Electric power meteorological disaster risk early warning method and system based on artificial intelligence

Abstract

The invention relates to an electric power meteorological disaster risk early warning method and system based on artificial intelligence. The method comprises the following steps: cleaning and preprocessing basic meteorological data including temperature, humidity, wind speed and precipitation to obtain a cleaned basic meteorological data set; on the basis of satellite cloud picture or radar image data, meteorological changes are modeled through a neural network model, assimilation features are obtained, and the basic meteorological data and the assimilation features are spliced to form a joint feature vector; and constructing a comprehensive disaster risk assessment model based on the meteorological factors, the power grid factors and the geographic information, carrying out training and multi-objective optimization on the comprehensive disaster risk assessment model, and carrying out real-time electric power meteorological disaster risk early warning by using the optimized comprehensive disaster risk assessment model. According to the invention, meteorological data, power grid equipment states and geographic information can be monitored in real time, meteorological disasters possibly suffered by power equipment can be accurately warned, and the safety and stability of a power system are improved.

Description

Technical Field

[0001] The present invention relates to the field of electric power meteorological disaster risk warning, and in particular to an artificial intelligence-based electric power meteorological disaster risk warning method and system. Background Art

[0002] In the operation of the power system, meteorological factors have a direct impact on the safety and stability of the power grid. For example, meteorological disasters such as wind, ice and snow, and lightning may cause power equipment failures and affect power supply stability. Existing traditional meteorological early warning systems usually rely on manual observations and basic meteorological data analysis, and have problems such as low monitoring accuracy, poor early warning timeliness, and insufficient real-time correlation analysis of power grid equipment status. The power system urgently needs an early warning system that can accurately monitor and analyze meteorological data in real time and is highly integrated with the operation status of the power grid. Summary of the invention

[0003] The present invention proposes an artificial intelligence-based power meteorological disaster risk warning method and system. The present invention can monitor meteorological data, power grid equipment status and geographical environment information in real time, and accurately predict the meteorological disasters that power equipment may suffer through intelligent algorithms, and issue warning signals in advance, which significantly improves the safety and stability of the power system.

[0004] The present invention is implemented by the following technical solution: an artificial intelligence-based power meteorological disaster risk early warning method, the steps are as follows: Step S1: Collect various meteorological data, geographic information and power grid operation data; Step S2: Clean and preprocess the basic meteorological data including temperature, humidity, wind speed, and precipitation to obtain a cleaned basic meteorological data set; based on satellite cloud images or radar image data, model the meteorological changes through a neural network model to obtain assimilated features, and concatenate the basic meteorological data and the assimilated features to form a joint feature vector; Step S3: construct a comprehensive disaster risk assessment model based on meteorological factors, power grid factors and geographic information, train and optimize the comprehensive disaster risk assessment model, and use the optimized comprehensive disaster risk assessment model to conduct real-time power meteorological disaster risk warning; the comprehensive disaster risk assessment model is expressed as: ; in, is the comprehensive disaster risk, T is temperature, H is humidity, W is wind speed, P is precipitation; L is grid load, C is grid equipment health status, G is geographic information, are the weight coefficients of meteorological factors, power grid factors, and geographic information, respectively. Respectively represent the 1st, 2nd, and 3rd meteorological risk weight coefficients, are the weights of wind speed, precipitation, and the combination of wind speed and precipitation, These are two coefficients obtained by fitting historical load data and load disaster data; It is the relationship coefficient between the health status of power grid equipment and disaster risk.

[0005] Further preferably, in step S2, basic meteorological data including temperature, humidity, wind speed and precipitation are cleaned by a Kalman filter algorithm.

[0006] Further preferably, in step S2, a convolutional neural network (CNN) is combined with a recurrent neural network (RNN) to model the meteorological changes and obtain assimilated features.

[0007] Further preferably, in step S2, the extraction process of assimilation features is expressed as: ; in, is the assimilation characteristic at time t, is the satellite cloud image or radar image at time t, is the total time, RNN and CNN represent recurrent neural network and convolutional neural network respectively; ; in, represents the i-th meteorological feature after meteorological assimilation, and n is the number of meteorological features.

[0008] Further optimization, the joint feature vector It is expressed as: ; Through the attention mechanism or weight distribution model, the basic meteorological data and assimilated features are weighted and fused to define a weight vector sequence. , then the joint eigenvector Calculate as follows: ; Among them, the weight vector sequence is obtained by training historical data or dynamically calculated using the attention mechanism. Represent the 1st, 2nd, …, mth weight coefficients respectively.

