60 results about "Real time forecasting" patented technology

The invention provides a distributed photovoltaic power generation grid-connected anti-countercurrent control system and control method. The system comprises a load prediction and optimization assembly, a countercurrent detection and response assembly and a control and optimization scheduling assembly. The method comprises the steps of receiving input of various data sources such as power grid dispatching data, user power utilization habits and weather forecast, predicting load requirements and power grid capacity in a period of time in the future in real time, and providing a decision basis for a control and dispatching component; receiving the load demand and the power grid capacity predicted by the load prediction and optimization component, and real-time power grid state data; when it is detected that countercurrent in the power grid is about to occur, an instruction for adjusting photovoltaic output is sent to the control and optimization scheduling assembly; receiving the predicted load demand and power grid capacity, and an instruction for adjusting photovoltaic output; and optimizing the scheduling strategy of photovoltaic power generation, dynamically adjusting the output of photovoltaic power generation, and sending a control signal to photovoltaic power generation equipment. According to the invention, power grid faults and loss caused by countercurrent can be reduced.

The invention discloses an intelligent traffic data processing method and system based on a 5G communication system, and relates to the technical field of traffic data processing, and the method comprises the steps: collecting data based on a 5G communication network; preprocessing the collected traffic data through an edge computing node; real-time prediction and risk assessment are carried out on traffic flow based on the cloud platform; sending real-time traffic information and optimization suggestions to a traffic management center and related vehicles; when an emergency occurs, a rapid response mechanism is triggered through a 5G network, and traffic resources are scheduled. The system collects, processes and analyzes traffic data in real time through the characteristics of high bandwidth and low delay of the 5G communication network and the distributed processing capability of edge calculation, achieves the real-time prediction and management of traffic flow and accident risks, and can improve the road utilization efficiency, reduce traffic accidents and improve the traffic safety through the intelligent decision and optimization control strategy. And the emergency response capability is improved, so that a more intelligent solution is provided for urban traffic management.

The invention belongs to the technical field of geological disaster monitoring and early warning, and particularly relates to a landslide catastrophe risk probability intelligent prediction method based on a two-phase substance point method. A two-phase substance point method is used for achieving rainfall-induced landslide catastrophe whole-process numerical simulation, then probability distribution of response quantities such as the sliding distance, the slump range and the slip mass volume is obtained, and the risk probability is predicted in combination with disaster-bearing body characteristics; according to the prior distribution of each parameter variable, a random sample is extracted as an input data set, a response quantity sample obtained through analog calculation is used as an output data set, an XGBoost model is adopted for training, and an agent model of a two-phase substance point method is constructed; and carrying out parameter inversion on slope monitoring data by utilizing Bayesian probability inversion, dynamically updating probability distribution of response quantity of each model, and predicting risk probability in real time according to the response quantity distribution. The physical significance is clear, and the prediction efficiency and the prediction precision are improved.

The invention relates to the technical field of unmanned aerial vehicle sensing, and discloses an unmanned aerial vehicle complex vector wind field sensing method and system, and the method comprises the steps: respectively recording the original data of a wind field sensing device in a wind tunnel simulation wind field test and the original data of an unmanned aerial vehicle carrying the wind field sensing device for a flight test, and constructing a feature data set; using the constructed feature data set to train an LSTM model, so that the LSTM model can predict wind field information at the current moment in real time according to historical wind field data, installing a wind field sensing device array on the unmanned aerial vehicle, collecting wind field information data in real time, and using the trained LSTM model to predict the wind field information in real time; and performing data fusion on prediction results of the plurality of sensors by adopting a Kalman filtering algorithm, and calculating wind field information such as three-dimensional wind speed, wind-induced air pressure change and turbulence. According to the method, real-time efficient prediction of the complex wind field environment in the air of the unmanned aerial vehicle is realized, and an effective means is provided for improving wind field detection and reliable operation of various unmanned aerial vehicles in the complex wind field environment.

The invention relates to the technical field of ecological cycle, in particular to an ecological cycle management method based on agricultural and pastoral integration. A data mining technology is adopted, soil sensor data, meteorological data, crop growth data and livestock health data are integrated, and a dynamically coupled nutrient management model is constructed; the nutrient requirements of crops and soil are predicted in real time by using algorithms such as deep learning, and a fertilization strategy is adjusted according to a prediction result, so that the limitation of static management is overcome, and precise fertilization is realized; the real-time monitoring technology is combined, and real-time adjustment is carried out according to the actual environment and crop requirements; nutrient supply is continuously optimized through a feedback mechanism system, the problem of lag in an existing management scheme is avoided, and it is guaranteed that crops obtain a proper amount of nutrients in different growth stages; historical data are deeply analyzed by using a data mining technology, key factors influencing nutrient supply and demand are mined, a scientific fertilization strategy is formulated, the utilization rate of data is remarkably improved, and uncertainty in management decision is reduced.

