32 results about "Atmospheric circulation" patented technology

The invention belongs to the field of building performance simulation, and provides a monthly meteorological prediction data hourly downscaling method based on distribution adjustment time mapping. Comprising the steps of representative city determination, typical meteorological year data collection, global climate mode and monthly average resolution data set determination, main variable data set acquisition, candidate statistical distribution determination, maximum likelihood estimation, optimal distribution screening, empirical cumulative distribution function construction, distribution correction, quantile calculation, quantile mapping, distribution adjustment and model prediction. According to the method, hourly downscaling of monthly prediction data of different weather variables in the atmospheric circulation model is realized through a specific distribution adjustment and quantile mapping method, so that the physical characteristics of the data and the statistical characteristics of historical data are kept, and the long-term trend and extreme weather conditions of GCM prediction can be reflected; and reliable future meteorological data is provided for building performance simulation.

The invention discloses a method for performing extreme precipitation simulation by using a CNN-LSTM model based on multi-source data fusion. The method comprises the following steps: firstly, acquiring multi-source data of a research area, calculating the extreme precipitation of the area, and screening the most important key atmospheric circulation index influencing the extreme precipitation of a target area by using an extreme gradient lifting model; performing unified preprocessing on the multi-source data; then constructing a space-time prediction model fusing a convolutional neural network and a long-short-term memory network, and generating a deep learning prediction model through multi-feature joint training; and carrying out space-time prediction on the extreme precipitation by utilizing the preprocessed multi-source data and a deep learning model. According to the method, the extreme rainfall data, the atmospheric circulation index and the terrain DEM data are fused in a unified input framework, so that the model can perform joint prediction on rainfall by simultaneously utilizing the space-time rainfall distribution characteristics, the atmospheric circulation mode and the terrain information, and the prediction and simulation precision is high.

The invention discloses a deep learning-based drainage basin abnormal rainfall event identification monitoring and weather behavior analysis method and system, and the method comprises the steps: obtaining the summer rainfall original data of a drainage basin, and carrying out the preprocessing; the abnormal rainfall event is divided into different subspace modes through a self-organizing mapping machine learning method; qualitatively analyzing a space-time evolution rule and an interaction mechanism of water vapor conveying characteristics and atmospheric circulation configuration in abnormal rainfall according to each subspace mode; evaluating the influence of different meteorological factors on extreme weather conditions through a Bayesian method, and quantitatively exploring abnormal rainfall event change factors; the system comprises a data preprocessing module, a self-organizing mapping mode classification module, a water vapor conveying and atmospheric circulation influence mechanism module and a Bayesian method meteorological factor influence analysis module. According to the method, accurate identification and monitoring of the drainage basin abnormal rainfall event are realized, the internal characteristics of rainfall data can be effectively learned, and the identification capability of the abnormal rainfall mode is enhanced.

The invention discloses a multi-stage coupled climate mode optimization and uncertainty quantification method and system, and relates to the technical field of climate prediction, and the method comprises the steps: obtaining a multi-source meteorological data set, and carrying out the preprocessing of the multi-source meteorological data set; on the basis of the preprocessed multi-source meteorological data set, a multi-dimensional evaluation index system is used for screening a GCM combination with the best adaptability to a research area; based on the screened GCM combination, establishing a segmented statistical relationship between the simulation data and the measured data in the historical period, obtaining a correction factor, and performing deviation correction on the original output of the mode by using the correction factor; and on the basis of a deviation correction result, fitting the GCM data in each scene by adopting a common least square method and a smooth fourth-order polynomial, respectively calculating the variance and proportion of each uncertainty component, and obtaining the total uncertainty through linear superposition. According to the invention, the problems of large atmospheric circulation mode selection deviation, low correction precision and insufficient uncertainty management in the prior art are solved.

The invention relates to a rainfall objective prediction method based on an intelligent algorithm, and the method comprises the steps: decomposing the midsummer rainfall observation data of a time sequence based on an empirical orthogonal function, obtaining a spatial mode and a time coefficient, calculating a circulation field mean value of a target time period based on an atmospheric circulation field monthly value data set predicted by a climate mode, and obtaining the rainfall objective prediction result. The method comprises the following steps: obtaining a circulation abnormal field through a difference value with a same-period climate state circulation field, carrying out correlation analysis on a time coefficient and the circulation abnormal field, identifying and extracting a first correlation region with high correlation, carrying out correlation analysis on the time coefficient and atmosphere reanalysis data, obtaining a second correlation region with high correlation, and identifying and extracting a second correlation region with high correlation. And extracting a circulation anomaly average value of a similar region in the first related region and the second related region as an input feature, training a prediction model according to the input feature and a corresponding time coefficient, obtaining the corresponding time coefficient according to the prediction model and circulation anomaly feature data in the second time period, and performing reduction to obtain rainfall prediction data. And the accuracy of rainfall prediction is improved.