[0009] Furthermore, for the joint feature vector, dynamic weighting is achieved through reinforcement learning or neural network to continuously optimize the weight coefficient.

[0010] Further preferably, in step S3, a multi-objective optimization (MOO) method is adopted to comprehensively consider the three objectives of accuracy, timeliness and computational efficiency, perform multi-objective optimization on the comprehensive disaster risk assessment model and dynamically adjust the optimization strategy.

[0011] Further optimization is performed by combining multiple objectives and forming a comprehensive optimization objective of the comprehensive disaster risk assessment model through weighted summation: ; in, is the weight coefficient of the accuracy index, is the weight coefficient of the timeliness index, represents the accuracy index, Indicates timeliness indicator.

[0012] Furthermore, in the multi-objective optimization process, a reinforcement learning (RL) mechanism was introduced to automatically adjust the optimization strategy with real-time feedback.

[0013] The present invention also provides an artificial intelligence-based electric power meteorological disaster risk early warning system, including a data acquisition terminal, a background service terminal, and a display terminal; The data collection terminal is responsible for collecting various meteorological data, geographic information and power grid operation data; The backend server processes and integrates meteorological data, geographic information and power grid operation data from multiple meteorological data sources; the backend server has built-in meteorological data preprocessing module, meteorological data assimilation module, feature splicing module and comprehensive power disaster risk assessment model; The meteorological data preprocessing module cleans and preprocesses the basic meteorological data including temperature, humidity, wind speed, and precipitation to obtain the cleaned basic meteorological data set; The meteorological data assimilation module combines convolutional neural network (CNN) and recurrent neural network (RNN) to model meteorological changes and obtain assimilated features; The feature stitching module fuses basic meteorological data with assimilated features; Evaluate the risk of power grid under different meteorological conditions through comprehensive disaster risk assessment model and generate corresponding early warning signals; The display terminal provides an interface for interacting with users, intuitively displaying relevant data and early warning signals.

[0014] The present invention can significantly improve the power company's ability to respond to meteorological disasters. It can not only effectively reduce equipment failures caused by meteorological disasters, but also help operation and maintenance personnel to provide timely warnings, take precautions in advance, and reduce losses caused by sudden meteorological disasters.

[0015] (1) Technology Integration: It integrates multiple advanced technologies such as meteorological monitoring, data analysis, predictive modeling, and intelligent early warning. It can collect multi-dimensional data in real time and combine meteorological data with power grid operation data to achieve accurate analysis.

[0016] (2) High-precision monitoring: This system uses high-precision sensors and monitoring devices to make up for the blind spots of traditional meteorological data and improve the accuracy of meteorological data. In particular, it is superior to traditional methods in monitoring the accuracy of specific areas and equipment in the power grid.

[0017] (3) Real-time warning: Compared with the delay of traditional meteorological warning systems, this system can achieve real-time monitoring and short-term warning at the 15-minute level, greatly improving the timeliness of the warning and reminding power companies to take measures earlier.

[0018] (4) Intelligence level: The present invention uses artificial intelligence algorithms (CNNs, RNNs, ANFIS) to process meteorological data and power grid equipment operation data. It can automatically identify and analyze the complex laws of the meteorological-power grid system, improve the prediction accuracy and dynamic adjustment capabilities, provide early warning and accurately guide decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work. Example 1

[0021] Reference Figure 1 , an artificial intelligence-based power meteorological disaster risk early warning method, the steps are as follows: Step S1: Collect various meteorological data, geographic information and power grid operation data; Step S2: Clean and preprocess the basic meteorological data including temperature, humidity, wind speed, and precipitation to obtain a cleaned basic meteorological data set; based on satellite cloud images or radar image data, model the meteorological changes through a neural network model to obtain assimilated features, and concatenate the basic meteorological data and the assimilated features to form a joint feature vector; Step S3: construct a comprehensive disaster risk assessment model based on meteorological factors, power grid factors and geographic information, train and optimize the comprehensive disaster risk assessment model, and use the optimized comprehensive disaster risk assessment model to provide real-time power meteorological disaster risk warning.

