180 results about "Flood forecasting" patented technology

The invention relates to the technical field of hydrological forecasting, in particular to a deep learning-based short-term rainfall and flood forecasting method and system, and the method comprises the following steps: constructing a deep learning-based radar echo epitaxial prediction model and a rainfall and flood forecasting model, training and testing the two constructed models by adopting the first data set and the second data set respectively; inputting the rainfall radar echo data of the target watershed in a certain period of time into the radar echo epitaxial prediction model, and predicting the change trend of the rainfall radar echo in a future period of time; and inputting the change trend of the rainfall radar echo in the future period of time into the rainfall and flood forecasting model, and forecasting the rainfall and runoff of the target drainage basin in the future period of time. According to the method, the precipitation and the runoff of the target drainage basin can be predicted more accurately in a short time, and the prediction capability and the decision level of regional water resource management are effectively improved.

The invention discloses an interpretable artificial intelligence flood forecasting method and device based on a drainage basin runoff and runoff mechanism, a storage medium and electronic equipment. The method comprises the following steps: acquiring original meteorological and hydrological data of a target watershed, preprocessing the original meteorological and hydrological data, and calculating watershed surface average rainfall data and watershed surface average evaporation data; constructing an XAJRNN layer based on a Xinanjiang model hydrological principle and a recurrent neural network unit structure; constructing an interpretable deep learning flood forecasting model based on an XAJRNN layer, a normalization layer and an LSTM layer; and inputting the basin surface average rainfall data and the basin surface average evaporation data into the trained interpretable deep learning flood forecasting model, and performing real-time flood forecasting. According to the method and the device, the problem of lack of interpretability and physical consistency of flood forecasting results of a deep learning model in the prior art is solved.

The invention relates to the field of flood forecasting, and discloses a flood forecasting method and system, a storage medium and computing equipment, and the method comprises the steps: topological modeling: taking a river catchment area surrounded by a water diversion line and a catchment area outlet hydrometric station as nodes of a graph, constructing a connection relation of edges according to the actual flow direction of a river and the connection relation of branches, and carrying out the topological modeling; therefore, topological graph structure data adaptive to a hydrological mechanism is constructed; converting month number characteristics, generating a dynamic periodic timestamp vector, splicing the dynamic periodic timestamp vector with topological graph structure data, and inputting an attention neural network for training; the attention neural network architecture comprises two independent feature channels which are used for processing historical data and forecast data respectively, the network supports a variable modular architecture, and a flood forecast result can be output according to needs after a model is trained. According to the method, the flood prediction precision and interpretability are improved, and a high-precision and interpretable space-time joint flood prediction scheme is realized.

The invention relates to the technical field of flood forecasting, in particular to a flood forecasting method, device and equipment based on field monitoring and real-time modelling, and the method comprises the steps: collecting a topographic data file and an earth surface data text of a target area to construct a sub-meter DEM digital elevation model; radar real-time base data, radar and ground rainfall historical observation data, brightness temperature data and geographical coordinate information of a target area are obtained to solve actual radar real-time base data, rain short-term forecast data and soil moisture data, and then the data are utilized to calculate a preset hydrological model to obtain a mountain torrent gully basin distributed hydrological model result; and establishing a hydrological-hydrodynamic coupling model according to the sub-meter DEM digital elevation model and the torrential flood gully basin distributed hydrological model so as to forecast the current flood state and the future flood state. Therefore, the problems that an existing mountainous area hydrological model cannot completely reflect the current state of a drainage basin, and the simulation error of a hydrodynamic model is large are solved.