The invention relates to the field of climate change influence assessment, and discloses a method for quantizing contribution of natural internal variability to drought intensity based on a circulation similarity method, and the method comprises the steps: data collection; evaluating the drought strength based on the difference between the rainfall distance and the potential evapotranspiration distance, and quantifying the relative contribution of rainfall deficit and evapotranspiration demand to the drought strength; based on a circulation similarity method, a regional large-scale atmospheric circulation background is obtained, and contribution of a natural internal variation rate to drought intensity, rainfall deficit and evapotranspiration requirements is evaluated; estimating long-term evolution trends of drought intensity, rainfall deficit and evapotranspiration demands under different carbon emission scenes in the future; the method has the advantages that the natural internal influence mechanism of drought intensity change is defined, the relative contribution of historical and future rainfall deficit and evapotranspiration requirements to the drought intensity is quantified, the reliability of future drought trend estimation is enhanced, theoretical support is provided for making disaster prevention and reduction policies, and the accuracy of drought risk prediction is improved.

The present invention provides a weather forecasting method and device for multi-scale numerical weather prediction model fusion. The weather forecasting method includes: obtaining target forecasting factors for multi-scale numerical weather prediction model fusion based on general circulation forecast data and convective weather forecast data; and inputting the target forecasting factors into a pre-trained deep learning forecasting model to obtain a weather forecasting result. In the present invention, by fusing the forecasting factors of numerical weather prediction models at different scales and inputting them into a pre-trained deep learning forecasting model, the weather forecast is obtained, which solves the defect in the prior art that the forecasting effects of weather systems at different scales cannot be taken into account simultaneously, realizes the fusion of multiple-scale numerical weather prediction models, and effectively improves the forecasting effects of weather systems at different scales.

This invention belongs to the field of building performance simulation and provides an hourly downscaling method for monthly weather forecast data based on distribution-adjusted time mapping. The method includes the following steps: identifying representative cities, collecting data from a typical meteorological year, determining a global climate model and a monthly average resolution dataset, acquiring a primary variable dataset, determining candidate statistical distributions, performing maximum likelihood estimation, screening optimal distributions, constructing an empirical cumulative distribution function, modifying distributions, calculating quantiles, mapping quantiles, adjusting distributions, and finally model prediction. By using specific distribution adjustment and quantile mapping methods, the method achieves hourly downscaling of monthly forecast data for different weather variables in an atmospheric circulation model (GCM). This method not only maintains the physical properties of the data and the statistical characteristics of historical data, but also reflects long-term trends and extreme weather conditions predicted by GCMs, providing reliable future weather data for building performance simulation.

This invention discloses a method for simulating extreme precipitation using a CNN-LSTM model based on multi-source data fusion. First, multi-source data of the study area are acquired, and the regional extreme precipitation is calculated. An extreme gradient boosting model is used to select the key atmospheric circulation indices that have the most significant impact on extreme precipitation in the target area. The multi-source data undergoes unified preprocessing. Then, a spatiotemporal prediction model fusing convolutional neural networks and long short-term memory networks is constructed, and a deep learning prediction model is generated through joint training of multiple features. Finally, the preprocessed multi-source data and the deep learning model are used to perform spatiotemporal prediction of extreme precipitation. This invention integrates extreme precipitation data with atmospheric circulation indices and topographic DEM data into a unified input framework, enabling the model to simultaneously utilize spatiotemporal precipitation distribution characteristics, atmospheric circulation patterns, and topographic information to jointly predict precipitation, resulting in high prediction and simulation accuracy.

The invention relates to the technical field of new energy prediction, and particularly provides a photovoltaic abnormal output prediction method and device based on an atmospheric circulation field, and the method comprises the steps: calculating the difference between potential height grid prediction data of a preset height layer of a preset date in numerical weather prediction and climate state data of the preset height layer corresponding to the preset date, obtaining a forecast distance flat field of a preset height layer on a preset date; and predicting the abnormality of the photovoltaic output of the preset date based on the similarity between the forecast distance flat field of the preset height layer of the preset date and each element in a pre-constructed circulation feature set. According to the technical scheme provided by the invention, the application level of atmospheric circulation characteristics is enhanced in photovoltaic output prediction, so that the prediction capability of photovoltaic output, especially abnormal output, is improved.

