Power grid high influence weather characteristic and mode identification method and system
By integrating meteorological data and grid operation data, advanced machine learning technology is used to identify and early warning of high-impact weather events in the power grid, solving the problem of power system not responding quickly enough in extreme weather events, achieving high accuracy and rapid response, reducing the operating risks and costs.
Patent Information
- Application Number
- CN202410789884.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-19
- Publication Date
- 2024-11-15
AI Technical Summary
The existing high-impact weather analysis in the meteorological field lacks customized feature analysis and identification for the power industry, resulting in the power system not responding quickly and flexibly enough in the face of extreme weather events, and lacks effective feature analysis and pattern recognition methods.
By acquiring meteorological data and grid operation data, feature extraction is used using principal component analysis (PCA), weather pattern recognition is combined with deep learning models, multi-category cross entropy loss function is used to optimize the model, and convolutional neural network, recurrent neural network and Integrated prediction of long-term and short-term memory networks to realize the identification and early warning of high-impact weather events in the power grid.
Accurate identification and rapid response to high-impact weather events in the power grid, improve the accuracy of weather identification, reduce the risks of power interruptions and facility damage, optimize operational costs, improve service quality, and through online learning and self-optimization Ensures the timeliness and adaptability of the predictive model.
Smart Images

Figure CN118966740A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the intersection of meteorological science and power system operation management technology, and in particular to a method and system for identifying characteristics and patterns of high-impact weather on power grids. Background Art
[0002] As climate change intensifies, extreme weather events such as heavy rainfall, blizzards, and low temperatures and freezing temperatures are becoming more frequent, posing a severe challenge to the stable operation of the power system. These weather events may not only cause damage to power transmission lines and facilities, but may also trigger sharp fluctuations in power demand, bringing great pressure to power grid operations.
[0003] The current high-impact weather in the meteorological field mainly refers to weather that has a high impact on general social life, and often lacks customized feature analysis and identification for specific industries (such as the power industry). At the same time, the existing power system management often does not respond quickly and flexibly when facing extreme weather events, and lacks effective feature analysis and pattern recognition methods. Therefore, it is necessary to develop a comprehensive technical solution that can combine the latest advances in meteorology and power engineering, provide targeted high-impact weather identification methods, and optimize the power system's response measures based on these results. Summary of the invention
[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.
[0005] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a method for identifying characteristics and patterns of high-impact weather in power grids to solve the problem that existing power system management lacks customized feature analysis and identification when facing extreme weather events.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0007] In a first aspect, the present invention provides a method for identifying characteristics and patterns of high-impact weather in a power grid, comprising:
[0008] Obtain meteorological data and power grid operation data;
[0009] Performing feature selection on the meteorological data and the power grid operation data to obtain a feature subset, and performing feature extraction on the feature subset using PCA to obtain a high-level feature set;
[0010] Based on the advanced feature set, a weather event data set that exceeds the threshold of the power grid response index is screened, the weather event data set is input into a deep learning model for training, the weather pattern that causes abnormal operation of the power system is obtained, and the deep learning model is optimized using a multi-class cross entropy loss function;
[0011] Real-time meteorological data and power grid operation data are input into the trained deep learning model for prediction to obtain prediction results. If the prediction results exceed the warning threshold, the warning mechanism is automatically triggered.
[0012] As a preferred solution of the method for identifying the characteristics and patterns of high-impact weather in the power grid described in the present invention, wherein: feature selection is performed on the meteorological data and the power grid operation data to obtain a feature subset, and the feature subset is extracted by PCA to obtain a high-level feature set including:
[0013] Using a machine learning algorithm to calculate the correlation between each weather feature in the meteorological data and the stability of the power grid, and obtaining a feature subset based on the correlation;
[0014] PCA converts the feature subset into a set of linearly unrelated principal components through orthogonal transformation, and obtains a high-level feature set after dimensionality reduction by selecting and projecting the meteorological data and power grid operation data onto the principal components.