[0022] Specifically, since the collected data comes from different sources (weather stations, power grid equipment, satellite data, etc.), and their formats and accuracy vary, data cleaning and preprocessing are required. The meteorological data preprocessing module uses the Kalman filter algorithm to filter out noise in the data to ensure the accuracy and timeliness of the data.

[0023] ; in, is the estimated state of meteorological data at the current moment, is the estimated state of meteorological data at the previous moment, is the Kalman gain, are observation data such as temperature and wind speed, The observation matrix is ​​obtained by Kalman filtering, which can eliminate random noise in the data and provide accurate data input for subsequent analysis. The basic meteorological data (such as temperature T, humidity H, wind speed W, precipitation P, etc.) are cleaned and preprocessed to obtain the cleaned basic meteorological data set. , is the temperature at time t, is the humidity at time t, is the wind speed at time t, is the precipitation at time t.

[0024] In step S2, the meteorological changes are modeled and assimilated features are obtained by combining convolutional neural networks (CNN) and recurrent neural networks (RNN). CNN is used to extract spatial features from meteorological images (such as satellite cloud images, radar data, etc.), while RNN is good at processing time series data and capturing the temporal characteristics of meteorological evolution.

[0025] Satellite cloud image or radar image data ,in is the satellite cloud image or radar image at time t, is the total time. After CNN processing, the spatial feature matrix of satellite cloud images or radar images can be extracted. Then, RNN performs time series modeling on these spatial features to obtain the prediction results of meteorological changes.

[0026] ; in, is the assimilation feature at time t, which is a high-dimensional vector that integrates the spatial and temporal variation information of meteorological data. is the satellite cloud image or radar image at time t, RNN and CNN represent recurrent neural network and convolutional neural network respectively.

[0027] ; in, It represents the i-th meteorological feature after meteorological assimilation (such as severe convection area, change in radar reflection intensity, etc.), and n is the number of meteorological features.

[0028] Basic meteorological data and assimilated features are concatenated to form a joint feature vector : ; Through the attention mechanism or weight distribution model, the basic meteorological data and assimilated features are weighted and fused to highlight the features that are more important for risk prediction, and a weight vector sequence is defined , then the joint eigenvector Calculate as follows: ; Among them, the weight vector sequence is obtained by training historical data or dynamically calculated using the attention mechanism. Represent the 1st, 2nd, …, mth weight coefficients respectively.

[0029] Furthermore, existing multi-source data fusion methods usually use static weight coefficients, but this method has poor adaptability to different meteorological scenarios. In order to enhance innovation, this application proposes a dynamic weighting mechanism, which automatically adjusts the weight coefficients of different meteorological data sources (such as temperature, humidity, wind speed, satellite cloud images, radar images, etc.) based on changes in meteorological conditions. The specific method is as follows: Dynamic weighting can be achieved through reinforcement learning or neural networks. For example, reinforcement learning can be used to continuously optimize weight coefficients and automatically adjust the contributions of different meteorological data sources according to changes in meteorological conditions.

[0030] ; in, is the weight coefficient of the i-th meteorological data source at time t, is the characteristic vector of the i-th meteorological data.

[0031] Existing disaster risk assessment models usually only consider meteorological factors and power grid factors, but in actual applications, the disaster risk of the power grid is also affected by multiple factors such as geographic information and equipment health status. Multimodal data fusion is introduced to integrate data of multiple modalities such as meteorological factors, power grid factors, and geographic information to improve the accuracy of risk assessment.

[0032] In step S3, a comprehensive disaster risk assessment model is constructed based on a multi-dimensional analysis of meteorological factors, power grid factors (power grid load and equipment health status), and external factors (such as geographic information). The comprehensive disaster risk assessment model combines meteorological factors and power grid operation status to calculate the comprehensive disaster risk through the following formula: : ; in, represents the risk function of meteorological factors, where T is temperature, H is humidity, W is wind speed, and P is precipitation; represents the risk function of the power grid factors, L is the power grid load, C is the health status of the power grid equipment, is the risk function of geographic information, and its value can be determined by the type of terrain, the location of the equipment, and whether it is a disaster-prone area. G is geographic information, They are the weight coefficients of meteorological factors, power grid factors, and geographic information, which are trained and optimized based on historical data. The specific steps are as follows.