The invention discloses a small-scale greenhouse gas emission prediction method based on a pre-training model, and belongs to the field of greenhouse gas emission prediction. According to the method, a greenhouse gas monitoring micro station is arranged in a park, data such as greenhouse gas concentration and meteorological factors are collected in real time, and high-precision traceability of a greenhouse gas emission source and accurate accounting of greenhouse gas emission can be achieved in combination with a high-precision meteorological numerical value mode, a backward trajectory tracking algorithm, a receptor model and a diffusion model; the greenhouse gas emission amount is predicted in real time by calling the pre-training model, and the prediction efficiency and accuracy are remarkably improved. Different from a mode that a traditional model depends on a large amount of historical data training, the pre-training model can be finely adjusted on the basis of a small amount of existing data, so that effective learning and prediction are realized. According to the method, a large amount of time and computing resources are saved, the environmental change can be quickly responded, and real-time and accurate emission prediction is provided.

The invention relates to the technical field of electric power icing prediction, and discloses an electric power line icing prediction method and system, and the method comprises the steps: obtaining electric power data and meteorological data, and carrying out the data preprocessing; and extracting a first feature of the line icing to obtain a second feature and a third feature of the line icing, constructing a line icing prediction model through a machine learning algorithm, predicting the icing condition of the power line in real time, and outputting a line icing thickness prediction result. According to the method, physical prediction and a machine learning model are fused, multi-dimensional information such as weather, line states and historical data is combined, the icing thickness prediction precision is remarkably improved, real-time monitoring and dynamic early warning can be achieved, minute-level icing thickness updating and risk grading are supported, ice melting or dispatching instructions are triggered in time, and the ice melting efficiency is improved. The method still has relatively high robustness in complex environments such as mountainous areas, high altitudes and extreme weather; the manual inspection frequency can be reduced, and the operation and maintenance cost is reduced.

The invention discloses a deep well rock mass instability failure time accurate forecasting and early warning system, and relates to the technical field of deep well rock mass failure early warning, and the deep well rock mass instability failure time accurate forecasting and early warning system comprises the following steps: collecting a plurality of deep well rock samples to carry out a uniaxial compression experiment, and collecting acoustic emission signals in real time based on an acoustic emission sensor to obtain original acoustic emission signal data; carrying out self-adaptive preprocessing, and carrying out data classification and integration to obtain rock sample acoustic emission parameter data; constructing an original damage forecasting model, and performing model training and verification optimization based on the rock sample acoustic emission parameter data to obtain a rock damage forecasting model; collecting an acoustic emission signal of the deep well rock mass, and predicting the instability failure time of the deep well rock mass in real time based on the rock failure forecasting model; the method is used for solving the problems that when an existing deep well rock mass damage early warning technology carries out early warning on rock mass instability damage through acoustic emission monitoring, single or few characteristic parameters are manually processed and analyzed, and the early warning reliability and timeliness are not high.

The invention discloses a multi-modal household storage intelligent scheduling method and system based on large model prediction, and relates to the technical field of energy storage, and the method comprises the steps: obtaining multi-modal data in real time, and carrying out the preprocessing operation, the multi-modal data comprising electricity price, load electricity demand, photovoltaic power generation and weather data; inputting the preprocessed multi-modal data into the large model, and generating prediction results of each preset time period in a future period, including prediction results of electricity price, load electricity demand and photovoltaic generating capacity; based on a prediction result of the large model, adopting a multi-objective optimization algorithm to generate an initial charging and discharging scheduling strategy of each preset time period; and applying the initial charging and discharging scheduling strategy to the household storage system, and continuously optimizing the large model according to an execution effect so as to adjust the charging and discharging scheduling strategy in a future preset time period in a rolling manner. Multi-modal input variables are predicted in real time by introducing a large model technology, and a charging and discharging scheduling strategy is optimized, so that the adaptability and efficiency of a household storage system are remarkably improved.

