecasting. At present, the real-time flood forecasting technology mainly adopts a physical process and a mathematical method for analyzing, modeling and calculating, and carries out real-time correction on the structure, parameters or model output of a model according to newly monitored rainfall, water level or flow data, thereby continuously improving the forecasting precision of process flood.
At present, a national hydrological database is built, various water level and rainfall measuring stations are constructed to a village in a refined mode, and a good platform is provided for hydrological information query and data analysis by combining with rapidly developed internet technology.
However, at present, the hydrological department in China mainly depends on a flood forecasting system to forecast, and most traditional hydrological algorithms are adopted to forecast the flood, and due to the fact that the structure and the algorithms of a system database are relatively fixed, even if all factors influencing the flood process are analyzed and calculated, the complicated and complicated relationships among the factors and the situations of rain, water and work with unusual changes are difficult to balance, and part of forecasts are easy to have low precision, short forecast period and low speed of forecasting operation, so that the flood prevention decision needs are difficult to meet.
In a document published by the inventor of the application, (document 1, a master thesis of Guangxi university, namely, "research and application of Hoxiu station flood forecast in Hoxiu basin based on neural network, Susheng by the author, 12 months in 2017), a flood forecasting method is provided, which is applied to Hoxiu station flood forecast in Hoxiu basin, and achieves certain breakthrough in solving the technical problems in the prior art, but the forecasting precision can only reach the third level, and the improvement needs exist.
Disclosure of Invention
The invention aims to provide a flood forecasting method and system based on a neural network, and the forecasting precision is improved.
Therefore, the flood forecasting method based on the neural network provided by the invention comprises the following steps: s1, adding the area average rainfall calculated by the Thiessen polygon method on the basis of a data warehouse taking the research object as a theme; s2, adjusting and expanding the data warehouse; and S3, performing neural network analysis modeling by adopting the rainfall and water conservancy project scheduling data of a plurality of continuous early periods of a plurality of flood fields and the water conservancy project scheduling data in the data warehouse after conversion according to the flood forecast synthetic flow method, and performing real-time recursion to obtain at least one of the flood peak flow and the water level of the forecast study object.
In some embodiments, the following technical features are also included:
step S1, according to the Thiessen polygon method, forming a Thiessen polygon network in the whole flow area or a certain area, wherein each Thiessen polygon corresponds to a rainfall observation point in the network, and the rainfall of the rainfall observation point represents the rainfall in the corresponding Thiessen polygon; therefore, the regional average rainfall is calculated through the rainfall of all the rainfall observation points in the network.
4. The method for forming the Thiessen polygonal net comprises the following steps: firstly, drawing a rain collecting area of a certain rainfall station of a drainage basin required by research on a topographic map, and marking the position of the rainfall station; then deriving a region map; and finally, according to the principle and the characteristics of the Thiessen polygon, constructing the Thiessen polygon on the regional map to obtain the Thiessen polygon network of the rain collecting region required by research.
4. The method and system for neural network-based flood forecasting according to claim 3, wherein the regional average rainfall is
Can be obtained according to the following formula:
wherein n is the number of Thiessen polygons, xiFor the rainfall of each rainfall observation point, a is the rainfall weight coefficient, which is the ratio of each corresponding Thiessen polygon to the area of the region.
In step S2, the adjusting and expanding the data warehouse includes: increasing the area average rainfall as a main factor influencing the water level of the rainfall station to be forecasted; and adjusting the flood discharge flow of the upstream reservoir into the delivery flow, wherein the delivery flow is the flood discharge flow, the power generation flow and the irrigation flow.
In step S3, according to the synthetic flow method, flow synthesis is performed on each factor according to the propagation time, that is, row-column displacement transformation is performed on the corresponding propagation time of the event sample data, corresponding area average rainfall is calculated for the rainfall data in the transformed data unit, the calculation result is used as a new factor to participate in neural network analysis, so that the data unit to be subjected to neural network analysis embodies the actual influence on the water level of the rainfall station to be forecasted, and the data of continuous multiple previous-period periods correspondingly transformed according to the synthetic flow method is added to the data unit to be subjected to neural network analysis.
In step S3, the number of flood process sessions that can be modeled for each neural network analysis is not limited. .
The forecast water level of the station to be forecasted is made at n +1, the rainfall station data of the station to be forecasted at n +1 is obtained through weather forecast, and forecast with forecast period of 1 hour is completed; recording the minimum value of the propagation time of the upstream station as m, adjusting and expanding a data unit for m hours in the future, namely data of continuous multiple early-stage time intervals of factors influencing a forecast object, which are correspondingly transformed according to a synthetic flow method, can be complete, reasonable and effective, and further the water level and the flow of the station to be forecasted when the station reaches n + m can be gradually and reasonably extrapolated, namely the forecast period is drawn to m hours; wherein m and n are positive integers.
The method comprises the steps of adopting the rainfall and water conservancy project scheduling data of a plurality of continuous early periods of a plurality of flood fields converted according to a flood forecast synthetic flow method as a data analysis unit to carry out neural network analysis modeling on at least one of flood peak flow and water level of a forecast object, carrying out preliminary precision inspection on a model obtained through neural network analysis by directly using super-alert flood event sample data which is not involved in the compilation of a flood forecast scheme, carrying out complete and standard precision evaluation and inspection on the scheme if the inspection result is accurate, repeatedly carrying out modeling, inspection and error correction, properly adjusting a forecast period according to the precision inspection result, and finally obtaining a set of relatively satisfactory flood forecast scheme.
The invention also provides a flood forecasting system based on the neural network, which adopts the method to forecast the flood.
The present invention also proposes a computer medium storing a computer program executable to implement the flood forecasting method described above.