The invention discloses a rainfall runoff simulation method for performing session flood data cleaning based on a dynamic threshold value. The rainfall runoff simulation method comprises the following steps: collecting actually measured rainfall and section flow data of a target drainage basin in a multi-year hour scale; performing flood event segmentation on the continuous rainfall runoff data sequence of the target drainage basin based on the flow hydrograph morphological characteristics, and extracting rainfall runoff data corresponding to each flood session; through preliminary rainfall screening, rainfall runoff total amount comparison screening and fine screening based on runoff coefficient truncation analysis, abnormal flood sessions with unreasonable magnitude relationships in the rainfall-runoff data are identified and eliminated; by screening key time points in the flood process, flood session data with an abnormal time sequence is eliminated; and taking the cleaned flood data as the input of the session hydrological model, and carrying out normal simulation. According to the method, the flood data cleaning process is standardized, unreasonable flood session data are eliminated in advance, input data are ensured to be clean, and thus the flood forecasting precision is improved.

The invention discloses a physical mechanism and deep learning coupled flood forecasting method, device and system. The flood forecasting method comprises the steps of obtaining DEM data, hydro meteorological data and drainage basin static attribute data of a target drainage basin; based on the DEM data, dividing a target drainage basin into a plurality of sub-drainage basins, extracting a river network, and calculating static attribute data of the river network, the river network comprising a plurality of river channels; inputting the hydrometeorological data, the drainage basin static attribute data and the river network static attribute data of each sub-drainage basin into a pre-trained distributed hybrid hydrological model to obtain simulated flow of a hydrological observation station, and completing flood forecasting; the distributed hybrid hydrological model is coupled with a physical mechanism and deep learning. On the premise that the distributed hybrid hydrological model has good interpretability, high-precision daily runoff and flood simulation can be achieved, and high calculation efficiency is achieved.

The invention provides an intelligent flood forecasting method and device based on an FQI-DMTSNet self-adaptive generative adversarial mechanism and a readable storage medium thereof.According to the method, data are preprocessed through multi-source data fusion and assimilation, time sequence features are extracted through a dynamic hierarchical multi-scale time sequence sensing network (DMTSNet), a scheduling strategy is optimized by combining fitting value iteration (FQI) reinforcement learning, and the real-time flood forecasting method based on the FQI-DMTSNet self-adaptive generative adversarial mechanism is obtained. And data enhancement and data-free area parameter migration are realized by means of an adaptive feature-gated generative adversarial network (AFG-GAN), so that a closed-loop collaborative forecasting system is formed. The problems that a traditional model is high in dependence on data, poor in extreme scene adaptability and the like are solved, data-free area parameter generation, extreme scene sample enhancement and regulation and control strategy optimization are achieved, the accuracy, adaptability and practicability of flood forecasting are improved, and the method is suitable for flood forecasting, hydrological dispatching, urban inland inundation early warning and other scenes.

The invention provides a power grid equipment large-range progressive real-time flood forecasting method based on a three-water-source Xinanjiang model, which abandons a basin homogenization hypothesis of a traditional lumped model, divides a target basin into subunits with spatial uniqueness and topological connectivity through a Thiessen polygon method, and accurately matches underlying surface spatial differentiation features; according to the method, the limitation of single parameter calibration is broken through, an NSGA-I I multi-objective optimization algorithm is adopted to be combined with historical flood data, a parameter optimization system containing multiple indexes such as a flood peak flow error and a peak present time error is constructed, a Pareto optimal solution set is generated, a parameter knowledge base is constructed in a classified mode, and the problem of poor parameter adaptability is solved; a geographically weighted regression method is used for fusing satellite remote sensing, ground rainfall station and hydrometric station data, a dynamic correction mechanism is established, model input is updated every 30 minutes, and the rainfall runoff calculation temporal-spatial resolution is remarkably improved; after subunit outlet flow riverway convergence calculation is completed based on a Muskingum method, power grid equipment space distribution and vulnerability threshold values are innovatively associated.