The invention relates to a drainage basin hydrological forecasting method and system based on big data, and the method comprises the following steps: collecting drainage basin hydrological related data through a big data platform, transmitting the data to an edge calculation node of a data collection end in real time, and carrying out the preprocessing and feature extraction; forming a data set by the data after feature extraction, and dividing the data set into a test set and a verification set through space-time cross validation; respectively constructing and training a random forest model and a long short-term memory network model LSTM by using the training set to obtain a trained random forest model and a trained long short-term memory network model LSTM; fusing the trained random forest model and the long short-term memory network model LSTM by using a Stacking method to obtain a watershed hydrological prediction model, and evaluating and optimizing the watershed hydrological prediction model by using test set data; newly-collected data are processed by the edge computing nodes and then fused into the original data set, and basin hydrological data are predicted in real time through the optimized basin hydrological prediction model.

The invention relates to the technical field of meteorological disaster early warning, and provides an urban waterlogging early warning method and system based on grid meteorological prediction and an intelligent terminal, and the method comprises a data collection and local prediction step, a cloud fusion and regional prediction step, and a waterlogging evaluation and early warning response step. The lightweight AI model is deployed at an edge sensing terminal and an early warning terminal through a cloud edge-end collaborative architecture, local real-time prediction of meteorological data and dynamic fusion evaluation of waterlogging risks are realized, and the accuracy and efficiency of urban waterlogging early warning can be significantly improved.

The invention relates to the field of water supply systems, and discloses an urban pipeline leakage monitoring system based on a DMA partition management technology, and the system comprises a data collection module which is used for collecting the data of a flowmeter, a pressure sensor and a hydrophone in a DMA partition; the data preprocessing module is used for preprocessing the data; the time sequence analysis module is used for performing seasonal and trend decomposition and nonlinear time sequence analysis on the preprocessed data; the leakage detection module is used for predicting and detecting leakage; the noise analyzing and positioning module is used for analyzing and determining the specific position of the leakage point of the pipe section through the noise data of the hydrophone; and the real-time prediction alarm module is used for processing data in real time and triggering alarm when leakage is detected. The system not only can predict a potential leakage event in advance, but also can quickly position and give an alarm when leakage occurs, so that the operation efficiency and safety of the water supply system are greatly improved.

The invention relates to the technical field of weather data processing, in particular to a lightning forecasting and unmanned aerial vehicle route recommendation method based on electric field data, which comprises the following steps: outputting a first probability of lightning on each line section on an inspection line section; outputting a second probability of lightning on each line section on the inspection line section; fusing the first probability and the second probability in real time to obtain a fused third probability; whether the third probability of lightning on each section of line exceeds a preset value or not is judged, and when it is judged that the third probability of lightning on at least one section of line exceeds the preset value, the area position of the line section with the third probability exceeding the preset value is sent to an unmanned aerial vehicle control module; planning an unmanned aerial vehicle route based on the regional position through the unmanned aerial vehicle control module to obtain a second route; according to the invention, the route can be changed in time according to the lightning condition predicted in real time, so that the unmanned aerial vehicle can avoid the position where the lightning probability is large, and the occurrence of the situation that the unmanned aerial vehicle is hit by the lightning is reduced.

The invention discloses a demisting method, a model training deployment method, a demisting device and a vehicle. The automatic demisting method comprises the steps of obtaining model parameters of a side window temperature prediction model; input parameters of the side window temperature prediction model are obtained, wherein the input parameters comprise the outside humidity, the outside temperature, the inside humidity, the inside temperature and air conditioner real-time parameters; obtaining a real-time predicted temperature based on the side window temperature prediction model and the input parameters; obtaining a dew point temperature; and the real-time predicted temperature is compared with the dew point temperature, and when the real-time predicted temperature is not higher than the dew point temperature, air conditioner real-time parameters are adjusted so that the real-time predicted temperature can be higher than the dew point temperature. According to the demisting method, the real-time parameters of the air conditioner can be adjusted in real time according to the driving environment and the working condition, so that the vehicle can provide the best comfort, prevent the side window from fogging and automatically demist after fogging in different weathers, and the riding experience is improved.

The invention discloses a road risk simulation avoidance system and method based on multi-mode Internet of Things perception, and belongs to the technical field of intelligent traffic, and the method comprises the following steps: collecting the road condition information and road section characteristic parameters of a current road section in real time; screening out a plurality of monitoring vehicles; extracting historical driving behavior data and road condition characteristics of the plurality of monitored vehicles; constructing a historical driving behavior model, and constructing a scene model; extracting historical traffic road condition information, and extracting static feature data through the scene model; constructing a spatial feature matrix, and constructing a spatial-temporal feature matrix; performing feature weighted fusion on the spatial feature matrix and the spatial-temporal feature matrix to obtain a multi-model fusion evaluation result, and calculating a vehicle collision probability; and a vehicle owner is prompted to avoid danger and adjust steering, and an auxiliary steering avoiding function is started through the intelligent danger avoiding module. The problems that a traditional monitoring function cannot achieve risk modeling early warning and risk avoidance and lacks a multi-vehicle collision real-time prediction function are solved.

