57 results about "Flood forecast" patented technology

The invention discloses a hydrological forecasting method based on multi-feature combination and a Transform model. The method comprises the following steps: acquiring measured data of a target watershed hydrological station, building a physical hydrological model, acquiring output data and derivative feature data of the physical hydrological model, and integrating to obtain basic hydrological data; creating enhanced hydrological physical features, and forming a multi-dimensional original feature pool; constructing a plurality of combination strategies based on the basic hydrological data and the multi-dimensional original feature pool; capturing a long-term dependency relationship of the hydrological time sequence by using an improved Transform model; and the model performance is improved through automatic hyper-parameter optimization. According to the method, the influence of different input feature combinations on the flood forecasting precision is highlighted, and effective technical support is provided for water resource management and flood control and disaster reduction.

The invention relates to the technical field of flood forecasting, in particular to a flood forecasting error real-time correction method and system, and the method comprises the steps: obtaining data under a unified time reference, calculating a water purification head, and obtaining the instantaneous flow; inquiring the theoretical efficiency, calculating the actual measurement efficiency, and carrying out difference calculation to obtain an electricity-water efficiency residual error; evaluating the electricity-water efficiency residual error to obtain a stability index; judging an operation phase, and calculating the confidence coefficient of the operation phase according to the stability index; generating a correction vector and correcting the uncorrected process line to obtain a corrected process line; projecting the correction process line according to a preset engineering feasible region to obtain first correction output; performing online checking and incremental correction on the first correction output to obtain final correction output; and updating the model initial state and the boundary condition of the next time window according to the final correction output. According to the method, the problem of error accumulation caused by phase misjudgment under the hysteresis effect and field disturbance of a traditional method is solved, and the real-time accuracy of flood forecasting is remarkably improved.

The invention relates to the technical field of urban hydrological simulation and flood forecasting, and particularly discloses a comprehensive unit line improvement method applicable to an urbanized region, which comprises the following steps: S1, constructing an urban rainstorm runoff model, and simulating the flow process of a drainage basin outlet section under rainstorm in different recurrence periods; s2, calculating the flow process of the drainage basin outlet section under the same design rainstorm, and reconstructing and converting the flow process into a centralized flow process; s3, distributing the centralized flow process into sub-flows of a water outlet and a river channel section; s4, on the basis of the actual maximum drainage capacity of the drainage port, storage and drainage response correction is carried out on the sub-flow process of the drainage port, and a time sequence flow process considering drainage capacity constraint is formed; s5, calculating the segmented delay of the propagation time lag to the corrected flow, and obtaining an improved total flow process; the improved urban rainstorm basin outlet flow simulation is realized by performing space-time reconstruction, space distribution, drainage capacity constraint correction and propagation time delay delay on the flow process.

The invention discloses an intelligent flood forecasting method for coupling error correction and joint modeling, and relates to a deep learning and uncertainty modeling technology. At the input end, constructing a future random rainfall scene through hourly dynamic normal disturbance; the method comprises the following steps: at a model end, introducing a multi-structure and multi-objective function combination based on Kolmogorov-Arnold Networks and Transform, and forming a multi-member ensemble forecast; at an error end, a probabilistic modeling method based on a numerable asymmetric Laplacian mixed density network is provided, and fine error correction is realized; a Vine copula function is adopted to construct high-dimensional joint distribution, and a Bayesian model averaging and expectation maximization algorithm is combined to realize weighted fusion of multi-member posterior results; according to the method, uncertainty in flood forecasting can be comprehensively described, the stability and adaptability of a forecasting system are improved, and the method is suitable for a basin-level real-time flood ensemble forecasting scene.

The invention relates to the technical field of hydrological simulation and flood forecasting, and particularly discloses a watershed runoff simulation method and device with a graph neural network fused with hydrological priori knowledge, and the method comprises the steps: carrying out the grid division of a target watershed, and building a watershed graph structure in combination with the hydrological priori knowledge; introducing a trainable weight into the graph neural network model, and performing time delay weighting based on a spatial distance to obtain a fusion network model; the meteorological driving data, the rainfall data and the drainage basin attributes serve as input, the convergence process is simulated through the fusion network model, and the predicted downstream runoff volume is obtained; and comparing the predicted downstream runoff volume with the actually measured downstream runoff volume, and optimizing all learnable parameters of the fusion network model. According to the method, the defects that a traditional physical model is large in calculation amount and long in consumed time are overcome, rapid simulation of the runoff process is achieved, and the accuracy and reliability of a data driving model in a runoff simulation task are improved by fusing hydrological priori knowledge and introducing physical constraints.