The invention discloses an intelligent early warning method and system for snow-melting flood. The method comprises the following steps: constructing digital twin integrating snow-melting confluence and hydrological hydrodynamic force; collecting multi-source monitoring information, and forming driving data through quality control and spatio-temporal interpolation; driving data are input into digital twinning, and model parameters and states are updated through data assimilation; simulating snow-melting runoff, water level and flow to obtain snow-melting flood probability prediction; the method comprises the following steps of: solving a river reservoir gate pump scheduling scheme by using a multi-objective evolutionary algorithm by taking reservoir outlet flow, gate opening and pump station starting and stopping as decision variables, calculating a water level, flow and a submerging range in digital twinning, and selecting a target scheme according to an evaluation rule to generate a scheduling instruction and graded early warning information; monitoring data is collected during scheduling execution and compared with probability prediction, and when the deviation exceeds a threshold value, data assimilation and scheduling scheme updating are triggered. According to the invention, the snow-melting flood forecasting precision and flood control scheduling collaboration are improved.

The invention discloses an intelligent water conservancy-oriented drainage basin flood forecasting system based on capacity expansion of a flood storage and detention area, relates to the technical field of intelligent water conservancy flood prevention forecasting and early warning, and aims to solve the problem of flood simulation forecasting errors caused by distortion of a water level-volume curve under a low storage capacity condition when internal subsections of a flood storage and detention area are started or partially outburst. The system comprises a main riverway hydrodynamic simulation forecasting module, a flood storage and detention area starting judgment module, a flood storage and detention area topology analysis module and a flood storage and detention area subsection expansion simulation module. The main river channel hydrodynamic simulation forecasting module adopts a hydrodynamic model to carry out flood simulation forecasting, the flood storage and detention area starting judgment module judges whether a flood storage and detention area is started or not according to user input data and a main river channel simulation result, and the flood storage and detention area topology analysis module determines the connection and capacity expansion sequence of each independent branch. The flood storage and detention area subsection capacity expansion simulation module realizes subsection capacity expansion simulation by dynamically calculating the outer boundary and the internal flood diversion flow, and calculation is scientific and efficient.

The invention relates to a river channel confluence stabilization and efficient calculation method of a distributed hydrological model CASC2D, and the method comprises the steps: employing a semi-implicit local inertial wave method in a river channel unit to carry out the stable solving of a water flow evolution process, effectively inhibiting the numerical oscillation, and improving the model calculation efficiency; meanwhile, parallel computing is carried out on the double-layer nested grid structure of the model through OpenMP, a thread safety control and dynamic scheduling mechanism is matched, synchronous waiting and data competition are remarkably reduced, and the parallel utilization rate is increased. The method is particularly suitable for flood simulation of small and medium-sized watersheds in mountainous and hilly areas, the model calculation efficiency can be improved by 15 times while high-precision simulation is kept, and the risk of numerical instability under the short-time high-intensity rainfall condition is remarkably reduced. The method does not need to depend on a GPU or a distributed cluster platform, has high universality, good portability and engineering realizability, and is suitable for various application scenes such as flood forecasting, risk analysis and real-time scheduling.

The present invention provides a Transformer-based method for predicting floods over varying time periods. The method comprises obtaining and preprocessing hydrological and meteorological data for equal time periods; constructing and training a Transformer model; redistributing water volume over varying time periods in the watershed; and, for each time step, inputting the redistributed rainfall data into the trained Transformer model to obtain a predicted runoff sequence. The method can automatically identify rainfall events across time periods and generate new, unequal-time rainfall sequences based on the maximum short-duration rainfall. The Transformer model is used to generate unequal-time flood sequences with higher temporal resolution. The method can more accurately reflect the relationship between maximum rainfall and flood peaks, improving the accuracy and reliability of flood simulations and providing stronger technical support for disaster prevention and mitigation.

The invention discloses an urban rainstorm waterlogging process simulation method based on deep learning, and relates to the technical field of flood disasters. Comprising the following steps: constructing an urban inland inundation rainfall-water depth data set; training a ConvLSTM deep learning model by using the rainfall-water depth data set, and constructing a flood forecasting system based on the deep learning model; and constructing a training set and a verification set based on simulation results under different rainfall scenes, carrying out comprehensive evaluation on the performance of the ConvLSTM model, and carrying out verification in an actual scene. The coupling model provided by the invention not only improves the accuracy and speed of urban flood forecasting, but also enhances the adaptability to extreme weather events, and the future research focuses on further optimization of the model structure.