The flood forecasting method and system based on the neural network are a continuous multi-period data recursion real-time flood forecasting method based on the neural network, on the basis of an established data warehouse taking a research object as a subject, the regional average rainfall calculated by the Thiessen polygon method is added, the original data warehouse is adjusted and expanded according to a flood forecasting synthetic flow method, the rain condition and the hydraulic engineering scheduling data of continuous multi-period early periods in the data warehouse are innovatively adopted as a data analysis unit to carry out neural network analysis modeling, the flood peak flow or/and the water level of the research object of the forecasting period are recurred in real time, and the forecasting precision is improved.
In some embodiments, the following benefits are also included:
the inventor proves that the real-time and accurate calculation result of the average rainfall capacity of the area plays an important role in the accuracy of flood forecast. Generally, the rainfall observed by a rainfall station can only represent the rainfall condition of a small range around the rainfall station, so the rainfall data of a single rainfall station cannot be used as the basis for forecasting and evaluating the flood, and the rainfall data of all the rainfall stations in the whole drainage basin or a certain area needs to be adopted to calculate the average rainfall of the area.
The data warehouse in the prior art is adjusted and expanded, and the data for neural network analysis in the adjusted and expanded data warehouse is more detailed and relatively increased, so that more powerful data support is provided for the neural network analysis.
The method comprises the steps of innovatively adopting water and rain conditions of a plurality of continuous flood fields and a plurality of early periods and hydraulic engineering scheduling data which are transformed according to the idea of a flood forecasting synthetic flow method as a data analysis unit to carry out neural network analysis modeling, and comprehensively representing the flood development process, so that the neural network analysis can fully dig the rules implicit in the flood process; the possibility that unreasonable fitting occurs when the sequence number used in the document 1 represents the flood development process for analysis is avoided; the number of flood process sessions that can be analyzed at a time is not limited.
In one embodiment of the present application, the accuracy of the test that is forecasted is successfully improved from class C to class B levels.
Drawings
FIG. 1 is a schematic flow chart of an embodiment of the present invention.
Fig. 2 is a schematic view of a water system in the Hejiang river basin to which the embodiment of the invention is applied.
Fig. 2a, 2b and 2c are schematic diagrams of the construction process of the thiessen polygon in the rain collecting area.
Fig. 3 is a causal graph (fishbone graph) of the change in the water level at the huzhou station.
FIG. 4 is a schematic diagram of a multi-layer feedforward neural network including an implied layer.
FIG. 5A is a schematic diagram of a synthesized flow rate according to an embodiment of the present invention.
Fig. 5B is a corresponding neural network relationship diagram.
Fig. 6 is a line graph of event sample accuracy assessments for an embodiment 2016.5.20 of the present invention.
Fig. 7 is a line graph of event sample accuracy assessments for an embodiment 2016.6.13 of the present invention.
Fig. 8 is a diagram showing the results of the forecast scenario test of the present embodiment.
Fig. 9 is a diagram showing the results of the prediction scheme of comparative document 1.
Detailed Description
The following embodiments of the present invention are proposed in conjunction with various disciplines such as data mining, neural networks, and flood forecasting.
Data mining is a multidisciplinary cross-domain. On the one hand, to find useful knowledge that is embedded in large datasets in a particular way, data mining must draw nutrients from other disciplines such as statistics, neural networks, information retrieval, high-performance computing, etc. On the other hand, other areas of discipline also require attention to the analysis and understanding of data from different perspectives; data mining also provides new opportunities and challenges for the development of these areas of discipline (zhangongling, yanggong. application of data mining in river flood level prediction [ J ] computer simulation, 2013 (30)). From an information processing perspective, people prefer computers to help analyze data and understand data, helping them make decisions based on rich data. Data mining (finding useful knowledge in extraordinary ways from a large amount of data) is then a natural need.
A neural network is a network that models the activity characteristics of the biological nervous system, a mathematical model of extensive and parallel information processing by computational units (neurons). The method is a common analysis algorithm which is already integrated into various data mining tools, has high fault-tolerant capability and learning capability by adjusting the mutual connection relationship among a large number of internal nodes according to the complexity of a system and the high-speed computing capability of a computer, and can fully approximate a complex function or nonlinear relationship on the premise of giving enough hidden units and enough training samples so as to achieve the purpose of processing information. In theory, neural networks can easily solve problems with hundreds of parameters, providing a relatively simple and efficient method for solving highly complex problems.
The traditional models of flood forecasting reflect some rules of hydrology, but because human has limited knowledge on hydrological and meteorological rules of a watershed, the laws of the nature are complex and changeable, the activities of human continuously influence the underlying surface of the watershed, and the establishment of various models cannot get rid of various assumptions on the simulation generalization of real hydrological phenomena, so that the better models cannot reflect objective rules comprehensively.
The method for forecasting the flood in real time based on the continuous multi-prophase data recursion of the neural network is characterized in that on the basis of an existing data warehouse taking a research object as a theme, the regional average rainfall calculated by a Thiessen polygon method is added, the original data warehouse is adjusted and expanded according to a flood forecasting synthetic flow method, the neural network analysis modeling is innovatively carried out by adopting the continuous multi-prophase rainfall conditions and the hydraulic engineering scheduling data in the data warehouse, and the flood peak flow (water level) of the research object in the forecasting period is obtained by recursion in real time. The flow chart is shown in fig. 1.
The present invention will be described in further detail below by taking the water level forecast of the huzhou station in the hujiang river basin as an example.
In recent years, the production and construction of the Hejiang river basin have changed, and the flood forecasting results of the Hujiang river basin in the nearly three years of Hejiang river stations are not accurate enough by the hydrology department, so that the inventor tries to apply the invention to the Hejiang river stations, and the detailed implementation mode is described in detail below.
Referring to fig. 2 (this figure is a topographic map), after a barrage in a congratulatory river basin section is removed (removed in 2016 in 4 months) in 2017 based on a neural network, event sample data which obviously influences the congratulatory river station water level is researched, so that a flood forecasting scheme with the forecasting precision of class iii is obtained, which is disclosed in document 1 (university of Guangxi: congratsu study and application of congratulatory river basin station flood forecasting based on the neural network, author Susheng, 2017 in 12 months).