The invention relates to the technical field of solar photovoltaic power stations, in particular to a self-adaptive solar photovoltaic power station grid-connected control method and system, and the method comprises the steps: obtaining the weather forecast data of a first region, and the historical power generation data and real-time power generation data of a photovoltaic power station; training an output power prediction model based on the weather forecast data and the historical power generation data, and dynamically updating the output power prediction model by using the real-time power generation data; predicting future photovoltaic power generation output power in real time through the updated output power prediction model to obtain predicted photovoltaic power generation output power; analyzing the predicted photovoltaic power generation output power in combination with the load demand of the current power grid to obtain a current fluctuation trend and a load change trend; the current fluctuation state is judged through the load state and the current fluctuation state, and fault current is obtained; and based on the predicted photovoltaic power generation output power and the current fluctuation trend, constructing a fault current compensation model to dynamically compensate the fault current, and adjusting the fault current detection strategy of the relay protection device.

The invention relates to the technical field of traffic control, and discloses a traffic signal lamp control method capable of performing real-time prediction on traffic flow, and the method comprises the steps: S1, collecting multi-source traffic flow real-time monitoring data through roadside sensing equipment, S2, carrying out the data preprocessing and feature fusion processing of the multi-source traffic flow real-time monitoring data, and S3, carrying out the feature fusion processing of the multi-source traffic flow real-time monitoring data. Generating a standardized traffic flow feature vector; by verifying the consistency of multi-source sensor data in real time and detecting a data deviation problem caused by illumination shielding or equipment abnormity, the system can eliminate sensor input distortion, ensure the input accuracy of a traffic flow prediction model, further reduce signal control errors, improve the overall control reliability, and improve the traffic flow prediction accuracy. Through adaptive adjustment of model weight parameters and real-time correction of prediction deviation, the system can prevent a control strategy from lagging behind traffic evolution, ensure timely update of a signal phase sequence, reduce the risk of queuing overflow at an intersection, and further enhance adaptive control capability.

The invention discloses a real-time estimation method suitable for a heavy pollution weather emergency pollutant emission reduction proportion, and the method is based on a heavy pollution weather emergency emission reduction list and pollution source online data, and is suitable for the estimation of a heavy pollution weather emergency pollutant total emission reduction proportion. The method is used for estimating the emission reduction proportion of each pollutant in the heavy pollution weather emergency period in real time, reflecting the daily actual emission reduction proportion of each pollutant in the urban heavy pollution period, judging whether heavy pollution emergency management and control measures are implemented in place or not, and improving the heavy pollution weather emergency effect. The emission reduction effect of various pollution sources is enhanced, and real-time prediction and estimation can help a decision maker to timely adjust a management and control strategy and optimize emission reduction measures according to the pollutant concentration change, so that the air pollution level is effectively reduced, and the air quality is improved.

The present invention relates to the field of weather data processing technology, and specifically to a lightning forecast and drone route recommendation method based on electric field data, including outputting a first probability of lightning on each section of an inspection route; outputting a second probability of lightning on each section of an inspection route; fusing the first probability and the second probability in real time to obtain a fused third probability; judging whether the third probability of lightning on each section of the route exceeds a predetermined value, and when judging that the third probability of lightning on at least one section of the route exceeds a predetermined value, sending the regional position of the route section where the third probability exceeds the predetermined value to a drone control module, and planning the drone route based on the regional position by the drone control module to obtain a second route. The present invention can timely change the route according to the real-time predicted lightning situation, so that the drone can avoid the location with a high probability of lightning, thereby reducing the occurrence of the drone being struck by lightning.

The invention provides a recursive least square method-based building thermal load and room temperature online prediction method and system. The method comprises the steps of establishing a second-order thermal resistance and thermal capacity RC network model comprising a room air heat balance equation and a virtual enclosure structure heat balance equation; performing discretization processing on the second-order RC network model; a polynomial Kalman smoothing device is adopted to process indoor and outdoor temperature data containing noise; solving the temperature and thermal resistance and thermal capacity parameters of the virtual enclosure structure in real time; and according to the identified thermal parameters, the historical temperature data and the weather forecast data, the indoor temperature and the building thermal load at the next moment are predicted in real time. According to the method, the non-measurable temperature nodes of the virtual enclosure structure are incorporated into the to-be-identified parameters, the second-order RC model is constructed, and the thermal dynamic process of the building is reflected more truly; the accuracy of thermal load and room temperature prediction is directly guaranteed through model precision improvement, a reliable basis is provided for HVAC system optimization control, energy waste is avoided, and the thermal comfort of a user is improved.