[0015] As a preferred solution of the method for identifying the characteristics and patterns of high-impact weather on the power grid described in the present invention, the weather event data set that exceeds the power grid reaction index threshold is screened based on the advanced feature set, including:
[0016] The weather event data set is labeled, and the labeled weather event data set is aligned with the meteorological data.
[0017] As a preferred solution of the method for identifying the characteristics and patterns of high-impact weather in the power grid described in the present invention, the weather event data set is input into the deep learning model to obtain the weather pattern that causes abnormal operation of the power system, including:
[0018] If the data volume of the weather event data set does not reach the data volume threshold, the advanced feature set is processed using transfer learning and data enhancement technology to increase sample diversity, a small sample data set is obtained, and the small sample data set is input into a deep learning model to obtain the weather pattern that causes abnormal operation of the power system;
[0019] If the data volume of the weather event data set exceeds the data volume threshold, a convolutional neural network, a recurrent neural network, and a long short-term memory network are used for integrated prediction.
[0020] As a preferred solution of the method for identifying the characteristics and patterns of high-impact weather in the power grid described in the present invention, the integrated prediction using convolutional neural networks, recurrent neural networks and long short-term memory networks includes:
[0021] Set the number of network layers, number of neurons, and learning rate for convolutional neural networks, recurrent neural networks, and long short-term memory networks;
[0022] When using the convolutional neural network for prediction, the weather data set is converted into a matrix form, and then the matrix data is converted into a heat map, key features are extracted according to the heat map, and a mapping relationship between the key features and the abnormal operation of the power system is established to obtain the convolutional neural network prediction result;
[0023] Using the recurrent neural network and the long short-term memory network to extract features from the time series data of the weather data set, respectively, calculate the probability that each weather event will cause abnormal operation of the power system, and obtain the prediction results of the recurrent neural network and the long short-term memory network;
[0024] The stacking method is used to combine the convolutional neural network prediction results, the recurrent neural network and the long short-term memory network prediction results to obtain the final prediction result.
[0025] As a preferred solution of the method for identifying the characteristics and patterns of high-impact weather in power grids described in the present invention, wherein: using a multi-category cross entropy loss function to optimize the deep learning model includes:
[0026] The multi-category cross entropy loss function is used to calculate the difference between the deep learning model prediction and the true label. The multi-category cross entropy loss function is expressed as:
[0027]
[0028] Where N is the number of weather event samples, M is the number of weather event categories, and y ic is a binary indicator indicating whether the sample weather event i belongs to the weather event category c, p ic is the probability that the model predicts that sample i belongs to category c.
[0029] As a preferred solution of the method for identifying the characteristics and patterns of high-impact weather in the power grid described in the present invention, if the prediction result exceeds the warning threshold, the automatic triggering of the warning mechanism includes:
[0030] Based on the final prediction results of the deep learning model, the probability values of different weather events are output, and the warning thresholds of different types of weather events and power system parameters are set;
[0031] The automatic early warning system is used to monitor the probability values of different weather events in real time. If the probability values of different weather events exceed the preset probability threshold, the early warning mechanism is automatically triggered.
[0032] In a second aspect, the present invention provides a system for identifying characteristics and patterns of high-impact weather in a power grid, comprising:
[0033] Data acquisition module, used to acquire meteorological data and power grid operation data;
[0034] A feature selection and extraction module, used for performing feature selection on the meteorological data and the power grid operation data to obtain a feature subset, wherein the feature subset is subjected to feature extraction using PCA to obtain an advanced feature set;
[0035] A model training module, for screening weather event data sets that exceed the threshold of power grid response indicators based on the advanced feature set, inputting the weather event data sets into the deep learning model for training, obtaining the weather patterns that cause abnormal operation of the power system, and optimizing the deep learning model using a multi-class cross entropy loss function;
[0036] The model prediction module is used to input real-time meteorological data and power grid operation data into the trained deep learning model to obtain prediction results. If the prediction results exceed the warning threshold, the warning mechanism is automatically triggered.