[0033] (a) Impact of temperature T and humidity H on the power grid Temperature and humidity have a direct impact on power grid disasters (such as icing, equipment icing, etc.), especially in cold weather, when the temperature drops, the risk of icing of power facilities will increase. This can be fitted through regression analysis or machine learning models based on meteorological data. Assume that the relationship between temperature T and humidity H is: ; in, They represent the 1st, 2nd, and 3rd meteorological risk weight coefficients, which are weights fitted from historical meteorological data and equipment disaster data, indicating the contribution of temperature and humidity to icing risk. Regression analysis or machine learning models based on meteorological data can be fitted by linear regression or support vector machine.

[0034] (b) Impact of wind speed W and precipitation P on the power grid Wind speed and precipitation are also crucial to the risk assessment of power equipment. Strong winds can cause equipment collapse, tree fall, line damage, and other problems, while precipitation can cause snow and ice accumulation. The risk contribution of wind speed W and precipitation P can be calculated using the following formula: ; in, are the weights of wind speed, precipitation, and the combination of wind speed and precipitation, respectively, which represent the nonlinear effects of wind speed and precipitation on risk. A polynomial regression model is used to fit these relationships.

[0035] (c) Calculation of overall meteorological risk Combining the above meteorological influencing factors, the risk assessment function of meteorological factors is It can be expressed as: ; This part of the model performs a weighted calculation based on the impact of weather changes on power equipment to obtain a numerical value that represents the risk of disaster.

[0036] (d) Influence of grid load L The increase in grid load L usually means that grid equipment operation is more tense, which may lead to an increase in equipment failure rate, especially under extreme weather conditions. The relationship between grid load and disaster risk can be described by the following formula: ; in, is the relationship function between power grid load and disaster risk, These are two coefficients obtained by fitting historical load data and load disaster data. The quadratic term of load L takes into account the nonlinear effect of grid load. As the load increases, the probability of a fault will increase exponentially.

[0037] (e) Impact of the health status of power grid equipment C The health status of power grid equipment C directly affects the disaster resistance of the power grid under severe weather conditions. Equipment that is aging, damaged, or improperly maintained is prone to failure. A linear model is used to describe the impact of equipment health status: ; in, is the relationship function between the health status of power grid equipment and disaster risk, It is the relationship coefficient between the health status of power grid equipment and disaster risk. The health status of equipment C is usually quantified by the maintenance record, age and other information of the equipment. The lower the value, the older the equipment is and the more susceptible it is to weather disasters.

[0038] (f) Comprehensive power grid disaster risk calculation Combining the load and equipment health status, the disaster risk calculation formula for power grid factors is: ; (g) Comprehensive disaster risk calculation Finally, the comprehensive disaster risk of meteorological factors and power grid factors is considered It can be expressed as: ; Through this comprehensive disaster risk assessment model, the risk of the power grid under different meteorological conditions can be assessed and corresponding early warning signals can be generated. If the value exceeds the set threshold (such as 0.75), the system will issue a warning signal to prompt the power operation and maintenance personnel to take countermeasures in advance, such as power outages, strengthening equipment inspections, and increasing staff on duty.

[0039] The present invention adopts a multi-objective optimization (MOO) method to comprehensively consider the three goals of accuracy, timeliness and computational efficiency. Different meteorological scenarios and power grid conditions may have different requirements for these three goals. Therefore, the comprehensive disaster risk assessment model is optimized with multiple objectives and the optimization strategy is adjusted dynamically to ensure that the performance of the comprehensive disaster risk assessment model reaches the best balance under different environments.

[0040] a. Optimization goal: Maximizing accuracy: Minimizing prediction errors ensures that disaster risk assessments are as close as possible to actual disaster situations.

[0041] ; in, represents the accuracy index, represents the actual disaster risk value, represents the predicted disaster risk value; Optimize timeliness: Reduce delays in data processing and forecasting to ensure that warnings are triggered in real time, especially in short-term warnings and rapidly changing meteorological conditions. Timeliness indicators Measured by processing time.

[0042] b. Comprehensive optimization objectives: Combining the above multiple objectives, through weighted summation, the final comprehensive optimization goal is formed: ; in, is the weight coefficient of the accuracy index, The weight coefficient of the timeliness index is dynamically adjusted according to different situations. For example, in extreme weather conditions (such as typhoons and blizzards), the accuracy requirement may be higher, so the weight of the accuracy index is increased, while in normal weather conditions, timeliness may be more important, so the weight of the timeliness index is increased.