The invention provides a multi-mode coupling flood forecasting method and system based on a gated cross attention mechanism, and relates to the technical field of flood forecasting. The method comprises the following steps: acquiring high-precision rainfall monitoring information, numerical forecast rainfall information and hydrological station session flood excerpt information of a predetermined drainage basin; constructing training, verification and test samples; a spatial-temporal feature coding module based on a convolutional neural network and Transform is constructed, and rainfall information is fully mined; a time sequence feature coding module based on a time sequence convolutional network is constructed, and runoff information is fully mined; constructing a multi-modal feature fusion module based on a gated cross attention mechanism, and fusing different modal features mined by the time sequence and spatial-temporal feature coding modules; and constructing a flood prediction module based on a self-attention mechanism to perform flood prediction. According to the invention, the accuracy and reliability of flood forecasting are improved based on the multi-modal feature fusion technology.

The invention discloses an LSTM-CAM-Transform flood forecasting model for long and short dependence feature fusion, and relates to the technical field of flood forecasting. According to the method, an LSTM-CAM-Transform flood forecasting model is trained by using a root-mean-square propagation method based on periodic data, the LSTM and the Transform are coupled by the model through a cross attention mechanism, and the short-term feature extraction capability of the LSTM and the long-term dependence modeling capability of the Transform are fused and constructed to obtain the LSTM-CAM-Transform flood forecasting model. And verifying the precision of the LSTM-CAM-Transform flood forecasting model by using the decision coefficient based on the verification period data, and performing flood forecasting by using the trained LSTM-CAM-Transform flood forecasting model. According to the invention, joint identification of flood long-term and short-term characteristics is realized, and the accuracy and reliability of flood forecasting are improved.

The invention discloses a hydrological model parameter calibration method based on automatic session flood segmentation, and belongs to the field of runoff forecasting. The method makes full use of the data characteristics of the drainage basin flow process, proposes a method for segmenting session flood according to the flow percentage, deduces a target function weighted by the volume of session flood on the basis of the method, and obtains a hydrological model parameter calibration result. Hydrological model parameter calibration with session flood as a target is realized; according to the method, on the basis of an original hydrological model parameter calibration method, the high flow proportion in the parameter calibration process is adjusted, the defect that the flood forecast error of the flood period session is large due to the fact that the low flow proportion in a traditional parameter calibration method is too large is overcome, and the flood forecast accuracy is improved.

The invention discloses a flood forecasting method for constructing a dynamic unit line library based on rainfall flood data feature classification. The flood forecasting method comprises the following steps: S1, collecting historical data; s2, data cleaning and preprocessing; s3, constructing a flood data classification model; s4, unit line calculation and storage; s5, real-time flood forecasting; s6, performing visualization and verification; the method comprises the following steps: receiving short-term rainfall forecast of a meteorological department in real time, synchronously acquiring measured data of a rainfall station and a water level station in a drainage basin, counting current early-stage influence rainfall, performing feature analysis, establishing a CNN + K-means model, dynamically outputting a forecast result, generating a forecast flood hydrograph according to a calculation process, and continuously performing automatic rolling forecast according to real-time data. A forecasting result is automatically updated, and a real-time data and forecasting result chart is generated; a more scientific basis is provided for flood control decision making, and disaster loss caused by defects of a traditional method is effectively reduced.