The application provides a wind power short-term power prediction method, system and device affected by icing. First, weather data and icing data on the surface of a fan blade are acquired, the icing data is analyzed to obtain an icing type, then changes in the weather data are monitored and analyzed to obtain weather prediction data, and finally, the wind power short-term power is predicted in real time according to the icing type and the weather prediction data. The application can observe the icing structure characteristics from multiple directions, subdivide the icing structure, facilitate more detailed understanding of the type of the icing, better control the change trend of the icing, and make more reasonable prediction on the short-term power.

The application provides a rain day subway passenger flow early warning method and electronic equipment. The method comprises the following steps: obtaining relevant subway stations and relevant time periods of the relevant subway stations arriving at a target subway station in a target time period. Passengers arriving at the relevant subway stations in the relevant time periods can arrive at the target subway station in the target time period; obtaining the actual umbrella-carrying probability of the passengers arriving at the relevant subway stations in the relevant time periods according to meteorological data; obtaining the off-station probability according to the real-time rainfall of the target subway station in the target time period and the actual umbrella-carrying probability; obtaining the total number of stranded passengers in the target time period based on the off-station probability and the number of passengers arriving at the target subway station in the target time period; and initiating early warning when the total number of stranded passengers is greater than a carrying threshold. In this way, the total number of stranded passengers of the subway station is predicted in real time, and early warning is carried out accordingly, so that the sudden passenger flow can be managed and guided in time and effectively, the passenger flow management level of the subway station under sudden weather conditions is improved, and the safety risk is reduced.

The invention relates to the field of parking lot management, in particular to a method for predicting parking space occupation of a parking lot and related equipment. The method comprises the steps that area type identification information and historical data of a parking lot are acquired, and the historical data comprise corresponding parking space occupation data and weather data in a preset time period; determining a target mode corresponding to the long-term prediction model based on the identification information, the historical data and the date of the day; inputting the historical data and the current day weather into a short-term prediction model and inputting the historical data and the current day weather into a long-term prediction model based on the target mode for prediction processing to obtain short-term prediction data and long-term prediction data respectively; the short-term prediction data and the long-term prediction data are subjected to junction point processing based on the target prediction time, the prediction result of the day is obtained, and the time of the long-term prediction data is larger than or equal to the target prediction time and larger than the time of the short-term prediction data. The parking space occupation of the parking lot can be predicted in real time, and the parking space occupation prediction precision of the parking lot is further improved.

The invention provides a BMS (Battery Management System) control method and system based on active identification and reverse compensation, which are applied to a battery management system, and the method comprises the following steps: determining an SOC (State of Charge) initial value based on a standing voltage and an environment temperature; calculating an SOC correction coefficient according to the influence factors and dynamically updating the SOC correction coefficient; fusing an ampere-hour integral method and the correction factor to complete SOC estimation; carrying out real-time prediction and correction on the SOC by adopting an extended Kalman filtering algorithm and fusing voltage and current observation values; obtaining a current SOH value in combination with the internal resistance change rate and the capacity attenuation factor; and extracting a characteristic curve and comparing the characteristic curve with a reference sample to realize health state identification, then triggering reverse compensation of SOC estimation parameters according to an identification result, dynamically adjusting estimation parameters such as effective capacity and internal resistance, and improving SOC estimation precision under an aging working condition. The method is flexible in parameter updating, high in SOC estimation robustness and suitable for dynamic energy management and health management of the lithium iron phosphate battery pack in a complex working environment.

The invention relates to the technical field of power grid control, and particularly discloses an energy storage power distribution optimization method and system in a new energy station, a terminal and a medium, and the system comprises an energy storage management module, a power estimation module, a power control module and an energy prediction module, the system operation efficiency is improved, the equipment safety is guaranteed, the power estimation module predicts and calculates the power demand of the energy storage unit in the next preset period in real time, the power control module distributes the power demand for maintaining the system stability among different energy storage units, and the energy prediction module is used for predicting the future energy demand and supply and optimizing the scheduling strategy; according to the method, energy resources can be effectively utilized, the operation cost is reduced, the strategy is dynamically adjusted according to real-time data and changing environmental conditions, a decision maker is helped to make a more scientific and accurate decision, the fluctuation of wind power integration power is reduced, and the task completion time is shortened.