On the basis of the thesis, the embodiment makes technical improvement, improves the forecasting precision, breaks through the limitation that the neural network in the document 1 can only analyze one flood process each time on the basis of fully considering the objective law of flood development, is continuously perfect, increases the evaluation on the forecasting timeliness, and provides and completes the technical scheme and the application of the embodiment. The content specifically includes the following:
1. construction of Thiessen polygons based on Arcgis
At present, a plurality of methods for calculating the average rainfall of the area are provided, and an arithmetic mean method, a numerical method, an isoline method, a Thiessen polygon method and the like are commonly used. Among the methods, the Thiessen polygon method is most suitable for the condition that rainfall stations or rainfall distribution in a drainage basin or area is not uniform, and the accuracy of average rainfall calculation can be greatly improved. And the upstream rain collection area of the Hezhou station is formed by combining mountains, small basins and hills, the terrain is complex, and the average rainfall in the area is most suitable to be calculated by using a Thiessen polygon method.
Thiessen polygon is proposed by netherlands meteorologist a.h. thiessen, a method of calculating regional average rainfall by the rainfall of discrete rainfall points. Namely, adjacent discrete rainfall points are connected into a triangle, and the vertical lines of the sides of the triangle are drawn, so that a polygon (namely a Thiessen polygon) is surrounded by the vertical lines around each discrete point, and the rainfall observed by the unique rainfall in the Thiessen polygon is used for representing the rainfall in the range of the polygon. According to the method, a Thiessen polygon network is formed in the whole watershed or a certain area, each Thiessen polygon in the network corresponds to a rainfall observation point, and the rainfall of the rainfall observation point represents the rainfall in the corresponding Thiessen polygon. Therefore, the average rainfall in the area can be calculated through the rainfall of all the rainfall observation points in the network.
Let the number of Thiessen polygons be n, and the rainfall at each rainfall observation point be xiThe ratio of each corresponding Thiessen polygon to the area of the region, i.e., the rainfall weight coefficient, is AiAverage rainfall of the area
Can be obtained according to the following formula:
in the embodiment, firstly, a congratulation river basin congratulation and river station rain collecting area (fig. 2a) required by research is sketched on a topographic map, and meanwhile, the position of a rainfall station is marked; then derive the region map (fig. 2 b); and finally, according to the principle and the characteristics of the Thiessen polygon, constructing the Thessen polygon on the regional map by taking Arcgis10.5 as a platform to obtain the Thiessen polygon net of the Hezhou station rain gathering region (from the tortoise stone reservoir dam to the Hezhou station) of the Hejiang river basin, which is required by research, as shown in figure 2 c.
At the same time, 9 rainfall weight coefficients A are calculated for the 9 rainfall stations in FIG. 2ci(ratio of each corresponding Thiessen polygon to area of the region) as in Table 1. The rainfall stations mentioned below will be indicated using the corresponding serial numbers in table 1.
TABLE 1 Table of regional rainfall weight coefficients calculated by Thiessen polygon method
2. Adjusting and expanding original data warehouse
The original data warehouse establishment mainly comprises dimension modeling and data preparation, a star mode is adopted according to main factors (see figure 3) influencing the Hezhou station water level (the mode is the prior art, and specifically, the Korean family bright data mining, concept and technology (original book 3 rd edition) [ M ]. Beijing: mechanical industry publisher, 2012.)
The method comprises the steps of carrying out dimensional modeling according to a dimensional modeling method in document 1, establishing a data warehouse by taking Microsoft excel as a platform according to a simple and easy principle, unifying a data structure, and completing the establishment of the data warehouse through the acquisition, cleaning, checking and integration of data. Compared with the document 1, the data warehouse is adjusted and expanded. The original data warehouse condition and the adjustment expansion are explained as follows:
1. the method is influenced by construction of a riverbank park in a congratulatory river city section in 2015 and dismantling of an upstream barrage (eight-step barrage) of a smart peak bridge in an urban area at the bottom of 3 months in 2016, a congratulatory river station is positioned about 600 meters upstream of the barrage, and underlying surfaces (factors such as riverbeds) of the congratulatory river city section have obvious changes, so that the data warehouse only records the water and rain condition and hydraulic engineering scheduling data which influence the water level of the congraty river station in 2016 in the original data warehouse.
2. Increasing the average rainfall in the area is the main factor affecting the water level in the congratulation station, as shown in fig. 3.
3. The flood discharge flow of the tortoise stone reservoir is adjusted to be the flow out of the reservoir, the flow out of the reservoir is the flow of discharging the flood, the flow of generating electricity and the flow of irrigation, the details of the data are further optimized, and the reliability and the accuracy of analysis are improved.
4. Since 2018, no flood occurs in the Hezhou station until the beginning of 10 months, the highest peak value is only 101.7m, the difference between the highest peak value and the alarm ring water level is 103.5m and 1.8m, the difference between the highest peak value and the ordinary water level is 100.8 m, the research significance is low, and therefore 2018 related data are not added into a data warehouse.
5. The unit meters of all water level data in the original data warehouse are adjusted to be centimeters, so that the magnitude relation between the unit meters of all water level data and rainfall data is reduced, and the accuracy is improved.
3. Establishment of flood forecasting scheme based on neural network
3.1 general description
According to the hujiang drainage basin huzhou station flood forecasting scheme researched and established in the embodiment, according to the basic situation of the upstream of the huzhou station in the hujiang drainage basin and the construction situation of hydraulic engineering facilities, the data source for flood forecasting by the hydrological department is referred, and relevant information such as rainfall, water level, and the output flow of the tortoise-stone reservoir required by the huzhou station flood forecasting is collected, wherein the rainfall and the water level are from the city hydrological office and are already completed in the document 1, the output flow of the tortoise-stone reservoir is from the tortoise-stone reservoir, and the current application completes the work, and the huzhou station flood data cannot be collected timely due to the time relation, so that the flood peak level is only forecasted. However, as will be apparent to those skilled in the art, the method proposed in the present application is equally applicable to peak flow prediction, provided that sufficient data is available.