The invention provides a watershed flood early warning system and method based on hydrological model and mobile application coupling, and belongs to the field of watershed flood early warning. In the system, a communication and data transmission module collects data for hydrological model simulation in a watershed area; the model simulation module simulates hydrological data of rainfall runoff in a watershed area in real time by using a hydrological model based on the collected data, and transmits a simulation result to the flood risk forecasting module through the communication and data transmission module; the flood risk forecasting module generates a flood risk map based on the received simulation result and transmits the flood risk map to the communication and data transmission module; and the mobile application display module receives real-time early warning information generated by the communication and data transmission module based on the flood risk map, and displays the real-time early warning information to a user. The watershed flood is simulated in real time through the hydrological model, the flood forecasting and early warning information is sent to people in a watershed flood submerging range through the mobile application, and the accuracy and timeliness of watershed flood early warning can be improved.

The invention discloses a flood forecast evaluation method adapted to reservoir dispatching characteristics, which comprises the following steps: respectively driving the same hydrological model by using historical observation rainfall and forecast rainfall, quantifying the error contribution of the hydrological model and rainfall forecast through double-track difference, and constructing a sample evaluation weight in combination with a rainfall score; constructing a weighting condition error distribution model based on the weight, and generating a reservoir flow forecasting scene set containing physical cause characteristics; and inputting the scene set into a reservoir scheduling model containing flood control and benefit-making rules, and outputting a comprehensive scheduling risk index containing a downstream over-alarm risk and water abandoning loss through water balance and downstream evolution calculation. According to the method, the statistical weight is corrected through error attribution, the abstract forecast error is converted into specific engineering risk and economic loss, the hidden risk working condition which is difficult to find by a traditional statistical index is effectively identified, and a quantitative basis is provided for risk decision making of reservoir dispatching and targeted optimization of a forecast system.

The invention discloses a large-scale flood risk dynamic assessment method for a power grid equipment flood-related area, and aims to solve the problems of risk assessment result distortion under complex hydrology, insufficient pertinence of power grid equipment flood risk assessment, insufficient risk assessment dynamic updating capability and the like in a traditional flood risk assessment method. According to the method, basic data such as meteorology, hydrology, terrain, remote sensing images and power grid historical flood conditions are integrated, and flood-prone areas are analyzed and determined; the method comprises the following steps: aiming at a flood-related area of power grid equipment, constructing a district distributed flood forecasting model based on a three-water-source Xinanjiang model, taking a hydrological forecasting model result as input, constructing a refined one-two-dimensional coupling flood routing analysis model, and simulating a flood inundation risk evolution process of the power grid equipment in different scenes; and constructing a power grid equipment multi-index flood inundation risk grading system to clarify the risk grade, constructing a large-scale flood risk assessment model, and outputting a dynamic risk assessment result by associating real-time data.

The invention discloses a Xinanjiang model parameter automatic calibration method based on a dual-objective optimization genetic algorithm, relates to the technical field of genetic algorithms and Xinanjiang models, and aims to solve the problems that a Xinanjiang three-water-source model has many parameters, many influence factors and large solution space, and a traditional method cannot well solve the problems. M points are randomly selected from a search space of Xinanjiang model parameters, the m points are coded, and a population is initialized; establishing a target function based on the error index, and obtaining the fitness of population individuals through the target function; carrying out evolution operation on the population through selection, recombination and variation to obtain breeding offspring; and performing reinsertion operation on the obtained breeding offspring to complete automatic calibration of the parameters of the Xinanjiang model, and completing the automatic calibration of the parameters of the Xinanjiang model. Individuals most suitable for the environment are obtained, and flood forecasting and water resource management are carried out.

The invention relates to the technical field of flood forecasting, in particular to a medium and small river flood risk level forecasting device based on areal rainfall, which comprises an information acquisition module used for acquiring hourly areal rainfall in each hydrological station area in a drainage basin and processing to obtain return periods of areal rainfall of different time lengths; the statistical module is used for acquiring monitoring data of each hydrological station, calculating a flood recurrence period, and then carrying out statistical analysis on the recurrence periods of different levels of flood in the drainage basin according to a standard flood level; and the different-level flood model module is used for counting the correlation between the recurrence periods of the different-time-length surface rainfall and the recurrence periods of the different-level flood so as to establish different-level flood models, and the forecasting precision is improved through the models.