This invention relates to the field of meteorological disaster early warning technology, and provides a method and system for urban flooding early warning based on grid-based meteorological forecasting and intelligent terminals. The method includes: data acquisition and local forecasting steps, cloud fusion and regional forecasting steps, and flooding assessment and early warning response steps. Through a cloud-edge-device collaborative architecture, a lightweight AI model is deployed on edge sensing terminals and early warning terminals, enabling real-time local forecasting of meteorological data and dynamic fusion assessment of flooding risk, which can significantly improve the accuracy and efficiency of urban flooding early warning.

The invention discloses an online source analysis method and system for atmospheric pollutants, and relates to the technical field of pollutant source analysis. According to the main technical scheme, actually-measured meteorological data are input into a concentration loss prediction model, and the concentration loss is obtained; processing the concentration loss amount and the actually measured concentration change amount to obtain a theoretical concentration change amount; and inputting the emission factor and the theoretical concentration variation into a chemical mass balance model to obtain the contribution degree of each pollution source. The influence of meteorological conditions on the concentration loss amount is considered, and the aim of more accurately identifying and quantifying the contribution degrees of different pollution sources is expected to be achieved. The influence of the current meteorological condition on the concentration loss amount can be predicted in real time, and the aim of improving the evaluation accuracy of the contribution degree of the pollution source is expected to be achieved. According to the method, secondary pollutants are generated, and loss caused by diffusion dilution, rainfall / snowy wet settlement and other reasons is considered, so that the purposes of effectively ensuring the accuracy of theoretical concentration variation and further improving the accuracy of analyzing the contribution degree of a pollution source are expected to be achieved.

The invention provides a rainy day subway passenger flow volume early warning method and electronic equipment, and the method comprises the steps: obtaining related subway stations arriving at a target subway station in a target time period and related time periods of the related subway stations, and enabling passengers to arrive at the target subway station in the target time period when entering the related time periods of the related subway stations; according to the meteorological data, the actual umbrella carrying probability of passengers entering the subway station in the related time period is obtained; according to the real-time rainfall and each actual umbrella carrying probability of the target subway station in the target time period, obtaining a station leaving probability; based on the station leaving probability and the number of passengers arriving at the target subway station in the target time period, the total number of detained passengers in the target time period is obtained; and when the total number of the detained passengers is greater than the bearing threshold value, early warning is initiated. Therefore, the total number of the passengers detained in the subway station is predicted in real time, and early warning is performed in a targeted manner, so that the sudden passenger flow can be managed and dredged timely and effectively, the passenger flow management level of the subway station under the sudden weather condition is improved, and the safety risk is reduced.

The invention discloses a weather prediction method and device based on deep learning, a medium and equipment, and the method comprises the steps: obtaining meteorological data and a weather prediction model based on deep learning, and enabling the weather prediction model based on deep learning to be determined according to n cascaded / parallel weather prediction sub-models based on semantic segmentation, the input end of each stage of weather prediction sub-model except the first stage of weather prediction sub-model is in communication connection with the output ends of the first n-1 stages of weather prediction sub-models; respectively inputting meteorological data into each weather prediction sub-model of each weather prediction model based on deep learning, and correspondingly obtaining each piece of meteorological prediction data within a period of time after the current moment; and after each piece of weather prediction data is weighted, the final weather prediction data in the to-be-predicted space range is output, and through the application of the deep learning method, the calculation efficiency of the weather prediction model is greatly improved, so that the real-time prediction requirement is met, and a quick-response weather prediction result is provided for various application scenes.

This invention provides a method for predicting water levels at multiple tide gauge stations during typhoons based on a dynamic attention graph cyclic network, comprising the following steps: acquiring water level data and meteorological data from multiple tide gauge stations during a typhoon; performing time-series feature processing on the acquired water level and meteorological data and creating a topology; defining a dynamic graph attention hierarchical aggregation network model structure and constructing a dynamic attention graph cyclic network model; training and optimizing the dynamic attention graph cyclic network model using the processed data and graph topology; and using the optimized dynamic attention graph cyclic network model to predict the water levels at multiple tide gauge stations along the coast in real time for the next A hours during a typhoon. This method effectively balances prediction accuracy and computational efficiency, solving the problems of insufficient prediction accuracy and wasted computing power.