In the embodiment, a plurality of flood forecasting schemes are planned and designed, and the most accurate one of the flood forecasting schemes is selected as the Huzhou station flood forecasting scheme in the Hejiang river basin through a plurality of improvements of neural network analysis. On the premise of ensuring the flood forecasting precision and reliability, the method evaluates and tests the Huzhou station flood forecasting scheme of the Hejiang river basin according to the relevant flood forecasting precision evaluation requirements.
3.2 Multi-layer feedforward neural network
The multi-layer feedforward neural network iteratively learns a set of weights for sample data prediction, which is a classification of the neural network including an input layer, one or more hidden layers and an output layer, as shown in fig. 4, and in general, the number of neurons of a hidden layer is proportional to a factor of the input. The multi-layer feedforward neural network is hereinafter referred to as a neural network.
Two important processes and functions of neural networks are learning and execution. The neural network can record the learning information of a plurality of times on the link weight of each node after learning an information mode for a plurality of times, so that the neural network is particularly sensitive to the information mode, and can make a more accurate inference conclusion or a predicted value when similar information which is learned by the neural network is executed again.
The working principle of the neural network is as follows: the independent variable (factor) is continuously transmitted from the input layer to the hidden layer and then to the output layer, the neuron of the hidden layer is equivalent to an adopted statistical analysis model, the processing mode of the received corresponding signal is continuously adjusted, and finally the dependent variable output by the output layer becomes a more correct inference conclusion or a predicted value.
In the neural network analysis, in order to prevent data overfitting, sample data is reasonably split, a training set and a verification set are formed according to a certain proportion, and the common proportion is 7:3 or 5: 5.
3.3 design flood forecasting scheme
3.3.1 design description
Observed from the data warehouse and found that the Hezhou station counts 3 super-alert floods since 2016, in the 3 flood events, the flow of the upstream tortoise stone reservoir is up to 450 cubic meters per second, the flood discharge time is long, the Hezhou station upstream of the Hejiang basin has strong rainfall influence to the large dam section of tortoise stone, the sample data is most abundant, and the sample data has high analytical value and representativeness and is incomparable to other events. The adoption of the above 3 event sample compiling and inspection forecasting schemes can improve the reliability of the schemes to the maximum extent, so the following flood forecasting schemes will mainly be performed around the above 3 events, the event pairs such as table 2.
Table 2 event comparison table
In combination with actual conditions, in the embodiment, the forecasting scheme is compiled and evaluated by using sample data of 2016 super-alert flood events (the first two events, namely 2016.5.20 and 2016.6.13, in table 2), and the design scheme is subjected to precision inspection by using 2017 super-alert flood event samples (the third event, namely 2017.7.2, in table 2) which do not participate in the forecasting scheme compilation.
3.3.2 forecasting method
Scheme design: on the basis of a data warehouse established in document 1 and taking a research object as a theme, adding the area average rainfall calculated by the Thiessen polygon method, adjusting and expanding the original data warehouse according to the idea of a flood forecasting synthetic flow method and combining with innovation points, innovating, using the continuous rainfall and hydraulic engineering scheduling data of a plurality of early periods in the data warehouse as a data analysis unit to perform neural network analysis modeling, and recurrently obtaining the flood peak flow (water level) of the research object in the forecast period in real time.
The implementation process comprises the following steps:
1. the row data in the augmented data warehouse is adjusted (i.e., the data units that are to be analyzed by the neural network are determined).
According to the idea of a synthetic flow method, a certain propagation time (shown in table 3) exists from the turtle stone reservoir to the congratulation station, the rainfall of each rainfall station and the flow (water level) of the Zhongshan station, the flow of each factor reaching the congratulation station at the same time is superposed according to the propagation time to carry out flow synthesis, namely, unit row data are subjected to row-column displacement transformation according to the propagation time in a data warehouse, and corresponding area average rainfall is calculated for the transformed row rainfall data (the calculation result is used as a new factor to participate in neural network analysis), so that each row of data embodies the actual influence on the water level of the congratulation station. As in fig. 5A.
The embodiment innovatively adopts all the hour-by-hour data of the main factors influencing the Hezhou station water level in the first 6 hours after the traffic is synthesized to forecast the Hezhou station water level in the next hour, that is, the number of data units is adjusted from 15 researched in document 1 to 79 (considering that the self-variable serial number itself cannot completely reflect the influence on the water level of the shizhou station, and the influence on the water level of the shizhou station is eliminated, and the continuous rainfall and hydraulic engineering scheduling data of a plurality of early periods are used as a data analysis unit to perform neural network analysis modeling, so that the neural network analysis can fully mine the rules implied by the flood process, the possibility that unreasonable fitting occurs when document 1 uses the serial number to represent the flood process to perform analysis is avoided, the number of flood process fields which can be analyzed each time is not limited, and the number of the flood process fields is 6+1 forecast object period data (see the synthetic flow diagram of fig. 5) is 79.
TABLE 3 propagation time table
| Site | Propagation time (hours) |
| Tortoise-rock |
8 |
| Clock station flow (water level) | 6 |
| Water |
1 |
| White |
8 |
| Hezhou rainfall station | 0 |
| Bell mountain |
6 |
| Rain amount station for |
8 |
| Fish |
7 |
| Water gap |
4 |
| High- |
7 |
| Dam |
6 |
2. The number of columns in the augmented data warehouse is adjusted (i.e., the number of data units to be analyzed by the neural network is determined).
The method adopts the continuous rainfall condition and hydraulic engineering scheduling data of a plurality of early periods as a data analysis unit to carry out neural network analysis modeling, so that the flood process field which can be analyzed each time is not limited, and the data units are integrated into the same table to sum up 381 columns, namely 381 period data.
Meanwhile, the sample data of the over-warning flood event in 2017 is adjusted and expanded by the method, and preparation is made for precision inspection.