[0043] In order to improve the adaptability and intelligence level of multi-objective optimization, the present invention introduces a reinforcement learning (RL) mechanism, which enables the system to automatically adjust the optimization strategy based on real-time feedback, thereby continuously improving performance in a changing meteorological and power grid environment. Reinforcement learning learns the optimal strategy through the interaction between the agent and the environment. In this system, the agent takes actions At (such as adjusting the warning threshold, modifying the weight coefficient, etc.) based on the state St (such as current meteorological conditions, power grid load, equipment health status, etc.), and evaluates the quality of the warning results through the reward function Rt, and adjusts the strategy in real time. The specific implementation process is as follows: 1. Status St: includes meteorological data (such as temperature, humidity, wind speed, etc.), power grid data (such as load, equipment health status, etc.), and historical warning feedback information.

[0044] 2. Action At: Actions taken based on the current state, such as adjusting the weights of certain meteorological features, optimizing the model's computing resource allocation, changing warning thresholds, etc.

[0045] 3. Reward function Rt: The system calculates the error between the actual disaster risk value and the predicted disaster risk value. Conduct reward evaluations.

[0046] ; represents the actual disaster risk value, represents the predicted disaster risk value; The reward function encourages the agent to improve accuracy while reducing computational costs and improving timeliness. Through reinforcement learning, the agent will learn the best optimization strategy in multiple iterations and continuously adjust the weight of the optimization target based on the feedback signal (i.e., error feedback and reward function). For example, if the system performs poorly under certain extreme weather conditions, the system will adjust the weight coefficient and the structure of the comprehensive disaster risk assessment model to improve the prediction accuracy.

[0047] By using historical data and real-time data, the input weights and parameters of the comprehensive disaster risk assessment model can be adjusted through an adaptive feedback mechanism, making each warning more accurate.

[0048] Example 2 An artificial intelligence-based electric power meteorological disaster risk early warning system includes a data collection terminal, a background service terminal, and a display terminal.

[0049] The data collection end is responsible for collecting various meteorological data, geographic information and power grid operation data. Specifically, it includes: meteorological data: temperature, humidity, wind speed, wind direction, rainfall, sunshine, visibility, weather radar, satellite cloud map, lightning monitoring data; power grid operation data: power grid equipment operation status, load, voltage, etc. The data collection end includes the automatic weather station of the Meteorological Bureau, and meteorological sensors and monitoring devices installed at specific locations by the power company.

[0050] Configure and deploy meteorological sensors and monitoring devices to obtain real-time meteorological data through the automatic weather stations of the Meteorological Bureau and equipment installed by power companies.

[0051] The backend server processes and integrates meteorological data, geographic information, and power grid operation data from multiple meteorological data sources to ensure the accuracy and timeliness of the data. The process includes multiple steps such as data cleaning, preprocessing, data fusion, and feature extraction to meet the accuracy requirements of power meteorological warnings.

[0052] The backend server has built-in meteorological data preprocessing module, meteorological data assimilation module, feature splicing module and comprehensive power disaster risk assessment model.

[0053] The meteorological data preprocessing module cleans and preprocesses the basic meteorological data including temperature, humidity, wind speed and precipitation to obtain the cleaned basic meteorological data set.

[0054] The meteorological data assimilation module models meteorological changes and obtains assimilated features through the combination of convolutional neural network (CNN) and recurrent neural network (RNN).

[0055] The feature stitching module fuses basic meteorological data with assimilated features.

[0056] The risks of the power grid under different meteorological conditions are evaluated through a comprehensive disaster risk assessment model, and corresponding early warning signals are generated.

[0057] The display terminal provides an interface for interaction with users, intuitively displaying relevant data and early warning signals. The display terminal has the following functions: (a) Real-time display of the operating status and location of power equipment, including power outage, emergency repair, and power restoration status.

[0058] (b) Provide sound and information warning reminders to ensure that operation and maintenance personnel receive warning signals as soon as possible.

[0059] (c) Support user query and interaction functions to obtain detailed meteorological and power equipment information.

[0060] The present invention is not limited to what is described in the specification and implementation modes, and therefore additional advantages and modifications can be easily realized by those skilled in the art. Therefore, without departing from the spirit and scope of the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details, representative devices, and illustrative examples shown and described herein.