A flood similarity intelligent analysis method based on multi-mode Transform and comparative learning comprises the steps that firstly, static features are processed through a grouping full-connection network, and the characterization capacity is enhanced through feature specific transformation; a CNN-Transform-Attention hybrid network is constructed to extract dynamic process line features, a convolutional layer captures local morphological features, a Transform encoder captures a global time sequence dependency relationship through a self-attention mechanism, key hydrological stages are adaptively focused through an attention weight, multi-modal features are adaptively integrated by adopting a gating fusion mechanism to generate unified embedded representation, and the dynamic process line features are extracted by adopting a convolutional neural network (CNN)-Transform-Attention hybrid network. A comparison learning and difficult sample mining strategy is introduced, discriminative characterization is learned in a low-dimensional embedding space through a comparison loss function training model based on the Euclidean distance, and the similarity is mapped into an interpretable probability of a [0, 1] interval; the technical problem that multi-source heterogeneous features are insufficient in utilization is effectively solved, accurate quantification and interpretable evaluation of flood similarity are achieved, and technical support is provided for flood forecasting, historical flood matching and flood control dispatching decision making.

The invention relates to the technical field of flood forecasting, and discloses a flood forecasting error correction method based on system differential response, and the method comprises the following steps: eliminating abnormal interference components of flood forecasting time series data of a target area, and obtaining purified time series data; constructing a differential response association map by taking the factors as nodes and the association strength as edge weights according to the association strength of the influence factors in the purified data; comparing the forecast data with the actual monitoring data to obtain initial error data; according to an atlas response conduction path, splitting the initial error into an error contribution component set according to an influence factor response weight; in combination with differential response characteristics of each factor, the forecast data is corrected in a targeted manner to obtain preliminary correction data; based on a flood routing basic rule, performing consistency verification on the preliminary correction forecast data, and if the preliminary correction forecast data accords with the rule, generating final correction forecast data; according to the invention, the flood forecast error correction efficiency can be improved.

The invention provides a flood forecasting method and device based on an intelligent optimization neural network, and the method comprises the following steps: obtaining original hydrological time series data, and carrying out the variational mode decomposition of the original hydrological time series data based on an optimal penalty factor alpha and an optimal decomposition number K, and obtaining K intrinsic mode functions; dividing the K intrinsic mode functions into high-frequency intrinsic mode functions, intermediate-frequency intrinsic mode functions and low-frequency intrinsic mode functions based on the center frequency of each intrinsic mode function in the flood flow prediction network, and fusing the high-frequency intrinsic mode functions, the intermediate-frequency intrinsic mode functions and the low-frequency intrinsic mode functions to obtain local time sequence features; and fusing the global time sequence features and the local time sequence features to obtain global-local fusion features, and inputting the global-local fusion features into a full connection layer to obtain a predicted flood flow value. According to the scheme, the optimal penalty factor alpha and the optimal decomposition number K are globally and automatically optimized through the optimization algorithm, and low efficiency and deviation caused by manual parameter adjustment are avoided.

The application belongs to the technical field of cross between artificial intelligence and hydrological prediction, and provides a flood prediction method based on a time sequence large model, which comprises the following steps: firstly, obtaining multivariate hydrological observation data of a target river basin and preprocessing to construct a multivariate time sequence matrix; secondly, dividing the matrix into a plurality of local time sequence blocks in the time dimension through a sliding window; thirdly, constructing an extreme value feature extraction module containing a maximum value operator and a first-order difference operator, extracting a local maximum value vector and a first-order difference absolute value vector for each local time sequence block; fourthly, splicing the two extracted vectors and the original local time sequence block in the feature dimension to obtain a reinforced block sequence; and finally, mapping the reinforced block sequence into input word units and inputting the input word units into a time sequence large language model for attention weight distribution and prediction, and outputting a hydrological flow prediction sequence of a future time window. The application can retain extreme value features, overcome cliff effect, and improve flood peak prediction accuracy.