3. Planning a forecast period
Under the condition that the data unit is complete, reasonable and effective, the neural network model can make more accurate prediction. As can be seen from fig. 5, the forecast water level at n +1 is made to the congratulatory state station at n, and the congratulatory state rainfall station data at n +1 is unknown but can be obtained through weather forecast, so that forecast with forecast period of 1 hour is completed. It can be seen from the observation of table 3 that, among 11 stations, the propagation time of 8 stations is more than 6 hours, and the rest 3 stations are rainfall data, and the rainfall can be obtained through weather forecast, so that the data units in the future of 6 hours can be complete, reasonable and effective, and further the congratulation station water level at n +6 hours can be gradually and reasonably promoted, that is, the forecast period is planned to be 6 hours.
4. Neural network analysis
Based on a data mining tool JMP10 platform, sample data of 2016 year total over-warning flood field times (2 field times in total) adjusted and expanded through the steps are selected from a data warehouse and loaded into a JMP for neural network modeling analysis, and output dependent variables (responses) are as follows: congratulation station level at n +1, input argument (factor): the method comprises the following steps of setting a neural network training set and a verification set according to the ratio of 2:1, setting the number of hidden nodes, namely the number of neurons in a hidden layer according to 15, wherein the n-time Hezhou station water level, (n + 1-propagation time) clock mountain station water level, (n + 1-propagation time) tortoise stone ex-warehouse flow, (n + 1-propagation time) data of all rainfall stations, corresponding area average rainfall and the like. And (6) modeling. And (3) carrying out preliminary precision inspection on the model obtained through the neural network analysis by directly using 2017-year over-warning flood event sample data which does not participate in the compilation of the flood forecasting scheme, and finally carrying out complete and standard precision assessment and inspection on the scheme if the inspection result is accurate, and repeatedly carrying out modeling, inspection and error correction to gradually obtain a relatively satisfactory flood forecasting scheme.
5. Determining a forecast scenario
According to the test, in order to ensure the forecasting precision, the forecasting period is reduced from 6 hours to 5 hours. The final prediction scheme is shown in fig. 5B.
The forecasting formula is as follows:
in the formula, h is the forecast water level of the Hezhou station in the next hour, and the Hezhou station water level in the next 5 hours can be recurred according to the designed forecast scheme; i represents hidden layer neurons of the neural network, 15 in total; k represents the weight coefficient of the neuron, R is a constant value and is obtained by a neural network by utilizing a digging tool jmp; h represents the fitting value of the neuron, is a hyperbolic tangent tanh function value, is calculated by 78 factors through a neural network by using a digging tool jmp, and has the following formula,
in the formula, T is a determination ratio before the tanh function is calculated, L is a limiting coefficient, and P is a weight coefficient of each factor corresponding to a neuron, which are obtained by a neural network by using a mining tool jmp; z is factor, m is factor number. The correspondence is shown in Table 3A below.
TABLE 3A
3.4 precision assessment
According to the Standard of the hydrologic information forecasting Specification (GB/T22482-.
3.4.1 precision assessment and verification of forecasting protocols
The certainty coefficient (DC) represents the coincidence degree between the flood forecasting process and the actual measurement process, and the accuracy of the flood forecasting scheme is divided into three grades of A, B and C according to the conditions that DC is more than 0.9, DC is more than or equal to 0.9 and more than or equal to 0.7, and DC is more than 0.7 and more than or equal to 0.5. In this embodiment, the accuracy of the flood forecasting process of the scheme is evaluated and checked by using the deterministic coefficient, and the deterministic coefficient calculation formula is as follows:
in the formula: DC-deterministic coefficient (taking 2 decimal places);
yc(i) -a predicted value;
y0(i) -an actual measurement value;
n is the length of the data sequence.
Selecting all data participating in flood forecasting to evaluate according to the requirements of hydrologic information forecasting standards, selecting all data not participating in flood forecasting to test, calculating forecasting results every 5 hours, wherein the evaluation results are shown in figures 6 and 7, the X axis is a time sequence, the Y axis is a water level, the unit is cm, the DC in figure 6 is 0.98, the DC in figure 7 is 0.98, and the smooth curve is an actually measured water level.
The test result of the forecasting scheme of this embodiment is as shown in fig. 8 (all data not participating in flood forecasting are selected for evaluation), where DC is 0.82, and the rounder one in the curve is the measured water level; the prediction scheme of reference 1 is shown in fig. 9, where DC is 0.77, and the rounder curve is the measured water level. In this embodiment, the height is 0.05 higher than that of document 1, and the flood peak water level and the flood peak occurrence time are predicted more accurately.
In conclusion, the DC values of the two flood processes participating in flood forecasting in the embodiment are both 0.98 according to the deterministic coefficient formula, and the evaluation precision reaches the first-level; the DC value of the flood process which does not participate in flood forecasting is 0.82, and the inspection precision is a level B level.
3.4.2 flood peak forecast aging evaluation
And (3) representing the effectiveness timeliness coefficient of flood peak forecasting time, and calculating according to the following formula: CET ═ EPF/TPF
In the formula:
CET-aging coefficient (2 decimal places);
EPF-effective forecast period (which refers to the time interval from the release forecast time to the occurrence of the flood peak (or forecast object) at the station, and takes 1 decimal fraction, and the unit is hour (h);
TPF-theoretical forecast period (1 decimal) is taken as the time interval from the main rainfall stopping or forecasting according to the occurrence of elements to the occurrence of the flood peak (or forecasting object) of the station, and the unit is hour (h).
The single river reach flood peak forecast aging grade is determined according to the table 4, when the CET is more than 1.00, the forecast is advanced, the flood peak forecast is issued when the flood peak forecast basis element does not appear, and the forecast aging does not reach the third grade, and the aging is unqualified. The forecasting time efficiency in the water level flow process can also be determined by comparing the forecasting value with the longest forecasting period with the regulation of the flood peak forecasting time efficiency grade.