The application provides a concrete carbonation depth transient prediction method and system based on a GCM, and the method comprises the following steps: obtaining working condition data of concrete under different climate conditions and carbonation depth of concrete tests, and establishing an empirical model of concrete carbonation depth; based on historical environmental factors of a region to be detected and environmental fluctuation rules, a GCM is used to mark a region where the concrete carbonation working condition data is obtained and to perform grid division, corresponding environmental factors in the target region are obtained, and a trained atmospheric circulation model environmental factor fitting formula is used to calculate and predict the atmospheric circulation environmental factor variables according to the environmental factors, the predicted atmospheric circulation environmental factor variables are substituted into the improved concrete carbonation depth empirical model, a transient prediction model of the concrete carbonation depth under different environmental conditions is established, and the obtained transient prediction model considers the fluctuation of the environmental factors, is more instantaneous in the real environment of the concrete carbonation test site, and has higher prediction accuracy.

The present invention relates to the field of climate forecasting technology, and in particular to a multi-model ensemble sea temperature three-step sub-seasonal climate forecasting method and system. The method comprises the following steps: obtaining sea surface temperature data and sea ice coverage data, and pre-processing the sea surface temperature data and the sea ice coverage data to obtain standard input data; constructing an input field for an atmospheric circulation model based on the standard input data; inputting the input field into the atmospheric circulation model, and obtaining atmospheric circulation information and surface information based on the atmospheric circulation model; and performing dynamic downscaling based on a regional climate model according to the atmospheric circulation information and the surface information to obtain high-precision sub-seasonal climate forecast results. The present invention effectively solves the problem of insufficient accuracy of existing forecasting technologies, improves the accuracy and stability of sub-seasonal climate forecasts, provides new ideas for sub-seasonal climate forecasts, improves sub-seasonal climate forecasting techniques, and enhances the accuracy of forecast products.

The present application provides a kind of drought forecasting method and system based on artificial intelligence and atmospheric circulation mechanism, comprising: S1, obtaining alternative precipitation runoff influence factor;S2, the importance of alternative precipitation runoff influence factor is sorted and the final precipitation runoff influence factor and precipitation runoff state number k value are determined;S3, with the Bayesian-NHMM model of minimum precipitation runoff prediction error as the best precipitation runoff prediction model, obtain precipitation and runoff prediction result;S4, realize the classification of land water storage forecast factor;S5, train machine learning model, and use the trained machine learning model as land water storage forecast model;S6, realize drought forecasting and early warning based on land water storage forecast model.The present application makes full use of the advantages of Bayesian-NHMM model and current machine learning model, makes up for the problem of low accuracy of traditional drought forecasting, and can deduce future drought risk and carry out drought risk early warning.

The invention discloses a multi-stage coupled climate mode optimization and uncertainty quantification method and system, and relates to the technical field of climate prediction, and the method comprises the steps: obtaining a multi-source meteorological data set, and carrying out the preprocessing of the multi-source meteorological data set; on the basis of the preprocessed multi-source meteorological data set, a multi-dimensional evaluation index system is used for screening a GCM combination with the best adaptability to a research area; based on the screened GCM combination, establishing a segmented statistical relationship between the simulation data and the measured data in the historical period, obtaining a correction factor, and performing deviation correction on the original output of the mode by using the correction factor; and on the basis of a deviation correction result, fitting the GCM data in each scene by adopting a common least square method and a smooth fourth-order polynomial, respectively calculating the variance and proportion of each uncertainty component, and obtaining the total uncertainty through linear superposition. According to the invention, the problems of large atmospheric circulation mode selection deviation, low correction precision and insufficient uncertainty management in the prior art are solved.

The invention provides a machine learning-based sub-season day-by-day rainfall prediction result correction method and system, and relates to the technical field of climate prediction. According to the method, in consideration of the deficiency of the prediction capability of the numerical mode on the circulation element of the predicted region at the time scale of the second season, the circulation element of the predicted region is not selected as the prediction feature, but the high-influence key region reflected by the numerical mode is objectively screened as the prediction feature; compared with some similar machine learning prediction methods, the method is more suitable for sub-seasonal climate prediction in feature selection, and mode data mining is more sufficient. Besides, in consideration of climate state differences in a year, data of adjacent report starting days and forecast days are used as samples, so that information inconsistency caused by importing atmospheric circulation information with great climate state differences into a model is avoided, generalization ability of the model is improved, and higher prediction precision and higher prediction speed are achieved.