[0037] In a third aspect, the present invention provides a computing device, comprising:
[0038] Memory and processor;
[0039] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method for identifying characteristics and patterns of high-impact weather in the power grid are implemented.
[0040] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for identifying characteristics and patterns of high-impact weather in the power grid.
[0041] Compared with the prior art, the present invention has the following beneficial effects: the present invention integrates meteorological data and power grid operation data, combined with advanced machine learning technology, to achieve accurate identification and rapid response to high-impact weather events on the power grid. This technology not only improves the accuracy of weather identification, but also provides sufficient preparation time for the power system through real-time monitoring and automated early warning mechanisms, reducing the risk of power outages and facility damage. At the same time, this technology can continuously optimize maintenance and emergency plans, reduce operating costs, and improve service quality. In addition, its online learning and self-optimization capabilities ensure the timeliness and adaptability of the prediction model. In general, the present invention effectively improves the power grid's ability to cope with extreme weather events and makes an important contribution to ensuring the stability and reliability of power supply. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:
[0043] Figure 1 A schematic diagram of the overall process of a method for identifying characteristics and patterns of high-impact weather in a power grid according to an embodiment of the present invention;
[0044] Figure 2 A schematic diagram of the training accuracy of a deep learning model for a method for recognizing the characteristics and patterns of high-impact weather in a power grid according to an embodiment of the present invention. DETAILED DESCRIPTION
[0045] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0047] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0048] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.
[0049] At the same time, in the description of the present invention, it should be noted that the orientations or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the system or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0050] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0051] Example 1
[0052] Reference Figure 1 , as an embodiment of the present invention, provides a method for identifying characteristics and patterns of high-impact weather in a power grid, comprising:
[0053] Prior to describing this embodiment, relevant terms are explained. The high-impact weather characteristics of the power grid are disastrous weather that poses a hidden danger to the stable operation of the power grid. For example, icing of transmission lines caused by low temperatures in winter may cause damage to towers or lines. For example, floods, landslides, mud-rock flows, etc. caused by heavy rainfall in summer may have an impact on transmission lines and substations. In addition, strong convective winds may knock down transmission towers. In the present invention, high-impact weather means the weather that causes the power grid to trip.
[0054] S100: Acquiring meteorological data and power grid operation data;
[0055] Meteorological data includes basic meteorological data such as wind speed, temperature, humidity, precipitation and air pressure, as well as records of extreme weather events such as storms, heavy rains, droughts, low temperature and freezing events, etc. Grid operation data includes load data, equipment failure records, maintenance logs, power flow and consumption, etc.;
[0056] Specifically, select reliable meteorological data providers, such as national or regional meteorological bureaus, international meteorological organizations, and private meteorological service companies, set the time range and frequency of data collection, ensure that the collected data can reflect the impact of meteorological changes in different seasons and time periods on the power grid, and determine the format and quality standards of the data to facilitate data integration and analysis and processing;
[0057] After obtaining meteorological data and power grid operation data, identify and process missing values, abnormal values and duplicate records in the meteorological data and power grid operation data to improve the accuracy and reliability of the data; for missing values, first delete the rows or columns containing missing values, and use the mean or median of the six adjacent values in the column to fill them. Missing values can be filled by interpolation or other data estimation techniques, and the present invention does not make specific restrictions;
[0058] Use the median or standard deviation to handle outliers. Taking the median as an example, the median absolute deviation (MAD) is a robust statistical parameter used to identify outliers. The calculation formula of MAD is:
[0059] MAD=median(|X i -median(X)|)
[0060] MAD stands for Median Absolute Deviation, which is a statistic that measures the degree of dispersion in a data set. Median(X) represents the median of the data set X, which is the value in the middle after all data points are sorted. If the number of data points is even, then the median is usually the average of the two middle numbers, X. i represents the i observation in the data set;
[0061] It should be noted that data points with an absolute value of MAD score greater than 2 were considered outliers and were eliminated;
[0062] Standardize the preprocessed meteorological data and power grid operation data. Specifically, convert the data from different sources into a unified format and unit; for example, convert temperature data into degrees Celsius and precipitation into millimeters. Standardization uses the minimum-maximum scaling method, and the formula is:
[0063]
[0064] Among them, X max , Xmin are the maximum and minimum values in the data set, respectively.