TABLE 4 single river reach flood peak forecast aging grade table
| Age rating | First (Rapid) | Second (in time) | Third (qualified) |
| Coefficient of aging property | CET≥0.95 | 0.95>CET≥0.85 | 0.85>CET≥0.70 |
Meanwhile, the CET of each aging grade is calculated by taking the following values of the operation time consumption value dh (including the water regime information receiving and processing time, dh being TPF-EPF) as upper limits, namely the grade a is less than or equal to 0.6h, the grade b is less than or equal to 0.8h, and the grade c is less than or equal to 1.0 h.
According to the provisions, the main content of the forecasting operation of the forecasting scheme is that forecasting foundation elements (each water level, rainfall and tortoise stone reservoir delivery flow data) at n hours are substituted into a forecasting formula to further recur to obtain the Hezhou station water level at n +5 hours, calculation can be carried out through Excel, small program compiling and data mining tools, and the forecasting value can be obtained by inputting data required by the forecasting formula; the main operation time consumption comprises water regime information receiving processing and forecast data analysis processing of three rainfall sites, the whole operation process time consumption dh can be controlled below 0.2h (namely 12 minutes), if a small program can be written and embedded into a mountain torrent disaster monitoring and early warning platform and forecast data analysis processing work of the three rainfall sites is done in advance, the main operation time consumption is only the water regime information receiving processing time consumption.
In conclusion, the TPF of the prediction scheme is 5h, the operation time dh is 0.2h, the CET is calculated to be 0.96, and the aging grade is class a.
3.4.3 Total analysis
Firstly, the accuracy test value of the forecasting scheme of the embodiment is improved compared with that of the document 1, the peak water level and the peak time are more accurate, and particularly, the peak water level and the peak time are accurate when the first peak is continuously forecasted, as shown in table 5.
Table 5 continuous real-time flood peak forecasting using the forecasting method of this embodiment
The method adopts the continuous rainfall conditions and the hydraulic engineering scheduling data in a plurality of early periods after conversion according to the idea of a flood forecasting synthetic flow method as a data analysis unit to carry out neural network analysis modeling, so that the development process of the flood is represented comprehensively, and the rules implied in the flood process can be fully mined by the neural network analysis; the possibility that unreasonable fitting occurs when the sequence number is used for representing the flood development process in the document 1 for analysis is avoided; the number of flood processes that can be analyzed at a time is not limited.
Thirdly, the forecast timeliness of the forecast scheme is evaluated in the embodiment, and the forecast timeliness of the first grade fully shows the high efficiency of the forecast scheme.
The forecast period of the flood forecasting scheme is 5 hours, the forecasting precision is undetermined grade B (only the flood data in 2017 are adopted for precision inspection, the number of years is not less than 2 according to regulations, the flood station does not flood by 10 months in 2018, the highest peak value process is not representational), and the forecasting timeliness is grade A, so that the flood forecasting can be carried out in real time and used for reference forecasting.
The research of the embodiment is not mature enough based on the literature 1, only uses the water and rain condition and the hydraulic engineering scheduling data in a single time period as a data analysis unit to analyze and model the neural network, uses sequence numbers to represent the flood development process to analyze and model the neural network, can only analyze the flood process data of one field at a time, innovatively adopts all chronological data which influence the main factors of the water level of the congratulation state station in the first 6 hours to forecast the water level of the congratulation state station in the next hour, and further recurs to obtain the water level of the congratulation state station in the next 5 hours (the data in the first 6 hours is input, the forecast water level after 5 hours is output, and the data is recorded once in each hour), breaks through the limitation that the neural network of the literature 1 can only analyze the flood process of one field at a time, adopts the existing flood peak process data of all the water fields to compile a forecast scheme, the forecasting precision is improved, and the forecasting of the flood peak water level and the peak present time is more accurate.
The water level prediction is taken as an example in the present embodiment, but the method can also be used for flow rate prediction.
The flood forecast of the embodiment is used for measuring and tailoring a forecast object, is driven by a service, is based on a data warehouse, can explore deep layer rules implicit in a flood process from a large amount of water conditions, rain conditions and engineering conditions data influencing a flood peak (water level) of the forecast object, simultaneously adds the area average rain quantity calculated by a Thiessen polygon method, adjusts and expands an original data warehouse according to a flood forecast synthetic flow method, innovatively adopts the water conditions and hydraulic engineering scheduling data of a plurality of continuous early periods of a plurality of flood fields transformed according to the idea of the flood forecast synthetic flow method as a data analysis unit to perform neural network analysis modeling, and obtains a flood forecast model with higher precision; by adopting a recursion algorithm and combining with the actual situation of the forecast object, the flood forecast period is prolonged to a certain extent; and (3) extracting a forecasting formula according to the obtained flood forecasting model, and inputting data required by the forecasting formula to obtain a real-time forecasting value, so that the forecasting operation timeliness is greatly improved. The method of the embodiment has the advantages of high forecasting precision, long forecasting period, high forecasting operation timeliness, real-time forecasting, low application implementation difficulty and short period, and can provide more powerful support for flood prevention decision and emergency rescue scheduling. In addition, the method can also be applied to reservoir hydropower station dispatching, so that certain economic benefit can be generated
The method of the embodiment can be applied to station and reservoir hydropower station forecasting with comprehensive hydrological data (water level stations, rainfall stations and hydraulic engineering dispatching data which have main influence on forecasting objects), can be programmed into a software module to be embedded into a mountain flood disaster monitoring and early warning platform, and provides more powerful support for flood prevention decision and emergency rescue dispatching.
Claims (10)
1. A flood forecasting method and system based on a neural network are characterized by comprising the following steps:
s1, adding the area average rainfall calculated by the Thiessen polygon method on the basis of a data warehouse taking the research object as a theme;
s2, adjusting and expanding the data warehouse;
and S3, performing neural network analysis modeling by adopting the rainfall and water conservancy project scheduling data of a plurality of continuous early periods of a plurality of flood fields and transformed by a flood forecast synthetic flow method in the data warehouse, and performing real-time recursion to obtain at least one of the flood peak flow and the water level of the forecast study object.
