Electric power disaster damage development trend prediction method based on typhoon scene
By combining smart grid, meteorological station and satellite remote sensing data, training and updating the power disaster loss development trend prediction model, the problem of low accuracy of traditional prediction methods is solved, and more accurate power disaster loss prediction is achieved, and disaster prevention is prevented by the power system. Disaster reduction provides strong support.
Patent Information
- Application Number
- CN202411001681.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-12-03
AI Technical Summary
Traditional power disaster loss development trend prediction methods rely on manual data collection, resulting in low prediction accuracy and inability to effectively deal with the damage to power facilities caused by typhoons.
Using the power disaster loss development trend prediction method based on typhoon scenarios, the power disaster loss development trend prediction model is carried out by obtaining smart grid, meteorological stations and satellite remote sensing data, and the power disaster loss development trend prediction model is used to carry out the power disaster loss development development. Trend forecast.
It improves the accuracy of forecasting the development trend of power disaster loss, can more accurately predict the damage probability, affected areas and recovery time of power facilities, and provides more effective disaster prevention, mitigation and emergency response support.
Smart Images

Figure CN119066607A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of electric power technology, and in particular to a method, device, computer equipment, computer-readable storage medium and computer program product for predicting the development trend of electric power disaster losses based on typhoon scenarios. Background Art
[0002] As global climate change intensifies, extreme weather events, especially typhoons, have become more frequent and intense, posing a serious threat to power infrastructure. The strong winds, heavy rains and floods brought by typhoons can not only directly damage power facilities such as transmission towers, power lines, substations, etc., but may also indirectly cause secondary disasters such as landslides and tree collapses, further aggravating the damage to the power system. These disasters will not only cause large-scale power outages, affecting residents' lives and industrial production, but also bring high repair costs and long recovery work. Therefore, how to accurately predict the development trend of power disaster losses has become an important research direction.
[0003] Traditional technology usually involves manually collecting relevant data, so that experts can use this data to predict the development trend of power disasters; however, this method mainly relies on subjective judgment, resulting in low accuracy in the prediction of the development trend of power disasters. Summary of the invention
[0004] Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product for predicting the development trend of power disasters based on typhoon scenarios, which can improve the accuracy of the prediction of the development trend of power disasters, in response to the above-mentioned technical problems.
[0005] In a first aspect, the present application provides a method for predicting the development trend of power disaster losses based on typhoon scenarios. The method comprises:
[0006] Acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data;
[0007] Performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data;
[0008] Using the data features of the historical multi-source data, the power disaster loss development trend prediction model to be trained is trained to obtain a trained power disaster loss development trend prediction model;
[0009] Performing model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model;
[0010] The data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing.
[0011] In one embodiment, the data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction result of the object to be predicted, including:
[0012] Acquire current multi-source data corresponding to the object to be predicted;
[0013] Performing feature engineering processing on the current multi-source data to obtain data features of the current multi-source data;
[0014] The data features of the current multi-source data are input into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction result of the object to be predicted; the power disaster loss development trend prediction result includes the damage probability of power facilities, the affected area, the degree of damage and the recovery time.
[0015] In one embodiment, performing feature engineering on the historical multi-source data to obtain data features of the historical multi-source data includes:
[0016] Performing feature selection processing, feature creation processing, feature encoding processing and feature interaction processing on the historical multi-source data to obtain data features of the historical multi-source data;
[0017] The performing feature engineering processing on the current multi-source data to obtain data features of the current multi-source data includes:
[0018] The current multi-source data is subjected to feature selection processing, feature creation processing, feature encoding processing and feature interaction processing to obtain data features of the current multi-source data.
[0019] In one embodiment, the method of using the data features of the historical multi-source data to train the power disaster loss development trend prediction model to be trained to obtain the trained power disaster loss development trend prediction model includes:
[0020] Selecting a target prediction model associated with the prediction of power disaster loss development trend from the candidate prediction models as the power disaster loss development trend prediction model to be trained;
[0021] Constructing a training set and a test set of the power disaster loss development trend prediction model to be trained according to the data characteristics of the historical multi-source data;
[0022] The training set and the test set are used to train the power disaster loss development trend prediction model to be trained to obtain the trained power disaster loss development trend prediction model.
[0023] In one embodiment, the updating process of the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model includes:
[0024] Performing performance identification on the trained power disaster loss development trend prediction model to obtain a performance identification result of the trained power disaster loss development trend prediction model;
[0025] According to the performance identification result, the trained power disaster loss development trend prediction model is subjected to multi-dimensional model updating processing to obtain the target power disaster loss development trend prediction model.
[0026] In one embodiment, the acquiring of historical multi-source data corresponding to the object to be predicted includes:
[0027] Acquire original multi-source data corresponding to the object to be predicted;
[0028] The original multi-source data is subjected to data preprocessing to obtain the historical multi-source data; the data preprocessing includes data cleaning processing, data conversion processing, data integration processing, data normalization processing and data sampling processing.
[0029] In a second aspect, the present application also provides a device for predicting the development trend of power disaster losses based on typhoon scenarios. The device comprises:
[0030] A data acquisition module, used to acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data;
[0031] A data processing module, used for performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data;
[0032] A model training module, used to train the power disaster loss development trend prediction model to be trained by utilizing the data features of the historical multi-source data, and obtain a trained power disaster loss development trend prediction model;
[0033] A model updating module is used to perform model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model;
[0034] A feature input module is used to input the data features of the current multi-source data corresponding to the object to be predicted into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing.
[0035] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:
[0036] Acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data;
[0037] Performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data;
[0038] Using the data features of the historical multi-source data, the power disaster loss development trend prediction model to be trained is trained to obtain a trained power disaster loss development trend prediction model;
[0039] Performing model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model;
[0040] The data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing.
[0041] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:
[0042] Acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data;
[0043] Performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data;
[0044] Using the data features of the historical multi-source data, the power disaster loss development trend prediction model to be trained is trained to obtain a trained power disaster loss development trend prediction model;
[0045] Performing model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model;
[0046] The data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing.