The present application relates to icing precursor signal forecasting technical field, especially in icing signal analysis method and system based on circulation index, integrated modern meteorological analysis tool and data processing technology, aims at improving the prediction accuracy and response efficiency of icing event, through the comprehensive use of the fifth generation global climate atmospheric reanalysis data provided by the European center for medium-range weather forecasts, combined with the actual icing record of the specific area transmission line, the analysis and calculation of key atmospheric circulation index, including Siberia high pressure intensity index, east Asia trough intensity index, etc., through these data support, can accurately track and predict the possibility and severity of icing occurrence, application of advance-lag correlation analysis method, further reveals the dynamic relationship between icing and atmospheric circulation index, provides scientific basis for formulating effective icing response measures.

The invention relates to the technical field of icing precursor signal prediction, in particular to an icing signal analysis method and system based on a circulation index, integrates a modern meteorological analysis tool and a data processing technology, and aims to improve the prediction accuracy and response efficiency of an icing event. The method comprises the following steps: analyzing and calculating a key atmospheric circulation index by comprehensively utilizing fifth-generation global climate atmosphere reanalysis data provided by an European mid-term weather forecast center in combination with an actual icing record of a power transmission line in a specific region; comprise Siberia high-pressure strength indexes, East Asia groove strength indexes and the like, and the possibility and severity of icing can be accurately tracked and predicted through the support of the data; a lead-lag correlation analysis method is applied to further reveal the dynamic relationship between the icing and the atmospheric circulation index, and a scientific basis is provided for making effective icing response measures.

The present application relates to a kind of water quantity and water quality automatic regulation and control method, device, storage medium and electronic equipment at watershed scale, the method comprises: constructing watershed SWAT model, the model parameter of watershed SWAT model is verified, obtain the watershed SWAT model after verification, obtain the meteorological data of target region in target period predicted by atmospheric circulation model;The target water quantity and water quality of the target region in the target period are obtained, meteorological data is input into the watershed SWAT model after verification, obtain the water quantity and water quality of target region in target period, the water quantity and water quality regulation and control measure parameter of best management measure in the watershed SWAT model after verification is adjusted using non-dominated sorting genetic method with elite strategy, until the deviation between water quantity and water quality of target period and target water quantity and water quality is within the threshold range for stopping, obtain the water quantity and water quality regulation and control measure of the target region, reduce the artificial cost, improve the efficiency and precision of water quantity and water quality regulation and control.

The invention provides an end-to-end meteorological prediction large model, and belongs to the technical field of meteorological prediction. According to the end-to-end meteorological prediction large model, original observation data of a satellite, a radar, a ground station and the like are directly accessed, and multi-source data feature fusion is realized through a space-time attention mechanism. Layers comprise a macroscopic layer, a microscopic layer and a physical constraint module, the macroscopic layer adopts a three-dimensional convolutional neural network to capture global atmospheric circulation features, and the microscopic layer introduces a graph neural network to process complex boundary conditions such as terrains and urban underlying surfaces. According to the end-to-end meteorological prediction large model, the precision is improved, and under a 100-meter grid, the root-mean-square error of temperature, wind speed and other elements is reduced. And the real-time performance is enhanced: the five-minute-level update supports real-time applications such as power load scheduling and low-altitude logistics route planning. And economical efficiency optimization: deploying based on edge computing nodes without a super computer, so that the hardware cost is reduced.

A kind of multi-modal quantitative inversion method of paleoatmospheric circulation based on spatiotemporal deep learning network, using the powerful non-euclidean space feature extraction capability of spatiotemporal graph neural network (ST-GNN), captures the nonlinear spatiotemporal lag effect of vegetation response to climate; At the same time, coupling empirical orthogonal function decomposition (EOF) technology, orthogonal mode is extracted in potential feature space to capture the macroscopic atmospheric teleconnection pattern in geological record. The present application provides a quantitative inversion paradigm that can integrate micro-ecological mechanism and macro-dynamic mode, solves the problem of mixed signals in paleoclimate records of Qinghai-Tibet Plateau and surrounding transition zone, and realizes high-precision and high-resolution restoration of the evolution process of paleoatmospheric circulation intensity. In the multi-source stratigraphic record verification, not only can the mixed signals be effectively separated, but also the superior robustness and generalization ability are shown when dealing with geological age error and data loss.