2. The flood forecasting method and system based on neural network as claimed in claim 1, wherein in step S1, a thiessen polygon network is formed in the whole flow field or a certain area according to the thiessen polygon method, each thiessen polygon in the network corresponds to a rainfall observation point, and the rainfall of the rainfall observation point represents the rainfall in the corresponding thiessen polygon; therefore, the regional average rainfall is calculated through the rainfall of all the rainfall observation points in the network.
3. The method and system for neural network based flood forecasting according to claim 2, wherein the method for forming a Thiessen polygonal mesh is: firstly, drawing a rain collecting area of a certain rainfall station of a drainage basin required by research on a topographic map, and marking the position of the rainfall station; then deriving a region map; and finally, according to the principle and the characteristics of the Thiessen polygon, constructing the Thiessen polygon on the regional map to obtain the Thiessen polygon network of the rain collecting region required by research.
4. The method and system for neural network-based flood forecasting according to claim 3, wherein the regional average rainfall is
Can be obtained according to the following formula:
wherein n is the number of Thiessen polygons, xiFor the rainfall of each rainfall observation point, a is the ratio of each corresponding Thiessen polygon to the area, i.e. the rainfall weight coefficient.
5. The method and system for neural network based flood forecasting according to claim 1, wherein the adjusting and expanding the data warehouse in step S2 includes: increasing the area average rainfall as a main factor influencing the water level of the rainfall station to be forecasted; and adjusting the flood discharge flow of the upstream reservoir into the delivery flow, wherein the delivery flow is the flood discharge flow, the power generation flow and the irrigation flow.
6. The flood forecasting method and system based on neural network as claimed in claim 7, wherein in step S3, according to the synthetic flow method, the flow synthesis is performed on each factor according to the propagation time, that is, the row-column displacement transformation is performed on the event sample data corresponding to the propagation time, and the corresponding area average rainfall is calculated for the rainfall data in the transformed data units, the calculation result is used as a new factor to participate in the neural network analysis, so that the data units to be subjected to the neural network analysis exhibit the actual influence on the water level of the rainfall station to be forecasted, and the data of the data units to be subjected to the neural network analysis are added with the data of the consecutive early periods transformed according to the synthetic flow method.
7. The method and system for neural network based flood forecasting according to claim 1, wherein in step S3, the number of flood process sessions that can be modeled for each neural network analysis is not limited.
8. The method and system for flood forecasting based on neural network as claimed in claim 7, wherein the forecast water level at n +1 is made for the station to be forecasted at n, the station rainfall data at n +1 is obtained by weather forecast, and the forecast with forecast period of 1 hour is completed; recording the minimum value of the propagation time of the upstream station as m, adjusting and expanding a data unit for m hours in the future, namely data of continuous multiple early-stage time intervals of factors influencing a forecast object, which are correspondingly transformed according to a synthetic flow method, can be complete, reasonable and effective, and further the water level and the flow of the station to be forecasted when the station reaches n + m can be gradually and reasonably extrapolated, namely the forecast period is drawn to m hours; wherein m and n are positive integers.
9. The flood forecasting method and system based on the neural network as claimed in claim 7, wherein the water and rain conditions and the water conservancy project scheduling data of a plurality of continuous early periods of a plurality of flood fields transformed according to the flood forecasting composite flow method are used as data analysis units to perform neural network analysis modeling on at least one of the peak flow and the water level of the forecasting object, the model obtained through the neural network analysis is subjected to preliminary precision inspection directly by using the sample data of the super-alert flood event which is not involved in the flood forecasting scheme compilation, if the inspection result is accurate, the scheme is subjected to complete and standard precision assessment and inspection, modeling, inspection and error correction are repeated, the forecasting period can be properly adjusted according to the precision inspection result, and finally a relatively satisfactory flood forecasting scheme is obtained.
10. A neural network based flood forecasting system for flood forecasting using the method of any one of claims 1 to 10.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2018113755421 | 2018-11-19 | ||
| CN201811375542 | 2018-11-19 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111199298A true CN111199298A (en) | 2020-05-26 |
| CN111199298B CN111199298B (en) | 2023-06-27 |
Family
ID=70746981
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811454308.8A Active CN111199298B (en) | 2018-11-19 | 2018-11-30 | Flood forecasting method and system based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111199298B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111738600A (en) * | 2020-06-23 | 2020-10-02 | 南通大学 | Urban road air quality evaluation method based on high-precision PM2.5 inversion result |
| CN112417554A (en) * | 2020-11-17 | 2021-02-26 | 中国水利水电科学研究院 | Method for identifying local flood control influence risk factors by adopting fishbone map to carry out cross-basin water transfer engineering |
| CN112967476A (en) * | 2021-04-09 | 2021-06-15 | 黑龙江文图测绘地理信息有限责任公司 | Flood forecasting system based on hyperbolic flow distribution model |
| CN113642699A (en) * | 2021-06-30 | 2021-11-12 | 天津大学 | Intelligent river flood forecasting system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1438386A (en) * | 2002-01-31 | 2003-08-27 | 株式会社东芝 | River-level prediction device |