[0047] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:
[0048] Acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data;
[0049] Performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data;
[0050] Using the data features of the historical multi-source data, the power disaster loss development trend prediction model to be trained is trained to obtain a trained power disaster loss development trend prediction model;
[0051] Performing model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model;
[0052] The data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing.
[0053] The above-mentioned method, device, computer equipment, computer-readable storage medium and computer program product for predicting the development trend of power disaster losses based on typhoon scenarios obtain historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes historical data of smart grid, historical data of meteorological station and historical data of satellite remote sensing; feature engineering processing is performed on the historical multi-source data to obtain data features of the historical multi-source data; the data features of the historical multi-source data are used to train the power disaster loss development trend prediction model to be trained to obtain a trained power disaster loss development trend prediction model; model update processing is performed on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model; the data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to predict the development trend of power disaster losses to obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing. This solution integrates historical multi-source data to train and update the power disaster loss development trend prediction model, which is conducive to improving the prediction accuracy of the power disaster loss development trend prediction model, thereby facilitating more accurate prediction of the power disaster loss development trend of the predicted object in typhoon scenarios, and providing strong support for disaster prevention, mitigation and emergency response of the power system. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0055] Figure 1 A schematic diagram of a flow chart of a method for predicting power disaster loss development trend based on a typhoon scenario in one embodiment;
[0056] Figure 2 A flowchart of the steps of feature input in one embodiment;
[0057] Figure 3 A flowchart of the steps of model training in one embodiment;
[0058] Figure 4 It is a structural block diagram of a device for predicting the development trend of power disaster losses based on a typhoon scenario in one embodiment;
[0059] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0061] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0062] In an exemplary embodiment, Figure 1 As shown, a method for predicting the development trend of power disaster losses based on typhoon scenarios is provided. This embodiment takes the application of this method to a terminal as an example; it can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be but is not limited to various personal computers, laptops, smart phones, tablet computers, etc.; the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. In this embodiment, the method includes the following steps:
[0063] Step S101, obtaining historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data.
[0064] Step S102, performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data.
[0065] Step S103, using the data features of the historical multi-source data, the power disaster loss development trend prediction model to be trained is trained to obtain a trained power disaster loss development trend prediction model.
[0066] Step S104, performing model update processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model.
[0067] Step S105, input the data features of the current multi-source data corresponding to the object to be predicted into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes the current data of the smart grid, the current data of the meteorological station and the current data of satellite remote sensing.
[0068] The object to be predicted may be a specific area or power facility for which a prediction of the development trend of power disaster losses is required.
[0069] The historical multi-source data may be a collection of historical data from various sources related to the object to be predicted. For example, the historical multi-source data may include smart grid operation data, weather station observation data, and satellite remote sensing image data over a period of time.
[0070] The smart grid historical data may be operation data recorded by the smart grid system over a period of time in the past. For example, the smart grid historical data may include information such as power load, equipment status and / or fault records.
[0071] The historical data of a weather station may be weather-related data observed by the weather station over a period of time in the past. For example, the historical data of a weather station may include historical records of meteorological elements such as wind speed, rainfall and / or air pressure.
[0072] The satellite remote sensing historical data may be images or data acquired by the satellite during earth observation over a period of time in the past. For example, the satellite remote sensing historical data may include information such as cloud images, land cover and / or vegetation index.
[0073] Among them, feature engineering processing can be the process of processing and transforming raw data to extract useful feature information. For example, feature engineering processing can include steps such as data cleaning, data standardization, data feature extraction and data feature selection.
[0074] The data features may be attributes or indicators extracted from the original data that can represent the essential characteristics of the data. For example, the data features may include statistical features, time series features, and / or spatial features.
[0075] Among them, the power disaster loss development trend prediction model can be a model used to predict the development trend of disaster losses in the power system under typhoon scenarios. For example, the power disaster loss development trend prediction model can be a prediction model based on machine learning or deep learning, such as a neural network model.
[0076] The model update process may be a process of optimizing or adjusting a trained model to improve the prediction performance of the model. For example, the model update process may include parameter fine-tuning, transfer learning, and / or online learning.
[0077] Among them, the target power disaster loss development trend prediction model can be the final model used for actual prediction tasks after training and updating. For example, the target power disaster loss development trend prediction model can be an optimization model that can accurately predict the power disaster loss development trend.
[0078] The current multi-source data may be a real-time data set from multiple sources related to the object to be predicted. For example, the current multi-source data may include real-time smart grid operation data, weather station observation data, and satellite remote sensing image data.
[0079] Among them, the prediction results of the development trend of power disaster losses can be the forecast information output by the model about the development of disaster losses predicted to occur in the power system in a certain period of time in the future under the typhoon scenario. For example, the prediction results of the development trend of power disaster losses can include the expected affected areas, fault types and / or changing trends in the degree of losses.
[0080] Optionally, the terminal obtains historical multi-source data corresponding to the object to be predicted, including smart grid historical data, meteorological station historical data and satellite remote sensing historical data; performs feature engineering processing on the historical multi-source data to extract valuable data features; uses the processed data features to train a pre-designed power disaster loss development trend prediction model to obtain a trained power disaster loss development trend prediction model; further updates and optimizes the trained power disaster loss development trend prediction model to improve its prediction performance, thereby obtaining a target power disaster loss development trend prediction model; inputs the features of the current multi-source data corresponding to the object to be predicted (including current smart grid data, current meteorological station data and current satellite remote sensing data) into the target power disaster loss development trend prediction model, performs power disaster loss development trend prediction, and obtains the power disaster loss development trend prediction result of the object to be predicted.