[0065] S102: performing feature selection on meteorological data and power grid operation data to obtain a feature subset, and performing feature extraction on the feature subset using PCA to obtain an advanced feature set;
[0066] Preferably, a machine learning algorithm is used to calculate the correlation between each weather feature in the meteorological data and the stability of the power grid, and a feature subset is obtained according to the correlation;
[0067] In another possible implementation case, meteorological data and power grid operation data are used as feature matrix X, target variable power system operation abnormal state is used as Y, and LASSO model is used for training. The strictness of feature selection is controlled by adjusting the regularization parameter λ. According to the trained LASSO model, the value of the coefficient is checked, and the features with zero coefficients are regarded as unimportant or redundant features and removed from the feature matrix X. The remaining features, that is, the features with non-zero coefficients, are used as feature subsets for subsequent prediction and analysis.
[0068] In another possible implementation case, meteorological data and power grid operation data are used as feature matrix X, and target variable power system operation abnormal state is used as Y. A random forest model is used for training to obtain the importance score of each feature. According to the feature importance score, the feature with the highest score is selected as the important feature. A threshold is set or a fixed number of features are selected. The selected important features are used as feature subsets for subsequent prediction and analysis.
[0069] Preferably, PCA converts the feature subset into a set of linearly uncorrelated principal components through orthogonal transformation, and obtains a high-level feature set after dimensionality reduction by selecting and projecting meteorological data and power grid operation data onto the principal components;
[0070] It should be noted that when extracting time series data of meteorological data, sliding window statistics are used to calculate statistics such as sliding mean and sliding standard deviation within the set time window to capture the local dynamic changes of data in the time series, and then high-level time series features are extracted, such as trend components, seasonal components, autocorrelation and partial autocorrelation coefficients, to reveal the long-term trends, periodic laws and inherent correlations of the data.
[0071] S104: Based on the advanced feature set, a weather event data set that exceeds the power grid response index threshold is screened, the weather event data set is input into the deep learning model for training, the weather pattern that causes abnormal operation of the power system is obtained, and the deep learning model is optimized using a multi-category cross entropy loss function;
[0072] It should be noted that weather characteristics are meteorological conditions that are directly related to the stability of the power grid and the operation of the power system, including temperature, humidity, wind speed and direction, precipitation, thunderstorm activity and atmospheric pressure. Changes in these characteristics may have an important impact on the operating efficiency and safety of power grid equipment; weather patterns refer to the specific combination or changing patterns of a series of weather characteristics in time and space. These weather patterns may include heat waves, storms and freezing weather, which have a comprehensive impact on the operation of the power grid, including risks such as equipment overheating, damage and line failures.
[0073] Preferably, the weather event data set is labeled, and the labeled weather event data set is aligned with the meteorological data;
[0074] Specifically, the weather event data is aligned with the high-level feature set obtained through PCA dimensionality reduction to ensure that the two can accurately match in time and geographical location. Based on the aligned weather event data, the load changes, equipment failure rates and maintenance frequencies of the power grid are quantitatively evaluated. The relationship between the high-level feature set including meteorological features and power grid features and the power grid response indicators is analyzed to obtain the impact of weather events on equipment failure rates and maintenance frequencies, thereby comprehensively quantifying the potential impact of weather events on power grid operations. The predicted value of the power grid response indicator during each weather event is compared with the set threshold. If an indicator exceeds the threshold, it is considered that the weather event has a high impact on the power grid operation. The selected high-impact weather events are further classified and labeled according to weather categories for subsequent analysis and prediction.