| CN102360468A (en) * | 2011-10-13 | 2012-02-22 | 贵州东方世纪科技有限责任公司 | Flood forecasting system |
| US20140304007A1 (en) * | 2013-04-05 | 2014-10-09 | The Travelers Indemnity Company | Systems and methods for account-level flood risk assessment |
| CN104615845A (en) * | 2014-12-19 | 2015-05-13 | 兰州大学 | Supercritical flood risk analysis method |
| CN105115573A (en) * | 2015-07-18 | 2015-12-02 | 厦门理工学院 | Correction method and device for flood flow forecasting |
| US20160273181A1 (en) * | 2015-03-20 | 2016-09-22 | Alexis Hannah Smith | Flood Prevention System and Method |
| CN106326656A (en) * | 2016-08-24 | 2017-01-11 | 东南大学 | Simulation and prediction method for extreme rainstorm flood level of engineering facility |
| CN106650767A (en) * | 2016-09-20 | 2017-05-10 | 河海大学 | Flood forecasting method based on cluster analysis and real time correction |
| CN108446436A (en) * | 2018-02-08 | 2018-08-24 | 广州地理研究所 | The spatial distribution method for early warning of storm flood nonlinear model rainwater loss parameter |
-
2018
- 2018-11-30 CN CN201811454308.8A patent/CN111199298B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1438386A (en) * | 2002-01-31 | 2003-08-27 | 株式会社东芝 | River-level prediction device |
| JP2003293344A (en) * | 2002-01-31 | 2003-10-15 | Toshiba Corp | River water level predicting apparatus |
| CN102360468A (en) * | 2011-10-13 | 2012-02-22 | 贵州东方世纪科技有限责任公司 | Flood forecasting system |
| US20140304007A1 (en) * | 2013-04-05 | 2014-10-09 | The Travelers Indemnity Company | Systems and methods for account-level flood risk assessment |
| CN104615845A (en) * | 2014-12-19 | 2015-05-13 | 兰州大学 | Supercritical flood risk analysis method |
| US20160273181A1 (en) * | 2015-03-20 | 2016-09-22 | Alexis Hannah Smith | Flood Prevention System and Method |
| CN105115573A (en) * | 2015-07-18 | 2015-12-02 | 厦门理工学院 | Correction method and device for flood flow forecasting |
| CN106326656A (en) * | 2016-08-24 | 2017-01-11 | 东南大学 | Simulation and prediction method for extreme rainstorm flood level of engineering facility |
| CN106650767A (en) * | 2016-09-20 | 2017-05-10 | 河海大学 | Flood forecasting method based on cluster analysis and real time correction |
| CN108446436A (en) * | 2018-02-08 | 2018-08-24 | 广州地理研究所 | The spatial distribution method for early warning of storm flood nonlinear model rainwater loss parameter |
Non-Patent Citations (4)
| Title |
|---|
| 洪小康: "渭河下游洪水预报的人工神经网络模型研究", vol. 1, no. 004, pages 93 - 96 * |
| 王一峰;王心亚;: "基于人工神经网络的倒水河洪水预报模型", 水资源研究, no. 02, pages 42 - 50 * |
| 苏盛: "基于神经网络的贺江流域贺州站洪水预报研究与应用", pages 24 - 43 * |
| 高鹏;刘正东;: "天然河道水位流量预测模型分析", 陕西水利, no. 1, pages 221 - 230 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111738600A (en) * | 2020-06-23 | 2020-10-02 | 南通大学 | Urban road air quality evaluation method based on high-precision PM2.5 inversion result |
| CN112417554A (en) * | 2020-11-17 | 2021-02-26 | 中国水利水电科学研究院 | Method for identifying local flood control influence risk factors by adopting fishbone map to carry out cross-basin water transfer engineering |
| CN112967476A (en) * | 2021-04-09 | 2021-06-15 | 黑龙江文图测绘地理信息有限责任公司 | Flood forecasting system based on hyperbolic flow distribution model |
| CN113642699A (en) * | 2021-06-30 | 2021-11-12 | 天津大学 | Intelligent river flood forecasting system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111199298B (en) | 2023-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111199298A (en) | Flood forecasting method and system based on neural network | |
| CN107463730B (en) | A kind of streamflow change attribution recognition methods considering Spatio-temporal Evolution of Land Use | |
| CN114443982A (en) | Large-area soil heavy metal detection and space-time distribution characteristic analysis method and system | |
| CN105678481A (en) | Pipeline health state assessment method based on random forest model | |
| CN104715292A (en) | City short-term water consumption prediction method based on least square support vector machine model | |
| CN101853290A (en) | Meteorological service performance evaluation method based on geographical information system (GIS) | |
| CN109408848A (en) | A kind of distributed attribution method considering Runoff Evolution temporal-spatial heterogeneity | |
| Fang et al. | Application of long short-term memory (LSTM) on the prediction of rainfall-runoff in karst area | |
| CN102184423B (en) | Full-automatic method for precisely extracting regional impervious surface remote sensing information | |
| CN114330812A (en) | Landslide disaster risk assessment method based on machine learning | |
| CN112182234A (en) | Drainage basin flood control planning data knowledge graph construction method | |
| CN116485584B (en) | Method and system for cooperative regulation and control of WEE of river with large bottom slope in alpine region | |
| CN104636834A (en) | Improved optimization method for joint probability programming model system | |
| Gao et al. | A multifactorial framework for short-term load forecasting system as well as the jinan’s case study | |
| CN117236199B (en) | Method and system for improving water quality and guaranteeing water safety of river and lake in urban water network area | |
| CN113902580B (en) | Historical farmland distribution reconstruction method based on random forest model | |
| An et al. | Identify and assess hydropower project’s multidimensional social impacts with rough set and projection pursuit model | |
| CN116611785A (en) | Power transmission and transformation project cost model construction method, system, equipment and medium based on big data | |
| Zhai et al. | Using spatial heterogeneity to strengthen the neighbourhood effects of urban growth simulation models | |
| CN110852496A (en) | Natural gas load prediction method based on LSTM recurrent neural network | |
| Yang et al. | Evaluation and Interpretation of Runoff Forecasting Models Based on Hybrid Deep Neural Networks | |
| CN116796799A (en) | Method for creating small-river basin flood rainfall threshold model in area without hydrologic data | |
| CN113673145A (en) | Surface subsidence prediction method based on PCA-LSTM neural network | |
| CN105676313B (en) | For bridge spanning the sea construction marine site wave field, the regional prediction method of wind speed field | |
| CN111797577A (en) | Method and system for evaluating adaptability of typical remediation engineering of estuary and river network |
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 |