[0081] In the above-mentioned method for predicting the development trend of power disaster losses based on typhoon scenarios, historical multi-source data corresponding to the object to be predicted is obtained; the historical multi-source data at least includes historical data of smart grid, historical data of meteorological station and historical data of satellite remote sensing; feature engineering processing is performed on the historical multi-source data to obtain data features of the historical multi-source data; the power disaster loss development trend prediction model to be trained is trained using the data features of the historical multi-source data to obtain a trained power disaster loss development trend prediction model; the trained power disaster loss development trend prediction model is updated to obtain a target power disaster loss development trend prediction model; the data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction to obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing. This solution integrates historical multi-source data to train and update the power disaster loss development trend prediction model, which is conducive to improving the prediction accuracy of the power disaster loss development trend prediction model, thereby facilitating more accurate prediction of the power disaster loss development trend of the predicted object in typhoon scenarios, and providing strong support for disaster prevention, mitigation and emergency response of the power system.
[0082] In an exemplary embodiment, referring to Figure 2 , the data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and the power disaster loss development trend prediction result of the object to be predicted is obtained, which specifically includes the following contents:
[0083] Step S201, obtaining current multi-source data corresponding to the object to be predicted;
[0084] Step S202, performing feature engineering processing on the current multi-source data to obtain data features of the current multi-source data;
[0085] Step S203, input the data features of the current multi-source data into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction results of the object to be predicted; the power disaster loss development trend prediction results include the damage probability of power facilities, the affected area, the degree of damage and the recovery time.
[0086] The damage probability of power facilities may be the possibility of power facilities being damaged by a typhoon as calculated by the target power disaster loss development trend prediction model. For example, the damage probability of power facilities may be that the probability of a certain transmission line failing within the next 24 hours is 80%.
[0087] The affected area may be a geographical area predicted by the target power disaster development trend prediction model to be affected by the typhoon and cause power supply problems.
[0088] The degree of damage may be the severity of damage to power facilities predicted by the target power disaster loss development trend prediction model. For example, the degree of damage may be divided into three levels: slight, moderate and severe, or a specific numerical value may be used to represent the degree of loss.
[0089] The restoration time may be the time required from the damage of power facilities to the restoration of normal power supply as predicted by the target power disaster development trend prediction model.
[0090] Optionally, the terminal obtains current multi-source data corresponding to the object to be predicted, including current data of the smart grid, current data of the meteorological station, and current data of satellite remote sensing; performs feature engineering processing on these current multi-source data to extract valuable data features, which may include key indicators such as grid load conditions, typhoon intensity, and rainfall; inputs the processed data features into a target power disaster loss development trend prediction model, and performs power disaster loss development trend prediction based on the input data features through the target power disaster loss development trend prediction model to obtain the power disaster loss development trend prediction result of the object to be predicted, including key information such as the damage probability of power facilities, the affected area, the degree of damage and the recovery time.
[0091] The technical solution provided in this embodiment obtains current multi-source data and performs feature engineering processing, and inputs the processed data features into the target power disaster loss development trend prediction model, which is conducive to making full use of data features, thereby facilitating the generation of more comprehensive and accurate power disaster loss development trend prediction results.
[0092] In an exemplary embodiment, feature engineering processing is performed on historical multi-source data to obtain data features of the historical multi-source data, which specifically includes the following contents: feature selection processing, feature creation processing, feature encoding processing and feature interaction processing are performed on the historical multi-source data to obtain data features of the historical multi-source data; feature engineering processing is performed on current multi-source data to obtain data features of the current multi-source data, which specifically includes the following contents: feature selection processing, feature creation processing, feature encoding processing and feature interaction processing are performed on the current multi-source data to obtain data features of the current multi-source data.
[0093] Among them, feature selection processing can be the processing of selecting the most relevant and important features from the original data (historical multi-source data). For example, feature selection processing can use correlation analysis, variance analysis or machine learning algorithms to screen out the features that have the greatest impact on power loss prediction, such as wind speed, rainfall or grid load.
[0094] Among them, the feature creation process can be a process of creating new and more meaningful features based on the original data. For example, the feature creation process can include calculating the combination of multiple original features, generating time series features, or constructing spatial features.
[0095] The feature encoding process may be a process of converting non-numeric features into numerical features. For example, the feature encoding process may include one-hot encoding, label encoding, or frequency encoding of categorical variables.
[0096] Among them, feature interaction processing can be processing that utilizes the interaction between different features. For example, feature interaction processing can include creating products, ratios or other combinations of features, such as calculating the product of wind speed and rainfall to represent the comprehensive meteorological impact, or calculating the ratio of grid load to temperature to reflect the relationship between electricity demand and climate.
[0097] Optionally, the terminal performs the same feature engineering process on the historical multi-source data and the current multi-source data. For example, the terminal performs feature selection processing to select the most relevant features through statistical analysis or machine learning algorithms; performs feature creation processing to construct new composite features based on the original features; performs feature encoding processing to convert non-numeric features into numerical features; performs feature interaction processing to identify the interaction relationship between features; through this series of processing, the original multi-source data (historical multi-source data or current multi-source data) is converted into a data feature set with more predictive value (data features of historical multi-source data or data features of current multi-source data).
[0098] The technical solution provided in this embodiment is conducive to extracting and constructing more valuable data features by performing feature engineering processing on historical multi-source data and current multi-source data, including feature selection processing, feature creation processing, feature encoding processing and feature interaction processing, thereby helping to enhance the model's ability to understand the power system and environmental factors and improve the accuracy of the prediction of power disaster development trends.
[0099] In an exemplary embodiment, referring to Figure 3 , using the data characteristics of historical multi-source data, the power disaster loss development trend prediction model to be trained is trained to obtain the trained power disaster loss development trend prediction model, which specifically includes the following contents:
[0100] Step S301, selecting a target prediction model associated with the prediction of power disaster loss development trend from the candidate prediction models as the power disaster loss development trend prediction model to be trained;
[0101] Step S302, constructing a training set and a test set of a power disaster loss development trend prediction model to be trained according to the data characteristics of historical multi-source data;
[0102] Step S303, using the training set and the test set, the power disaster loss development trend prediction model to be trained is trained to obtain a trained power disaster loss development trend prediction model.