[0075] It should be noted that the thresholds of grid response indicators are set based on historical data and the actual needs of grid operation. These thresholds can be set according to the grid's bearing capacity, safety standards, and the impact of historical events;
[0076] Preferably, if the data volume of the weather event data set does not reach the data volume threshold, the advanced feature set is processed using transfer learning and data enhancement technology to increase sample diversity, obtain a small sample data set, and input the small sample data set into the deep learning model to obtain the weather pattern that causes abnormal operation of the power system;
[0077] Preferably, if the data volume of the weather event data set exceeds the data volume threshold, a convolutional neural network, a recurrent neural network and a long short-term memory network are used for integrated prediction;
[0078] It should be noted that when setting the data volume threshold for processing weather event data sets, comprehensive considerations should be made based on actual conditions, such as historical experience, expert knowledge, model performance evaluation, computing resources, time costs, and other factors; specifically, assuming that historical data analysis shows that when the weather event data set size is less than 1,000 samples, the use of transfer learning and data enhancement techniques can effectively improve model performance; and when the data set size exceeds 5,000 samples, directly using the integrated prediction of the deep learning model can obtain better results. In this case, the data volume threshold can be set between 1,000 and 5,000, and the specific value can be fine-tuned based on actual conditions;
[0079] Preferably, the number of network layers, the number of neurons and the learning rate of the convolutional neural network, the recurrent neural network and the long short-term memory network are set;
[0080] In another possible embodiment, a grid search or Bayesian optimization method is used to find the optimal number of network layers, number of neurons, learning rate and other hyperparameters;
[0081] Preferably, when using a convolutional neural network for prediction, the weather data set is converted into a matrix form, and then the matrix data is converted into a heat map, key features are extracted according to the heat map, and a mapping relationship between the key features and the abnormal operation of the power system is established to obtain the convolutional neural network prediction result;
[0082] Preferably, a recurrent neural network and a long short-term memory network are used to extract features from the time series data of the weather data set, respectively, to calculate the probability that each weather event will cause abnormal operation of the power system, and to obtain the prediction results of the recurrent neural network and the long short-term memory network;
[0083] Preferably, a stacking method is used to combine the convolutional neural network prediction results, the recurrent neural network and the long short-term memory network prediction results to obtain the final prediction result.
[0084] Preferably, a multi-class cross entropy loss function is used to calculate the difference between the deep learning model prediction and the true label. The multi-class cross entropy loss function is expressed as:
[0085]
[0086] Where N is the number of weather event samples, M is the number of weather event categories, and y ic is a binary indicator indicating whether the sample weather event i belongs to the weather event category c, p ic is the probability that the model predicts that sample i belongs to category c;
[0087] In another embodiment, a mini-batch method is used for model training to improve training efficiency and reduce memory consumption; or L1 / L2 regularization and dropout are used to alleviate the overfitting problem, especially in the case of small sample data;
[0088] It should be noted that the present invention constructs an efficient and flexible framework to evaluate the potential impact of weather events on power grid operations; based on high-level feature sets, weather events that have a significant impact on power grid reaction indicators are screened out, and deep learning models are used for training to identify weather patterns that cause abnormal operation of the power system; when the amount of data is insufficient, transfer learning and data enhancement techniques are used to enhance the generalization ability of the model; and when the amount of data is sufficient, integrated prediction of convolutional neural networks, recurrent neural networks, and long short-term memory networks is used to obtain more accurate prediction results; reaction indicator thresholds are set according to historical data and actual power grid needs, and a multi-category cross entropy loss function is used to optimize model performance; in addition, the present invention further improves the accuracy of the prediction by combining the prediction results of different networks through a stacking method. The present invention not only takes into account the accuracy and efficiency of the model, but also can alleviate the overfitting problem through methods such as regularization and small batch training, providing powerful decision support for power grid operations.