The application discloses a flood forecasting method fusing a physical mechanism and a conditional denoising diffusion model, and comprises the following steps: obtaining multi-source hydrological data of a basin to be forecasted; driving a preset hydrological model to execute multi-stage sequential parameter optimization to obtain a basic forecasting sequence and an initial residual sequence; extracting multi-dimensional conditional variables based on the hydrological data and the initial residual sequence, and using a multi-dimensional evaluation criterion to perform dimension reduction screening to obtain a fusion conditional vector; inputting the fusion conditional vector as a guide condition into a pre-trained conditional denoising diffusion model, performing iterative denoising sampling through a multi-scale denoising operator to generate a forecasting residual probability sequence set; and coupling the basic forecasting sequence and the forecasting residual probability sequence set to obtain deterministic forecasting and probabilistic forecasting results of the basin to be forecasted. The application realizes a leap from single-point forecasting to probabilistic distribution deduction, can accurately obtain a flood deviation boundary under a changing hydrological environmental condition, and enhances the reliability of disaster prevention and mitigation decision-making.

The invention discloses a cascade reservoir joint compensation scheduling method considering flood sources and reservoir capacity levels, and belongs to the field of reservoir joint flood control scheduling. In order to solve the problems that a traditional scheduling strategy is poor in adaptability and a reservoir impounding role is fuzzy, the method is implemented through the following steps that the range of cascade reservoirs included in scheduling is determined, and engineering characteristic data such as flood control water levels and flood control high water levels are collected; calculating the amount of flood needing to be retained and stored based on the flood forecasting process of the downstream flood control point; a control reservoir and a matching reservoir are divided according to the hydraulic topological relation, the impounding capacity is distributed by combining the flood volume proportion of each reservoir and the remaining flood control reservoir capacity proportion, full-reservoir impounding distribution is completed through secondary distribution and recursion, the dispatching effect is checked, and the flow regulation and control factors are dynamically adjusted. The method is simple, practical, reasonable, reliable and capable of quickly adapting to different water regimes and work conditions, accurately positioning the reservoir retaining role, improving the overall flood retaining benefit of the cascade reservoir group, conforming to actual dispatching characteristics, being easy to popularize and providing scientific support for real-time dispatching decision of a reservoir operation management department.

This invention provides a computer vision-based method for optimizing and intelligently monitoring urban flooding simulation, belonging to the field of flood model monitoring technology. The method includes: determining the study area and establishing a one-dimensional hydrodynamic model of the water network based on initial modeling data; using computer vision, segmenting and extracting land use data of the study area using the SegNet and U-Net algorithms, reshaping the underlying surface, constructing new land use data, and calculating new uncertainty parameters; comparing and analyzing the performance of the one-dimensional hydrodynamic model before and after updating the uncertainty parameters; and constructing a one- or two-dimensional surface coupling model combining the integrated flood forecasting model ITF-FLOOD with land surface flux. This one- or two-dimensional surface coupling model is used to simulate the depth and area of ​​surface water accumulation within the study area, completing the optimization and intelligent monitoring of urban flooding simulation. This invention can improve the simulation accuracy of urban flooding models and provide a more intelligent and convenient monitoring method for flood-prone areas.

The invention discloses a water conservancy remote sensing data low-dimensional feature efficient extraction and prediction method based on deep metric learning, and the method comprises the steps: focusing multi-source data driving and integrated model optimization, collecting flood-related multi-source data, and carrying out the cleaning and standardization processing to construct a data set; a depth metric learning model containing time sequence and space feature extraction branches is built, and a high-discrimination feature vector is output by combining the target function with log-ex approximation; splicing the features and fusing the features into manual time sequence features to form a fused feature matrix; constructing an integrated base model set, dynamically adjusting parameters by means of intra-class / inter-class distances, and performing time sequence cross validation training; constructing a second-level training set by using a base model prediction result, training a second-level model learning weight, and establishing a final integrated regression prediction model; and inputting real-time data to obtain a forecast result, and carrying out iterative optimization according to the evaluation index. Through small-watershed and large-watershed tests, the method adapts to watersheds of different sizes, the flood forecasting precision and stability are effectively improved, and reliable technical support is provided for flood control decision making.