This invention discloses a regional precipitation prediction method and system based on multi-source heterogeneous deep integration, belonging to the field of precipitation prediction technology. The method includes: classifying multi-source climate factors into atmospheric circulation, sea surface temperature, and comprehensive factor groups according to their physical properties; performing correlation screening, standardization, and PCA dimensionality reduction on each group of factors; constructing base models for each group of factors on a station-by-station basis using various machine learning algorithms, and employing a station-by-station hyperparameter search strategy; concatenating the out-of-place predictions of all base models and using the result as input to the ensemble layer of a deep neural network, learning dynamic ensemble weights station-by-station; and finally outputting the precipitation prediction results. This invention effectively eliminates the systematic bias of a single algorithm through heterogeneous integration and improves the model's adaptability to spatial heterogeneity through station-by-station hyperparameter search, exhibiting advantages such as high prediction accuracy, good interdecadal stability, and strong predictive ability for extreme precipitation events.

The invention relates to the technical field of climate forecasting, in particular to a multi-mode set sea temperature three-step seasonal climate forecasting method and system, and the method comprises the following steps: obtaining sea surface temperature data and sea ice coverage rate data; preprocessing the sea surface temperature data and the sea ice coverage rate data to obtain standard input data; constructing an input field of an atmospheric circulation mode based on the standard input data; inputting the input field into the atmospheric circulation mode, and obtaining atmospheric circulation information and earth surface information based on the atmospheric circulation mode; and according to the atmospheric circulation information and the earth surface information, performing dynamic downscaling based on a regional climate mode to obtain a high-precision sub-seasonal climate prediction result. According to the method, the problem of insufficient precision of the existing prediction technology is effectively solved, the accuracy and stability of the sub-seasonal climate prediction are improved, a new idea is provided for the sub-seasonal climate prediction, the sub-seasonal climate prediction skill is improved, and the prediction product precision is improved.

The invention provides a cold season composite low-temperature heavy snow event attribution analysis method based on circulation typing, which comprises the following steps of: performing climate mode deviation correction calculation and snowfall calculation on the basis of actually measured meteorological and hydrological data, global climate mode data and atmospheric circulation data of a collected research area; using a self-organizing mapping neural network to carry out circulation typing on atmospheric circulation variables, identifying composite extreme low temperature heavy snow events under different circulation types, and carrying out event feature analysis; based on the circulating current type frequency change, performing external forced detection attribution analysis and constraint estimation analysis of a global climate mode in a future period by adopting a regularization optimal fingerprint method; a climatic change separation method is adopted, and separation of thermal action, dynamic action and thermal-dynamic interaction is carried out through the frequency change trend of a composite extreme event. The invention provides a new perspective for the attribution of the climate change and human activity influence of the composite extreme climate event.

The application discloses a kind of regional precipitation isotope virtual monitoring method and system based on atmospheric circulation model, belong to hydrology meteorological monitoring and isotope hydrology technical field, method includes: based on engineering application demand obtains the reference point of target monitoring area, obtains the precipitation isotope simulation data at reference point and forms reference data sequence;Obtain the virtual precipitation isotope simulation data of multiple atmospheric circulation models, form candidate model set;From candidate model set, the virtual precipitation isotope simulation data of the position where reference point is located is extracted, and mode data sequence is obtained;The matching degree between mode data sequence and reference data sequence is calculated, and the optimal model and corresponding virtual precipitation isotope simulation data are selected from candidate model set;Based on the virtual precipitation isotope simulation data corresponding to optimal model, corresponding reference point in target monitoring area is filled with missing monitoring data or corrected original monitoring data, and the continuity of historical monitoring data of observation station is improved.

This application discloses a drought-flood transition prediction method, system, and storage medium that integrates artificial intelligence and physical mechanisms. The method includes: collecting multi-source data within a selected drought-flood transition study area; optimizing key factors affecting terrestrial water storage at each grid point within the study area; constructing and calibrating a temporal convolutional network deep learning model; using a long series of key factor sets to drive the deep learning model and reconstruct a long series of terrestrial water storage inversion datasets; extracting atmospheric circulation characteristics during drought-flood transition events and drought-flood transition events to construct a series of atmospheric circulation-drought-flood transition intensity pairings; deriving a nonlinear scaling relationship between drought-flood transition intensity and atmospheric circulation characteristics; and predicting the intensity and risk of drought-flood transitions at each grid point in the future. This invention provides a new perspective for drought-flood transition prediction, has important scientific significance for disaster prevention and mitigation, and provides a theoretical and technical foundation for assessing the environmental and hazard effects of Earth system evolution.