The invention discloses a reservoir flood forecasting method based on physical constraint and space-time double-flow coupling. The reservoir flood forecasting method comprises the following steps: S1, acquiring multi-source hydrometeorological data; s2, decomposing the multi-source hydro meteorological data into a historical state sequence and a future driving sequence; s3, performing feature extraction on the historical state sequence data through the physical enhanced long-short term memory network of the historical inertial feature extraction branch to obtain historical inertial features, and performing feature extraction on a future driving sequence through the time domain convolutional network of the future forced feature extraction branch to obtain future forced features; s4, performing weighted fusion on the historical inertial features and the future forced features to generate fusion features; and S5, inputting the fused features into a decoder to obtain a predicted water level increment, and superposing the predicted water level increment to the current water level to obtain a predicted value. The prediction timeliness is improved, the prediction precision of the water recession stage is improved, and the physical consistency of the prediction result is enhanced.

The invention discloses a torrential flood forecasting and early warning method and device based on refined rainfall analysis, and the method comprises the following steps: 1, carrying out the investigation and evaluation of torrential flood, and determining the critical rainfall of each torrential flood risk region; 2, reorganizing rainfall site monitoring rainfall data in real time and performing spatial interpolation; 3, multi-source data integration-correction processing of short-term and imminent rainfall forecast and medium and short-term rainfall forecast; step 4, carrying out real-time rolling refined splicing on the monitored rainfall of the spatial interpolation, the integrated-corrected short-term and imminent forecast rainfall and the integrated-corrected medium-short term forecast rainfall in the time dimension; step 5; and carrying out real-time rolling analysis on the mountain torrent disaster risk based on the refined rainfall data, and carrying out advanced early warning. The method has the advantages that mountain torrent disaster risks are analyzed in advance through high-time-efficiency and high-precision refined rainfall, early warning is carried out, precious time can be reserved for risk avoiding transfer, and life and property losses are reduced.

The invention provides a large-scale watershed hydrological modeling method based on variable grids, and belongs to the field of large-scale distributed hydrological modeling. The large-scale watershed hydrological modeling method comprises the following steps: firstly, determining the size of a watershed basic grid according to the spatial resolution of available basic data and the calculation time consumption of a distributed hydrological model under different resolutions; secondly, determining a grid level according to a rainfall level and a mountain torrent prevention and control level of a flood event; and finally, according to the grid levels of the different partitions, the basic grids are aggregated into coarse grids of different levels, and basin variable grid hydrological modeling is completed. The method is suitable for a large-scale watershed flood forecasting system, and can meet the timeliness demand of flood control decision while guaranteeing the refined modeling of a key area.

The invention belongs to the technical field of hydrological forecasting, and particularly provides a method for forecasting future session flood based on peak rainstorm distribution characteristics, and the method comprises the steps: collecting and reorganizing data, and processing the data into a time sequence with equal time intervals; obtaining space-time distribution of the flood peak formation rainstorm amount of each session of the research basin; calculating the rainfall process of each sub-basin of the flood to be forecasted, and analyzing the spatial and temporal distribution characteristics of peak rainstorm; performing similarity analysis on the spatial and temporal distribution characteristics of the peak storm of the to-be-forecasted flood and the historical session flood to obtain a session flood which is highly similar to the spatial and temporal distribution characteristics of the peak storm of the to-be-forecasted flood and is close to the peak storm amount, and taking the session flood as a to-be-calibrated session flood; establishing a multi-objective function; optimizing the objective function to obtain an optimal parameter of the hydrological model; and performing hydrological forecasting on the to-be-forecasted flood. According to the method, the parameters of the hydrological model for forecasting the flood in the future session with high accuracy can be obtained, so that a high-accuracy flood forecasting result is obtained.