The invention belongs to the technical field of flood forecasting, and particularly relates to a method for calculating post-wave water level by sudden flood forecasting flow, which comprises the following steps: S1, acquiring hydrological observation and flow forecasting data, and determining a calculation control node; s2, calculating a first adaptive flow and a first water level before the node 2; s3, calculating a first adaptive flow and a first water level behind the node 2; s4, calculating a second adaptive flow and a second water level before the node 2; s5, calculating a second adaptive flow and a second water level after the node 2; s6, according to the first adaptive flow, the first water level, the second adaptive flow and the second water level, the water surface gradient process is calculated; and S7, outputting a post-wave water level forecast calculation result according to preset data submission ending time. According to the method, two post-wave flood indicating flow hydrographs and corresponding water level hydrographs can be generated, and the technical level of post-wave water level forecasting of the flood is effectively improved.

The application provides a set forecast precipitation information space-time error correction method and system based on image super-resolution, and relates to the technical field of precipitation forecast.The application corrects the space-time error of set forecast precipitation information based on the image super-resolution idea and a deep learning algorithm, improves the spatial resolution of set forecast precipitation information, and improves the forecast precision.The method has outstanding advantages in improving the set forecast precipitation space-time structure similarity and the precision of rain or no rain certainty classification forecast compared with the traditional statistical post-processing method, and provides high-precision data support for the application of set forecast precipitation information in flood forecast business.

The invention discloses a hydrological mechanism model parameter optimization method based on MoE, and the method comprises the steps: carrying out the session division and standardization of the historical data of a drainage basin at a data preprocessing stage, and extracting the key hydrological features of each session; in the parameter analysis stage, hydrological mechanism model parameters are layered by adopting a global sensitivity analysis method to form high, medium and low sensitive parameter sets; in a hybrid expert model construction stage, a multi-path expert network and a gating mechanism are introduced, and differential optimization and dynamic fusion of different sensitive level parameters are realized in combination with a feature enhancement module. According to the method, the efficiency and uniqueness of parameter calibration of an existing hydrological mechanism model are improved, and the precision and stability of the method in flood forecasting are remarkably enhanced.

The application belongs to the technical field of hydrological prediction, and particularly discloses a runoff prediction model considering extreme runoff and a construction method and system thereof, which comprises the following steps: collecting and preprocessing prediction factors of a research area to form a prediction factor set; determining the contribution degrees of the prediction factors and obtaining N prediction factors with the highest contribution degrees through preliminary screening; sequentially eliminating the N prediction factors from small to large contribution degrees and respectively inputting the preliminary prediction model for training and testing; selecting a group of prediction schemes with the highest Nash efficiency coefficient as a common model for the test results; simultaneously fitting the predicted runoff and the actual runoff output by the preliminary prediction model in the range exceeding the extreme threshold of runoff, and selecting a group of prediction schemes with the best fitting as an extreme value model; and coupling the common model and the extreme value model to obtain a comprehensive runoff prediction model. The application can effectively consider long-term runoff prediction and extreme flood prediction, and is suitable for popularization and use.

This invention provides a lightweight, edge-side LSTM-based integrated flood forecasting method based on ONNX. It acquires flow data and processes missing and outlier values, performs sliding window sampling and standardization, and constructs and trains an LSTM flood forecasting model based on the processed sample data. The trained LSTM flood forecasting model is then converted into an ONNX model and optimized through computational graph optimization, constant folding, quantization optimization, and hardware adaptation optimization to obtain a lightweight flood forecasting model. This lightweight flood forecasting model is then deployed on edge devices to perform local rolling inference on real-time collected flow monitoring data and output flood forecast results. This invention enables closed-loop operation of data acquisition, local inference, and forecast output on the edge, reducing model size, memory usage, and power consumption, and improving the real-time performance and reliability of flood forecasting in remote areas and scenarios with weak networks.

This invention provides a method for determining hydrological model parameters in data-free areas by combining dynamic and static features. Through multi-dimensional feature construction, improved dimensionality reduction methods, precise similarity partitioning, and parameter transfer, the method effectively improves the accuracy of determining hydrological model parameters in data-free areas. It provides a reliable technical method for water resource management, flood forecasting, and other work in data-free areas and has broad application prospects.