[0103] Among them, the candidate prediction models can be a series of machine learning or statistical models that can be used to predict the development trend of power disaster losses. For example, the candidate prediction models can include linear regression, decision tree, random forest, support vector machine, neural network and other models.
[0104] Among them, the target prediction model can be a model selected from the candidate prediction models that is most suitable for the task of predicting the development trend of power disaster losses. For example, the target prediction model can be a long short-term memory network (LSTM) or a gradient boosting decision tree (GBDT) model that performs best after preliminary evaluation.
[0105] The training set may be a subset of historical multi-source data used to train the model. For example, the training set may include a portion of historical data over a period of time and its corresponding power disaster development trend labels.
[0106] The test set may be a subset of historical multi-source data used to evaluate model performance. For example, the test set may include another part of historical data over a period of time in the past and its corresponding power disaster development trend labels.
[0107] Optionally, the terminal selects a target prediction model suitable for the power disaster loss development trend prediction task from a pre-prepared candidate prediction model library as the power disaster loss development trend prediction model to be trained, wherein the selection criteria may include factors such as the model's prediction ability, computational efficiency, and interpretability; constructs a training set and a test set based on data characteristics of historical multi-source data; uses the training set to train the power disaster loss development trend prediction model to be trained, and uses the test set to evaluate the performance of the power disaster loss development trend prediction model to be trained, and obtains the final trained power disaster loss development trend prediction model through multiple rounds of iterative optimization.
[0108] The technical solution provided in this embodiment, by selecting a suitable target prediction model from the candidate prediction models and using the data features of historical multi-source data to construct a training set and a test set for model training, is conducive to obtaining the model that is most suitable for the task of predicting the development trend of power disaster losses, thereby helping to improve the accuracy and generalization ability of the power disaster loss development trend prediction model.
[0109] In an exemplary embodiment, a model update process is performed on a trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model, which specifically includes the following contents: performance identification is performed on the trained power disaster loss development trend prediction model to obtain a performance identification result of the trained power disaster loss development trend prediction model; based on the performance identification result, a multi-dimensional model update process is performed on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model.
[0110] Among them, performance identification can be a process of comprehensively evaluating the trained power disaster loss development trend prediction model. For example, performance identification can include the calculation and analysis of indicators such as the model's prediction accuracy, recall rate, and root mean square error.
[0111] Among them, the performance identification result can be a quantitative evaluation of the performance of the trained power disaster loss development trend prediction model. For example, the performance identification result can include the prediction performance data of the model under different time scales, different geographical regions, and different weather conditions.
[0112] Among them, the multi-dimensional model update processing can be a process of comprehensively optimizing the model based on the performance identification results. For example, the multi-dimensional model update processing can include update operations in multiple aspects such as model structure adjustment, feature engineering optimization, hyperparameter tuning and / or incremental learning.
[0113] Optionally, the terminal performs a comprehensive performance identification on the trained power disaster loss development trend prediction model, which may include, for example, using a validation data set to evaluate the prediction accuracy, stability and generalization ability of the model in different scenarios, and obtain the performance identification results of the trained power disaster loss development trend prediction model. The performance identification results may include specific values of each evaluation indicator, as well as an analysis of the performance of the model under different conditions. Based on the performance identification results, the trained power disaster loss development trend prediction model is updated in multiple dimensions, such as including model structure adjustment, feature engineering optimization, hyperparameter tuning and other aspects, so as to solve problems found in performance identification and improve the overall performance of the model, thereby obtaining the target power disaster loss development trend prediction model.
[0114] The technical solution provided in this embodiment, by performing performance identification on the trained model and performing multi-dimensional update processing based on the performance identification results, is conducive to comprehensively evaluating the model performance and optimizing the model in a targeted manner, thereby facilitating improving the accuracy of the prediction of the development trend of power disaster losses.
[0115] In an exemplary embodiment, obtaining historical multi-source data corresponding to the object to be predicted specifically includes the following contents: obtaining original multi-source data corresponding to the object to be predicted; performing data preprocessing on the original multi-source data to obtain historical multi-source data; data preprocessing includes data cleaning processing, data conversion processing, data integration processing, data normalization processing and data sampling processing.
[0116] The original multi-source data may be unprocessed original data collected from different sources, for example, the original multi-source data may include smart grid original data, weather station original data, and satellite remote sensing original data.
[0117] Among them, data cleaning processing can be the processing of error correction and outlier processing of original multi-source data. For example, data cleaning processing can include operations such as removing duplicate data, filling missing values, and correcting erroneous values.
[0118] The data conversion process may be a process of converting data into a format or scale suitable for analysis. For example, the data conversion process may include operations such as data type conversion, unit unification, and time series alignment.
[0119] Among them, data integration processing can be the processing of merging data from different sources into a consistent data set. For example, data integration processing can include operations such as merging records from different databases, resolving data conflicts, and unifying data standards.
[0120] Among them, data normalization processing can be a process of converting data into a specific range to make features of different scales comparable.
[0121] Among them, data sampling processing can be a process of selecting representative data from the data to reduce the computational burden of data processing.
[0122] Optionally, the terminal obtains original multi-source data corresponding to the object to be predicted; performs a series of data preprocessing on the original multi-source data, including data cleaning processing, data conversion processing, data integration processing, data normalization processing and data sampling processing, so as to obtain historical multi-source data.
[0123] The technical solution provided in this embodiment is beneficial to improving data quality by preprocessing the original multi-source data, thereby providing high-quality and standardized input data for the power disaster loss development trend prediction model to be trained, thereby improving the training effect and prediction accuracy of the model.
[0124] The following is an application example to illustrate the power disaster loss development trend prediction method based on typhoon scenarios provided by this application. This application example uses the method applied to a terminal as an example.