Claims

1. An artificial intelligence-based power meteorological disaster risk early warning method, characterized in that: Here are the steps: Step S1: Collect various meteorological data, geographic information and power grid operation data; Step S2: Clean and preprocess the basic meteorological data including temperature, humidity, wind speed, and precipitation to obtain a cleaned basic meteorological data set; based on satellite cloud images or radar image data, model the meteorological changes through a neural network model to obtain assimilated features, and concatenate the basic meteorological data and the assimilated features to form a joint feature vector; Step S3: construct a comprehensive disaster risk assessment model based on meteorological factors, power grid factors and geographic information, train and optimize the comprehensive disaster risk assessment model, and use the optimized comprehensive disaster risk assessment model to conduct real-time power meteorological disaster risk warning; The comprehensive disaster risk assessment model is expressed as: ； in, is the comprehensive disaster risk, T is temperature, H is humidity, W is wind speed, P is precipitation; L is grid load, C is grid equipment health status, G is geographic information, are the weight coefficients of meteorological factors, power grid factors, and geographic information, respectively. Respectively represent the 1st, 2nd, and 3rd meteorological risk weight coefficients, are the weights of wind speed, precipitation, and the combination of wind speed and precipitation, These are two coefficients obtained by fitting historical load data and load disaster data; It is the relationship coefficient between the health status of power grid equipment and disaster risk.

2. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 1 is characterized in that: In step S2, the basic meteorological data including temperature, humidity, wind speed and precipitation are cleaned by the Kalman filter algorithm.

3. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 1 is characterized in that: In step S2, the meteorological changes are modeled by combining the convolutional neural network with the recurrent neural network to obtain the assimilated features.

4. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 3 is characterized in that: In step S2, the extraction process of assimilation features is expressed as: ； in, is the assimilation characteristic at time t, is the satellite cloud image or radar image at time t, is the total time, RNN and CNN represent recurrent neural network and convolutional neural network respectively; ； in, represents the i-th meteorological feature after meteorological assimilation, and n is the number of meteorological features.

5. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 4 is characterized in that: Joint eigenvector It is expressed as: ； Through the attention mechanism or weight distribution model, the basic meteorological data and assimilated features are weighted and fused to define a weight vector sequence. , then the joint eigenvector Calculate as follows: ； Among them, the weight vector sequence is obtained by training historical data or dynamically calculated using the attention mechanism. Represent the 1st, 2nd, …, mth weight coefficients respectively.

6. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 5 is characterized in that: For the joint feature vector, dynamic weighting is achieved through reinforcement learning or neural network to continuously optimize the weight coefficient.

7. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 1 is characterized in that: In step S3, a multi-objective optimization method is used to comprehensively consider the three objectives of accuracy, timeliness and computational efficiency, perform multi-objective optimization on the comprehensive disaster risk assessment model and dynamically adjust the optimization strategy.

8. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 7 is characterized in that: Combining multiple objectives, through weighted summation, the comprehensive optimization objective of the comprehensive disaster risk assessment model is formed: ； in, is the weight coefficient of the accuracy index, is the weight coefficient of the timeliness index, represents the accuracy index, Indicates timeliness indicator.

9. The artificial intelligence-based electric power meteorological disaster risk early warning method according to claim 8 is characterized in that: In the multi-objective optimization process, a reinforcement learning mechanism is introduced to automatically adjust the optimization strategy with real-time feedback.

10. An artificial intelligence-based electric power meteorological disaster risk early warning system, used to implement the artificial intelligence-based electric power meteorological disaster risk early warning method according to any one of claims 1 to 9, characterized in that: Including data collection terminal, backend service terminal and display terminal; The data collection terminal is responsible for collecting various meteorological data, geographic information and power grid operation data; The backend server processes and integrates meteorological data, geographic information and power grid operation data from multiple meteorological data sources; the backend server has built-in meteorological data preprocessing module, meteorological data assimilation module, feature splicing module and comprehensive power disaster risk assessment model; The meteorological data preprocessing module cleans and preprocesses the basic meteorological data including temperature, humidity, wind speed, and precipitation to obtain the cleaned basic meteorological data set; The meteorological data assimilation module models meteorological changes and obtains assimilated features through the combination of convolutional neural networks and recurrent neural networks; The feature stitching module fuses basic meteorological data with assimilated features; Evaluate the risk of power grid under different meteorological conditions through comprehensive disaster risk assessment model and generate corresponding early warning signals; The display terminal provides an interface for interacting with users, intuitively displaying relevant data and early warning signals.

Images