The application discloses a flood forecasting method based on time-varying parameters, and relates to the technical field of flood forecasting. The method comprises the following steps: determining the correlation between a target runoff recession coefficient and a target time-varying parameter and a preset reference time-invariant parameter by considering the target time-varying parameter and the reference time-invariant parameter; constructing a time-varying parameter reservoir model based on the correlation; and finally, performing flood forecasting based on the time-varying parameter reservoir model. The method can effectively consider the time-varying runoff intensity, and can analyze the whole system from a macro perspective, and can generalize the slope confluence process into a wide and shallow open channel flow process, so that the accuracy of flood forecasting is improved.

The application discloses a kind of face, block and point scale linkage flood warning method based on river forecast water level in plain area, according to the flood face scale early warning of plain representative site, for having face scale early warning area, using GIS technology and the spatial autocorrelation of plain area water level in flood forecast process, calculate submerged grid and depth, carry out block, point scale early warning.The method is a kind of simple, fast method.It includes the following steps: collecting plain area basic data;A kind of face, block and point multiscale linkage early warning mechanism;Compare regional representative site forecast water level and characteristic water level, dike elevation, and carry out face scale flood risk early warning;With face scale early warning area as unit, obtain regional forecast water level face data, generate the submerged grid and water depth of regional unit, and carry out block scale early warning;With face scale early warning area as unit, in combination with submerged face, submerged water depth and flood risk point data, utilize spatial analysis technique to carry out point scale early warning.

This invention discloses a multi-station collaborative interval forecasting method for watershed water levels based on conformal prediction, aiming to improve the safety and reliability of flood forecasting. The method includes: first, acquiring and denoising historical water level sequences of the target and associated stations; second, optimizing the historical window length based on automatic machine learning, and using a gated recurrent unit neural network to obtain the predicted water level values ​​of the target stations; third, combining river network topology information, performing topological weighted aggregation of historical prediction residuals from multiple stations to construct a spatially weighted residual pool that quantifies spatial uncertainty; and finally, introducing a threshold neighborhood risk weighting mechanism, whereby when the predicted point value approaches the warning water level, the confidence level is dynamically increased using a nonlinear function, and a larger residual quantile threshold is selected to generate a widened prediction interval. This invention integrates river network topology information, achieving automatic parameter optimization and risk-oriented dynamic interval adjustment, effectively ensuring coverage of warning water levels.

The application discloses a river network branch point hydrodynamic simulation method based on mechanism data fusion, relates to the field of hydraulic engineering, and comprises the following steps: acquiring simulation data by performing dynamic simulation on a local high-precision hydrodynamic model based on set boundary working conditions; acquiring a proxy model by performing model training based on a time sequence training sample set and pre-set branch point physical constraints; embedding the proxy model into a one-dimensional river network hydrodynamic model; and updating original branch point hydraulic states in the one-dimensional river network hydrodynamic model by using standard branch point results output by the proxy model. The model is trained based on a time sequence sample set and pre-set branch point physical constraints to acquire a proxy model, and intermediate parameters of the one-dimensional river network hydrodynamic model are updated by using inference results of the proxy model, so that the simulation efficiency of the one-dimensional river network hydrodynamic model is ensured, the overall accuracy of simulation results is improved, and real-time flood forecasting and engineering scheduling requirements are effectively supported.

The application relates to a flood cooperative forecasting management system and method based on big data analysis, and belongs to the technical field of big data analysis. The method comprises the following steps: obtaining a grid slope set of a target monitoring river basin according to overhead terrain data, obtaining a sub-river basin grid set and a sub-river basin set according to the grid slope set, and the division of the sub-river basin can help improve the accuracy of the location of the flood breach position of the target monitoring river basin; constructing a sub-river basin terrain parameter set according to the sub-river basin set, performing feature extraction on the sub-river basin terrain parameter set to obtain a sub-river basin terrain feature set, and obtaining referenceable flood data by analyzing the sub-river basin terrain feature set, so as to solve the problem of data scarcity of the target monitoring river basin, improve the referenceability of the data, process the referenceable flood data and a real-time monitoring data set through a flood risk assessment model to obtain a flood risk coefficient, obtain a flood forecasting list according to the flood risk coefficient, and improve the management efficiency of flood forecasting.