[0125] This application example deeply integrates deep learning algorithms into the priority decision-making process of power line repair. By building an intelligent model that can learn and predict autonomously, it realizes the accurate prediction of power line failures and the dynamic optimization of the repair sequence. Compared with traditional manual decision-making or fixed rule-based methods, this solution can more effectively process massive amounts of real-time data, including meteorological information, geographic information, power grid status data, etc., so that when faced with large-scale damage to the power grid caused by extreme weather such as typhoons, it can quickly identify key lines and guide the optimal allocation of resources, greatly improving the efficiency of repair work and the speed of power grid recovery. In addition, the solution also has the ability of self-learning and iteration, and can continuously optimize the decision-making model based on the feedback information of each repair, further improving the accuracy and practicality of the decision.
[0126] This application example integrates real-time information collected by smart grid sensors, meteorological data, satellite remote sensing and IoT devices, and uses data preprocessing, feature engineering, machine learning model training, and real-time monitoring and prediction technology to accurately predict potential disaster losses of power facilities under extreme weather conditions such as typhoons. The core of the system is to use supervised learning algorithms, such as random forests or gradient boosting trees, to establish a disaster loss prediction model, combining the physical characteristics of power facilities, historical disaster loss records, topography, weather forecasts and other factors to generate disaster loss development trend information, thereby helping power departments and emergency management agencies to take preventive measures in advance, reduce disaster risks, and ensure the stability and security of power supply. The specific technical implementation plan is as follows:
[0127] Step 1: Data collection and preprocessing.
[0128] By collecting multiple data such as smart grid, weather station, satellite remote sensing, etc., and cleaning, converting and integrating them, a solid foundation is laid for subsequent analysis. Ensuring the quality and applicability of data is the prerequisite for building a reliable prediction model.
[0129] (1) Data collection:
[0130] Data collection is the process of collecting raw data related to the project. The data sources mainly include:
[0131] Smart grid terminals: Terminals contain smart meters, sensors or other IoT devices that continuously monitor the operating status of power facilities, such as current, voltage, power factor, temperature, etc., as well as environmental factors that may affect the stability of the facilities, such as humidity and wind speed.
[0132] Weather station data: Meteorological data obtained from weather stations, including typhoon path forecasts, wind speed, rainfall, air pressure, temperature, etc., which help assess the impact of weather conditions on power facilities.
[0133] Satellite remote sensing data: Use satellite images to identify and monitor environmental changes around power facilities, such as land use type (urban, forest, water), terrain features (mountains, rivers), vegetation cover, etc. This information is critical to understanding the vulnerability of power facilities.
[0134] (2) Data preprocessing:
[0135] Data preprocessing is the process of converting raw data into a clean, standardized form that can be used for analysis. This typically involves the following steps:
[0136] Data cleaning: Check and handle errors, inconsistencies, and missing values in the data. For example, outliers may be caused by sensor failures and need to be identified and corrected or removed.
[0137] Data conversion: Converting data into a uniform format, such as converting timestamps to a uniform time zone, or encoding non-numeric data into a numeric representation.
[0138] Data integration: Combining data from different sources to create a unified data set. This may involve resolving data conflicts and ensuring that all data uses the same units and scale.
[0139] Data normalization: Scale the data to the same range, such as between 0 and 1, to eliminate the impact of dimension and make comparisons between different features fairer.
[0140] Data Sampling: If the dataset is very large, you may want to perform sampling to reduce computational cost while keeping the data representative.
[0141] Feature selection: Identify which data attributes (features) are most critical to the model’s predictive power and remove irrelevant or redundant features to improve model efficiency and accuracy.
[0142] Step 2: Feature Engineering.
[0143] Through feature selection, creation and coding, we extract information that is crucial to predicting power losses, laying a solid foundation for model training. This step builds a bridge between data collection and model building, and its refined operation is crucial to improving the performance of the prediction model.
[0144] Feature selection: Select the most influential features from the original dataset and remove those with low correlation with the target variable to reduce the complexity of the model and the risk of overfitting.
[0145] Feature creation: Creating new features based on existing features. These new features may capture relationships or patterns that are not obvious in the original data. For example, trends, periodicity, or lag features can be extracted from time series data.
[0146] Feature encoding: Convert non-numerical features (such as categorical data) into numerical form to facilitate model processing. Encoding methods include One-Hot Encoding, Label Encoding, or Embedding.
[0147] Feature Interactions: Explore the interactions between features and create cross-features to capture more complex patterns. For example, combining features of geographic location and weather conditions may have more predictive value than considering either one alone.
[0148] Step 3: Model building and training.
[0149] Select appropriate machine learning algorithms, such as random forests or neural networks, use the divided training sets to train the model, and evaluate the generalization ability of the model through cross-validation and test sets to ensure that the model can accurately predict the development trend of power disaster losses.
[0150] (1) Model selection:
[0151] Algorithm selection: Based on the nature of the problem and the characteristics of the data, choose an appropriate machine learning algorithm. For regression or classification tasks such as predicting power losses, you can consider using algorithms such as random forests, gradient boosting trees, support vector machines (SVMs), neural networks, etc.
[0152] Model evaluation criteria: Define metrics for evaluating model performance, such as mean square error (MSE), root mean square error (RMSE), R² score (coefficient of determination), etc. for regression tasks; precision, recall, etc. for classification tasks.
[0153] (2) Divide the data set:
[0154] Training set and test set: The preprocessed data set is divided into training set and test set. Usually 70%-80% of the data is used for training, and the rest is used to test the generalization ability of the model.
[0155] Cross-validation: The k-fold cross-validation method is used to divide the dataset into k (k represents a numerical value) subsets, and k-1 subsets are used in turn to train the model. The remaining subsets are used as validation sets to evaluate the stability of the model and avoid overfitting.
[0156] (3) Feature Scaling:
[0157] Standardization / Normalization: For some algorithms (such as SVM and neural networks), feature scaling can improve the convergence speed and predictive performance of the model. Standardization (scaling the features to a mean of 0 and a standard deviation of 1) and normalization (scaling the features to a range of 0 to 1) are two commonly used methods.
[0158] (4) Model training:
[0159] Hyperparameter tuning: Use grid search, random search, or Bayesian optimization to find the best hyperparameter combination for the model to optimize the model’s predictive ability.