The invention discloses a spring rainfall prediction method based on atmospheric circulation abnormal signals. The method comprises the following steps: S1, obtaining and preprocessing multi-source data; s2, key circulation system identification and signal extraction; s3, construction of a synergistic effect index: constructing a circulating current synergistic effect index for comprehensively representing the configuration states of the plurality of circulating current systems in the specific region based on the circulating current signals extracted in the S2; s4, establishing a rainfall prediction model; and S5, testing a prediction result and integrating services. According to the method, abnormal signals of a plurality of key circulation systems including arctic oscillation and remote correlation wave train thereof, a south branch groove, subtropical high pressure, a water vapor transfer channel and a cold air activity path are systematically identified and quantified from multi-source data, so that the problems of high dependence on a single factor and incomplete physical mechanism of a traditional method are solved; and a comprehensive and three-dimensional circulation background support is provided for prediction.

The invention discloses a story line probability coupling attribution method for a heat wave flood space composite event, and relates to the field of climate change. The method comprises the following steps: firstly, quantifying a spatially compounded heat wave flood extreme event by utilizing a compound index; based on a story line attribution framework, similar circulation is constructed by extracting circulation characteristics and according to a multi-dimensional Euclidean distance, so that contribution of dynamic factors of atmospheric circulation and thermodynamic factors of man-made force to events is disassembled and evaluated; then, under a probability attribution framework, based on the Euclidean distance, the occurrence probability and the recurrence period of similar events in historical and future specific periods are evaluated, and a time change sequence of the future occurrence probability is analyzed; according to the method, quantification, driving factor contribution analysis and future risk estimation of the spatial composite extreme events are realized, and a scientific basis is provided for mechanism understanding, forecasting and early warning and disaster prevention and reduction of the events.

The invention discloses a power grid safety high influence weather dynamics conceptual model construction method and system, and relates to the technical field of meteorological prediction, and the method comprises the steps: analyzing geographic data influencing power grid safety, and processing and interpolating irregular terrain data based on a Kriging interpolation method; key factors are extracted based on geographic data, and weather processes influencing power grid safety are classified according to seasons and influence mechanisms; key configuration of various weather processes is obtained, and a power grid safety high influence weather dynamics conceptual model library is constructed through terrain correction. According to the method, key factors including an atmospheric circulation factor, a convection indication factor and a ground meteorological factor are extracted, so that the accuracy of weather process classification is improved; the comprehensive dynamic conceptual model library is constructed, so that the comprehensiveness and the accuracy of dynamic evaluation of the power grid safety are improved, and the applicability and the reliability of the model library are improved.

The invention discloses a long-term-ultra-long-term runoff prediction method based on dynamic prediction factors and artificial intelligence. The method comprises the following steps: S1, data collection and preprocessing; s2, determining a month to be predicted and prediction timeliness, setting a training set and a test set, and carrying out standardization processing on data; s3, taking the reservoir inflow of the to-be-predicted month of the key reservoir as a prediction object, taking the atmospheric circulation index and the sea temperature index of 24 months before the report starting date as candidate prediction factors according to the prediction time efficiency, and screening an optimal prediction factor combination by combining correlation analysis and a nonlinear stepwise regression method; s4, constructing a GRU runoff prediction model of the upstream of the Yangtze River and training the model; s5, carrying out Yangtze river upstream reservoir flow real-time prediction based on the dynamic prediction factor and a GRU prediction model; by setting the sliding training window and dynamically updating the prediction factor combination and the prediction model parameters, the prediction skill and the prediction period of the reservoir inflow of the key reservoir at the upstream of the Yangtze River are remarkably improved, and an effective basis is provided for hydropower dispatching and management.

The invention discloses a wind speed prediction method and device, electronic equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring a climatic factor index of a first time interval in a predetermined area; inputting the climatic factor index of the first time interval into a wind speed prediction model to obtain a wind speed of a second time interval in the predetermined area; wherein the wind speed prediction model is trained by the following steps: obtaining training data, the training data comprising climate factor indexes and historical wind speeds in a predetermined area, and the climate factor indexes comprising a plurality of atmospheric circulation indexes and a plurality of sea temperature indexes; the climatic factor index of a third time interval in the preset area is used as input feature data, the historical wind speed of a fourth time interval is used as label data, a wind speed prediction model reflecting the relation between the climatic factor index and the wind speed is trained, and the third time interval is earlier than the fourth time interval. Therefore, the problem of poor wind speed prediction effect can be solved.