The invention discloses a method for forecasting flood in a drainage basin influenced by strong human activities based on a hybrid model. The method comprises the following steps: firstly, constructing a distributed hydrological model combining sub-drainage basin division and a Muskingum method, and performing calibration; using a hydrological model simulation error and multi-step watershed average rainfall based on river channel propagation time as characteristics, and using a Bayesian optimized LSTM (Long Short Term Memory) to construct an error correction model; and finally coupling the two models to realize real-time forecasting. Verification of the Lanxi river basin shows that compared with a traditional model, the Nash efficiency coefficient of the mixed model is remarkably improved, the flood peak relative error is smaller than or equal to 8%, the peak current time difference is smaller than or equal to 2 hours, precision is stable within the 12-hour forecast period, a long series of continuous historical data is not needed to serve as support, and only data in the session flood period are needed; the method is suitable for hour-scale flood forecasting of the watershed lack of data and strong human activity, an efficient and reliable technical means is provided for flood control and disaster reduction, and the method has wide engineering application prospects.

The invention discloses a river and lake regulation and storage system flood routing intelligent forecasting method based on a dynamic storage and discharge relation, and the method comprises the steps: constructing a space-time dynamic association diagram structure which reflects the waterpower relation of a drainage basin, and mapping multi-source hydrological data to generate a model input set with a unified space-time scale; establishing an aggregation reservoir parameterized structure model which comprises a dynamic water level-reservoir capacity relation driven by inflow change characteristics and a composite water level-flow relation integrating a time lag effect and backwater jacking feedback; constructing a flood regulation calculation equation set, and according to the real-time delay state in the composite relationship, adopting an adaptive solution mechanism to switch among different algorithm channels so as to solve the equation set; and correcting the model state by using the real-time observation data and executing rolling prediction. The method effectively solves the problems that in a complex river and lake system, dynamic characteristics of the storage and discharge relation are difficult to depict, the non-in-phase evolution process is unstable to solve, and high-dimensional parameters are difficult to calibrate, and the continuity and physical consistency of flood forecasting are improved.

The invention relates to the technical field of hydrology and water conservancy, and particularly discloses an intelligent real-time watershed flood early warning method and system, and the method comprises the steps: watershed data acquisition, fusion forecasting model construction, multi-stage early warning threshold calculation, early warning generation and online evaluation. According to the scheme, the real-time residual error of the physical hydrological model is constructed, the complex mode that the residual error changes along with the real-time rainfall intensity and the drainage basin state is learned by adopting the residual error prediction model based on machine learning, and an interpretable and steady forecast baseline and high-precision error correction are provided through a forward coupling and feedback coupling mixed architecture; the overall robustness of flood forecasting is effectively improved; dynamic inversion is carried out by using a fusion forecasting model, flood risk differences under different soil conditions are identified, the matching degree of early warning signals and actual risks is greatly improved, a comprehensive risk value is calculated in real time on a unified grid division system, and a risk zoning map is refreshed in real time along with updating of rainfall forecasting, so that the real-time prediction of the flood risk is realized. And the practicability of early warning information and the pertinence of emergency response are effectively improved.

The invention discloses a multi-target flood forecasting method based on coupling of a Xinanjiang model and deep learning. The method comprises the steps of 1, extracting a target watershed digital water system and subunit boundaries, and calculating watershed area rainfall; 2, hydrological data of a target drainage basin are collected, a data set is constructed, and an hourly rainfall-evaporation sequence and a secondary flood data set of the target drainage basin are obtained; 3, calibrating parameters of the Xinanjiang three-water-source model by adopting an NSGA-II optimization algorithm, generating a Pareto optimal solution set, and selecting a robustness parameter combination; and 4, jointly inputting the physical state variable output by the optimized Xinanjiang model and the original hydrological and meteorological data into a PatchTST deep learning model, capturing the nonlinear relationship between the physical state and the flow through a Transform architecture, generating corrected flow, and finally outputting a high-precision forecasting result. The method has the characteristics of multi-target collaborative parameter optimization, consideration of precision and physical consistency, accurate conversion of drainage basin geographic information and end-to-end high-robustness design.