[0160] Training process: Using the training set data, adjust the model parameters so that the model can learn the patterns and relationships in the data. This process may require multiple iterations until the model achieves satisfactory performance on the training set.
[0161] Regularization: To prevent overfitting, L1 or L2 regularization can be used to limit the complexity of the model by adding a penalty term.
[0162] (5) Model evaluation:
[0163] Performance Evaluation: Evaluate the performance of the model on the test set to ensure that the model not only performs well on the training data but also has good generalization ability on unseen data.
[0164] Error Analysis: Examine the prediction error of the model to identify model weaknesses and biases that may require further adjustments to the model or feature selection.
[0165] (6) Model verification:
[0166] Verify model stability: Use the results of cross-validation to check whether the performance of the model on different data subsets is consistent and evaluate the stability and robustness of the model.
[0167] Model interpretability: For some algorithms (such as decision trees and random forests), it is possible to analyze the importance of features and understand which features have the greatest impact on the prediction results, which is very important for business decision-making and building trust in the model.
[0168] Step 4: Model evaluation and optimization.
[0169] An independent test data set was used to verify its predictive ability, and indicators such as mean square error and accuracy were used to measure the effectiveness of the model. In response to the deficiencies found in the evaluation, strategies such as parameter tuning, iterative improvements in feature engineering, and regularization were implemented to enhance the generalization ability and predictive accuracy of the model, ensuring that it can reliably predict the development trend of power disaster losses in practical applications.
[0170] (1) Model evaluation:
[0171] Performance indicator selection: Choose appropriate performance indicators based on the nature of the prediction task. For regression tasks, the indicators include mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), and coefficient of determination (R² Score). For classification tasks, you may focus on accuracy, precision, and recall.
[0172] Use the test set to evaluate: The model is applied to the test set that has not been used in the previous training to evaluate the generalization ability of the model. This is a key step to verify the performance of the model on unknown data.
[0173] Error Analysis: In-depth analysis of prediction errors to identify which types of data points the model has poor prediction performance on, helping to understand the limitations and biases of the model.
[0174] Model interpretability: For some models (such as decision trees and logistic regression), analyze feature importance, understand which features contribute most to the prediction results, and improve the transparency and trust of the model.
[0175] (2) Model optimization:
[0176] Parameter tuning: Based on the results of the initial model evaluation, adjust the model parameters using methods such as grid search, random search, or Bayesian optimization to further improve the model’s predictive performance.
[0177] Feature Engineering Iteration: Based on the issues found in the model evaluation, it may be necessary to revisit the feature selection and engineering process, introduce new features, or modify existing features to improve model performance.
[0178] Ensemble learning: Consider using ensemble learning methods to improve the stability and prediction accuracy of the model by combining the prediction results of multiple models.
[0179] Regularization techniques: If the model shows signs of overfitting, you can increase the regularization parameter, such as L1 or L2 regularization, to reduce the model complexity and prevent overfitting.
[0180] Model ensemble: Consider using multiple models of different types and integrating their predictions through voting or weighted averaging to improve the robustness of the predictions.
[0181] Continuous learning: Establish a continuous learning framework for the model, allowing the model to automatically update when new data is received, maintaining the timeliness and adaptability of the model.
[0182] Step 5: Real-time monitoring and model calling.
[0183] Through real-time monitoring and model calling, it is ensured that the prediction model can be seamlessly integrated into actual operations, and efficient processing and analysis of real-time data can be achieved. By receiving real-time power system status and meteorological data, instant prediction is made and the development trend of power disasters is output. The prediction results are visualized in real time to assist power system managers and emergency response teams in making quick decisions. At the same time, the system integrates an alarm mechanism, which automatically triggers an alarm when the prediction results reach the preset threshold to ensure timely response.
[0184] (1) Real-time data collection and processing:
[0185] Data stream access: Establish an efficient data stream pipeline to obtain real-time power system status and environmental data from multiple sources such as smart grid terminals, weather stations, satellite remote sensing, etc. The data may be transmitted to the real-time monitoring server through API (application programming interface), FTP (file transfer protocol) or message queue.
[0186] Data preprocessing: The real-time monitoring server cleans, formats, and extracts features from the collected data to ensure that the data meets the requirements of the model input. This includes operations such as removing outliers, filling missing data, and converting data formats.
[0187] Feature Engineering: Dynamically update feature sets, adjust feature weights or create new features based on the latest data to reflect the vulnerability and environmental conditions of power facilities in the current environment.
[0188] (2) Model calling and prediction:
[0189] Model call: After preprocessing and feature engineering are completed, the real-time monitoring server calls the deployed prediction model and inputs the current data into the model for analysis and prediction.
[0190] Real-time prediction: The model predicts the potential damage extent, development trend and possible restoration time of power facilities based on the latest power system status and environmental data.
[0191] Result output: The prediction results output by the model include key information such as the probability of damage to power facilities, prediction of the affected area, expected degree of damage and recovery time.
[0192] (3) Result analysis and visualization:
[0193] Result visualization: The forecast results are displayed through a real-time monitoring dashboard, which may include hot spots on a map, time series forecast charts, alarm threshold indicators, etc., to help power system managers intuitively understand the risk status of the power system.
[0194] Alarm triggering: When the prediction result reaches the preset warning threshold, the alarm system is automatically triggered to notify relevant personnel to take emergency measures, such as starting the backup power supply, dispatching the maintenance team, etc.
[0195] (4) Model monitoring and optimization:
[0196] Model optimization: Based on real-time prediction results and actual disaster loss data, regularly evaluate the prediction accuracy and timeliness of the model, and adjust model parameters or retrain the model when necessary to improve prediction accuracy.
[0197] Feedback loop: Establish a feedback loop between the model and actual operations, collect comparisons between prediction results and actual disaster loss data, and use them for continuous learning and optimization of the model to ensure that the model can continuously adapt to new environmental changes and power system needs.