The application discloses a flood forecasting method and device based on historical cases, a storage medium and a computer. The method comprises the following steps: obtaining flood forecasting data of a to-be-predicted river basin, wherein the flood forecasting data comprises river basin environment data; obtaining a plurality of river basin flood cases, wherein the river basin flood cases comprise historical flood forecasting information and historical river basin environment data; determining the case similarity between the historical river basin environment data of each river basin flood case and the river basin environment data, and determining the river basin flood case corresponding to the case similarity higher than a preset threshold as a target case; determining the similarity weight and the confidence weight of each target case; for each target case, correcting the historical flood forecasting information thereof based on the similarity weight and the confidence weight, to obtain the corrected forecasting information of each target case; and determining the flood forecasting information of the to-be-predicted river basin based on the plurality of corrected forecasting information. The above scheme can improve the accuracy of flood forecasting.

The application discloses a kind of agent collaborative processing method and system for flood forecasting service, including receiving user initiated flood forecasting service request data and session identification;Parsing flood forecasting service request data, determine target service task;When the size of service document is less than the preset threshold, the content of service document is directly injected into the current session context;When the size of service document is not less than the preset threshold, the service document is blocked and vectorized, and the relevant document block is returned based on the similarity between the user request and the document block;Based on the injected session context, the corresponding executable object is called to execute the target service task, and the execution result is obtained;The execution result is fed back to the session corresponding to the session identification.The application can realize the cooperation of flood forecasting service document knowledge utilization, professional task execution, achievement automatic reference and continuous processing.

The invention discloses a mechanism data dual-drive flood forecasting method and system, belongs to the technical field of computers, and aims to realize the flood forecasting through a central routing agent dynamic allocation calculation mode in combination with physical mechanism model forecasting and generative state error model analysis and assisted by physical consistency forced verification. The problems that in a traditional method, a physical mechanism model is low in calculation efficiency, and a data driving model lacks physical constraints are effectively solved. Therefore, deep cooperation of real-time physical verification and dynamic error compensation in the forecasting process is achieved, and therefore the physical credibility, the calculation efficiency and the adaptive capacity to extreme flood events are improved in the flood forecasting process at the same time.

The invention discloses a hydrodynamic modeling method for flood control four-pre-platform construction, and particularly relates to the technical field of hydrodynamic model and flood forecast deduction, and the method comprises the steps: obtaining drainage basin basic data and flow data simulated by a hydrological model, and carrying out the preprocessing; the method comprises the following steps: constructing a drainage basin one-dimensional hydrodynamic model, extracting boundary conditions based on drainage basin basic data and flow data simulated by a hydrological model, arranging a drainage basin topological relation, carding a corresponding relation of the boundary conditions, taking one-dimensional section data contained in the boundary conditions as time sequence input, and sketching a modeling range of the drainage basin two-dimensional hydrodynamic model; and according to the drainage basin modeling range, constructing a drainage basin two-dimensional hydrodynamic model with a closed and independent partition scheme so as to facilitate parallel calculation of flood routing and realize evaluation and optimization of different flood control schemes.

The invention discloses a four-dimensional evaluation method for an intelligent algorithm flood forecasting result, and aims to overcome the technical defects that an existing flood forecasting algorithm evaluation system is single in dimension, lacks space-time universality consideration and does not give consideration to flood type adaptability and decision reliability. According to the method, a four-dimensional comprehensive evaluation system covering space-time expansibility, performance indexes, flood type adaptability, interpretability and uncertainty quantification is constructed. According to the method, each dimension index is mapped to a 0-10 quantized interval through a standardized normalization method, comprehensive and objective evaluation of the intelligent algorithm is realized in combination with a weighted comprehensive evaluation model, and the problems of algorithm selection blindness, limited application scene, high decision risk and the like are solved. Experimental verification shows that the method can effectively identify technical advantages and disadvantages of different algorithms, provides a scientific basis for model selection, optimization and engineering application of a flood forecasting algorithm, and is especially suitable for algorithm evaluation in a multi-scale and multi-type flood scene.