[0198] Among them, the algorithm technologies related to this application example are as follows:
[0199] (1) Machine Learning Algorithms:
[0200] The core of this application example relies on advanced machine learning algorithms, including supervised learning models such as random forests, support vector machines, and gradient boosting trees (XGBoost) to build a precise prediction framework. These algorithms identify disaster loss patterns and development trends by learning historical data of power facilities in typhoon scenarios, providing a scientific basis for disaster warning and response strategies for power systems. Random forests, with their powerful feature selection capabilities and generalization performance, excel in identifying the potential vulnerabilities of power facilities; SVMs are good at finding optimal boundaries in high-dimensional space, and are particularly effective in identifying nonlinear disaster loss patterns; gradient boosting trees significantly improve prediction accuracy and model stability by integrating multiple weak learners, making them an ideal choice for processing large-scale data sets.
[0201] (2) Feature Engineering Technology:
[0202] Feature engineering technology is the cornerstone of the innovative prediction model of this application example. It extracts the most critical information for predicting power disasters from the original data by deriving new features, feature selection and feature encoding. For example, by calculating the power load change rate, wind speed level, terrain characteristics, etc., a comprehensive index reflecting the vulnerability and environmental conditions of power facilities is constructed. Feature selection technology eliminates non-critical features to avoid overfitting of the model, while feature encoding converts categorical data into numerical form so that the model can be processed effectively. This series of operations significantly improves the predictive ability and interpretability of the model, ensuring that the model remains highly accurate and reliable in complex scenarios.
[0203] (3) Data preprocessing technology:
[0204] Data preprocessing technology ensures the quality of model input data and is a prerequisite for the stability and accuracy of the prediction model. This application example uses data cleaning, missing value processing, data standardization, and data integration to effectively purify the data set, reduce noise interference, and make the data suitable for the training of machine learning models. Data cleaning removes outliers and erroneous records, missing value processing fills in gaps through interpolation technology, data standardization scales numerical features to a uniform range to eliminate dimensional effects, and data integration integrates data from different sources to ensure data integrity. These preprocessing steps work together to provide clean, consistent, and high-quality data support for model training, thereby improving the generalization ability and prediction effect of the model.
[0205] (4) Real-time data stream processing:
[0206] Real-time data stream processing technologies, including message queues and stream processing frameworks, are key to achieving instant response of models. By establishing efficient data stream pipelines, streaming data from smart grids and weather stations are received and processed in real time to ensure that the model can quickly capture the latest changes in the power system status. These technologies not only support high-concurrency data ingestion, but also calculate and analyze data in real time, enabling the model to respond instantly to emergencies such as typhoons, provide timely warning information, and provide real-time data support for dynamic management and emergency decision-making of power systems.
[0207] (5) Model evaluation and optimization:
[0208] Model evaluation and optimization are the core links to ensure the long-term effectiveness of the prediction model. This application example uses cross-validation, grid search, Bayesian optimization and other technologies to comprehensively evaluate the performance of the model and tune the model parameters to improve the generalization ability and prediction accuracy of the model. Cross-validation evaluates the stability and prediction ability of the model by splitting the data set into multiple subsets and repeatedly training and testing the model; grid search and Bayesian optimization search for the best combination in the hyperparameter space to improve model performance. The comprehensive application of these technologies ensures that the model can still maintain a high degree of prediction accuracy and reliability when facing different scenarios, providing solid scientific support for disaster prediction and management of power systems.
[0209] Among them, this application example has the following advantages:
[0210] (1) Accurate prediction capability: By integrating machine learning and deep learning algorithms, it is possible to accurately predict the potential damage patterns and development trends of power facilities in typhoon scenarios, providing a powerful tool for disaster warning of power systems.
[0211] (2) Efficient data preprocessing: The use of technologies such as data cleaning, missing value processing, data standardization, and data integration ensures the data quality of the input model, reduces noise interference, makes the data more suitable for the training of machine learning models, and improves the accuracy and stability of predictions.
[0212] (3) Intelligent feature engineering: By deriving new features, feature selection and feature encoding, the most critical information for predicting power losses is extracted from the original data, enhancing the predictive ability and interpretability of the model.
[0213] (4) Model optimization and generalization: The use of cross-validation, grid search, and Bayesian optimization techniques can effectively evaluate and tune model parameters, improve the model's generalization ability and prediction accuracy, and ensure that the model can maintain a high level of prediction performance in different scenarios.
[0214] The technical solution provided by this application example, by integrating advanced machine learning and deep learning algorithms, can achieve accurate prediction of potential disaster losses of power facilities in extreme weather such as typhoons. Combined with real-time data stream processing and optimized feature engineering, it ensures the real-time response capability and high prediction accuracy of the prediction model. At the same time, through efficient data preprocessing and model parameter optimization, it improves data quality and model generalization ability, and ultimately provides real-time decision support for power system managers, significantly enhancing the disaster resistance and operational stability of the power system, effectively reducing the losses caused by disasters, and achieving efficient allocation of resources and cost savings.
[0215] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0216] Based on the same inventive concept, the embodiment of the present application also provides a typhoon scenario-based power disaster loss development trend prediction device for implementing the above-mentioned typhoon scenario-based power disaster loss development trend prediction method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more typhoon scenario-based power disaster loss development trend prediction device embodiments provided below can refer to the above-mentioned limitations on the typhoon scenario-based power disaster loss development trend prediction method, and will not be repeated here.
[0217] In an exemplary embodiment, Figure 4 As shown, a device for predicting the development trend of power disaster losses based on typhoon scenarios is provided. The device 400 for predicting the development trend of power disaster losses based on typhoon scenarios may include:
[0218] The data acquisition module 401 is used to acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data;
[0219] The data processing module 402 is used to perform feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data;
[0220] The model training module 403 is used to train the power disaster loss development trend prediction model to be trained by using the data features of the historical multi-source data to obtain a trained power disaster loss development trend prediction model;
[0221] The model updating module 404 is used to perform model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model;
[0222] The feature input module 405 is used to input the data features of the current multi-source data corresponding to the object to be predicted into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes the current data of the smart grid, the current data of the meteorological station and the current data of satellite remote sensing.
[0223] In an exemplary embodiment, the feature input module 405 is also used to obtain current multi-source data corresponding to the object to be predicted; perform feature engineering processing on the current multi-source data to obtain data features of the current multi-source data; input the data features of the current multi-source data into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction results of the object to be predicted; the power disaster loss development trend prediction results include the damage probability of power facilities, the affected area, the degree of damage and the recovery time.
[0224] In an exemplary embodiment, the data processing module 402 is also used to perform feature selection processing, feature creation processing, feature encoding processing and feature interaction processing on historical multi-source data to obtain data features of the historical multi-source data; the feature input module 405 is also used to perform feature selection processing, feature creation processing, feature encoding processing and feature interaction processing on current multi-source data to obtain data features of current multi-source data.
[0225] In an exemplary embodiment, the model training module 403 is also used to select a target prediction model associated with the power disaster loss development trend prediction from the candidate prediction models as the power disaster loss development trend prediction model to be trained; construct a training set and a test set of the power disaster loss development trend prediction model to be trained according to the data characteristics of the historical multi-source data; use the training set and the test set to train the power disaster loss development trend prediction model to be trained to obtain a trained power disaster loss development trend prediction model.
[0226] In an exemplary embodiment, the model update module 404 is also used to perform performance identification on the trained power disaster loss development trend prediction model to obtain a performance identification result of the trained power disaster loss development trend prediction model; based on the performance identification result, the trained power disaster loss development trend prediction model is subjected to multi-dimensional model update processing to obtain a target power disaster loss development trend prediction model.
[0227] In an exemplary embodiment, the data acquisition module 401 is also used to acquire the original multi-source data corresponding to the object to be predicted; perform data preprocessing on the original multi-source data to obtain historical multi-source data; data preprocessing includes data cleaning processing, data conversion processing, data integration processing, data normalization processing and data sampling processing.
[0228] Each module in the above-mentioned power disaster loss development trend prediction device based on typhoon scenarios can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.
[0229] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for predicting the development trend of power disaster losses based on typhoon scenarios is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.
[0230] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0231] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.
[0232] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.
[0233] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0234] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.
[0235] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0236] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A method for predicting the development trend of power disaster losses based on typhoon scenarios, characterized in that: The method comprises: Acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data; Performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data; Using the data features of the historical multi-source data, the power disaster loss development trend prediction model to be trained is trained to obtain a trained power disaster loss development trend prediction model; Performing model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model; The data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing.
2. The method according to claim 1, characterized in that: The data features of the current multi-source data corresponding to the object to be predicted are input into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction result of the object to be predicted, including: Acquire current multi-source data corresponding to the object to be predicted; Performing feature engineering processing on the current multi-source data to obtain data features of the current multi-source data; The data features of the current multi-source data are input into the target power disaster loss development trend prediction model to perform power disaster loss development trend prediction, and obtain the power disaster loss development trend prediction result of the object to be predicted; the power disaster loss development trend prediction result includes the damage probability of power facilities, the affected area, the degree of damage and the recovery time.
3. The method according to claim 2, characterized in that The performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data includes: Performing feature selection processing, feature creation processing, feature encoding processing and feature interaction processing on the historical multi-source data to obtain data features of the historical multi-source data; The performing feature engineering processing on the current multi-source data to obtain data features of the current multi-source data includes: The current multi-source data is subjected to feature selection processing, feature creation processing, feature encoding processing and feature interaction processing to obtain data features of the current multi-source data.
4. The method according to claim 1, characterized in that The method of using the data features of the historical multi-source data to train the power disaster loss development trend prediction model to be trained to obtain the trained power disaster loss development trend prediction model includes: Selecting a target prediction model associated with the prediction of power disaster loss development trend from the candidate prediction models as the power disaster loss development trend prediction model to be trained; Constructing a training set and a test set of the power disaster loss development trend prediction model to be trained according to the data characteristics of the historical multi-source data; The training set and the test set are used to train the power disaster loss development trend prediction model to be trained to obtain the trained power disaster loss development trend prediction model.
5. The method according to claim 1, characterized in that The updating process of the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model includes: Performing performance identification on the trained power disaster loss development trend prediction model to obtain a performance identification result of the trained power disaster loss development trend prediction model; According to the performance identification result, the trained power disaster loss development trend prediction model is subjected to multi-dimensional model updating processing to obtain the target power disaster loss development trend prediction model.
6. The method according to any one of claims 1 to 5, characterized in that: The obtaining of historical multi-source data corresponding to the object to be predicted includes: Acquire original multi-source data corresponding to the object to be predicted; The original multi-source data is subjected to data preprocessing to obtain the historical multi-source data; the data preprocessing includes data cleaning processing, data conversion processing, data integration processing, data normalization processing and data sampling processing.
7. A device for predicting the development trend of power disaster losses based on typhoon scenarios, characterized in that: The device comprises: A data acquisition module, used to acquire historical multi-source data corresponding to the object to be predicted; the historical multi-source data at least includes smart grid historical data, weather station historical data and satellite remote sensing historical data; A data processing module, used for performing feature engineering processing on the historical multi-source data to obtain data features of the historical multi-source data; A model training module, used to train the power disaster loss development trend prediction model to be trained by utilizing the data features of the historical multi-source data, and obtain a trained power disaster loss development trend prediction model; A model updating module is used to perform model updating processing on the trained power disaster loss development trend prediction model to obtain a target power disaster loss development trend prediction model; A feature input module is used to input the data features of the current multi-source data corresponding to the object to be predicted into the target power disaster loss development trend prediction model to predict the power disaster loss development trend, and obtain the power disaster loss development trend prediction result of the object to be predicted; the current multi-source data at least includes current data of smart grid, current data of meteorological station and current data of satellite remote sensing.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
Citation Information
Cited By
Metal wire bending prediction method and equipment based on clamp holding posture detection and medium
CN120012302A
Metal wire bending prediction method, device and medium based on clamp holding posture detection
CN120012302B