[0089] S106: Input the real-time meteorological data and power grid operation data into the trained deep learning model to obtain the prediction result. If the prediction result exceeds the warning threshold, the warning mechanism is automatically triggered;
[0090] Preferably, based on the final prediction results of the deep learning model, the probability values of different weather events are output, and the warning thresholds of different types of weather events and power system parameters are set;
[0091] Specifically, during the training phase of the deep learning model, the model is trained through a learning algorithm to understand and parse the input real-time meteorological data and power grid operation data, and then predict its impact on the power grid response indicators. Once the model training is completed, it can accept new real-time data input and make predictions based on it;
[0092] For predicting the probability of different weather events, the deep learning model uses activation functions such as softmax or sigmoid in its output layer to ensure that the output probability value is within a reasonable range, between 0 and 1, and the sum of the probabilities of all weather events is 1;
[0093] In another possible implementation case, to predict different weather events such as heavy rain, strong wind, and lightning, the deep learning model will output a probability vector. The length of this vector is the same as the number of weather event categories. Each element of the vector is a specific probability value, which represents the confidence of the model in the occurrence of the corresponding weather event category. For example, assuming that the model predicts the probability of three weather events: heavy rain, strong wind, and lightning, the deep learning model outputs a vector similar to [0.1, 0.6, 0.3], which means that under the current weather conditions, the probability of heavy rain is 10%, the probability of strong wind is 60%, and the probability of lightning is 30%;
[0094] Preferably, an automated early warning system is used to monitor the probability values of different weather events in real time, and if the probability values of different weather events exceed a preset probability threshold, an early warning mechanism is automatically triggered;
[0095] It should be noted that the setting of warning thresholds is a comprehensive consideration process. By analyzing historical data, we understand the impact of different weather events on power grid operations and determine key indicators such as load changes and equipment failure rates. Based on the historical performance of key indicators, we preliminarily set warning thresholds. In actual operation, we dynamically adjust the thresholds based on the model prediction results and power grid operation performance, and continuously optimize through testing and verification to ensure the accuracy and reliability of the warning system.
[0096] It should also be noted that by inputting real-time meteorological data and power grid operation data into the deep learning model, accurate prediction of power grid operation indicators is achieved. Based on the probability values of different weather events output by the model and combined with historical data analysis, warning thresholds are set, and real-time monitoring and warning triggering are carried out through an automated early warning system, thereby comprehensively assessing power grid operation risks and improving the stability, reliability and safety of power grid operations.
[0097] The above is a schematic scheme of a method for identifying high-impact weather characteristics and patterns of a power grid in this embodiment. It should be noted that the technical scheme of the system for identifying high-impact weather characteristics and patterns of the power grid and the technical scheme of the above-mentioned method for identifying high-impact weather characteristics and patterns of the power grid belong to the same concept. For details not described in detail in the technical scheme of the system for identifying high-impact weather characteristics and patterns of the power grid in this embodiment, please refer to the description of the technical scheme of the above-mentioned method for identifying high-impact weather characteristics and patterns of the power grid.
[0098] The power grid high-impact weather feature and pattern recognition system in this embodiment includes:
[0099] Obtain meteorological data and power grid operation data;
[0100] Feature selection is performed on meteorological data and power grid operation data to obtain feature subsets, and PCA is used to extract features from feature subsets to obtain advanced feature sets;
[0101] Based on the advanced feature set, the weather event dataset that exceeds the threshold of the power grid response index is screened, and the weather event dataset is input into the deep learning model for training to obtain the weather pattern that causes abnormal operation of the power system, and the multi-class cross entropy loss function is used to optimize the deep learning model;
[0102] Real-time meteorological data and power grid operation data are input into the trained deep learning model to obtain prediction results. If the prediction results exceed the warning threshold, the warning mechanism is automatically triggered.
[0103] This embodiment also provides a computing device suitable for identifying characteristics and patterns of weather with high impact on power grids, including:
[0104] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the method for realizing high-impact weather characteristics and pattern recognition of the power grid as proposed in the above embodiment.
[0105] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing the high-impact weather characteristics and pattern recognition method for the power grid as proposed in the above embodiment is implemented.
[0106] The storage medium proposed in this embodiment and the method for realizing the high-impact weather characteristics and pattern recognition of the power grid proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.
[0107] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.
[0108] Example 2
[0109] Reference Figure 2Tables 1-2 are an embodiment of the present invention, which provides a method for identifying the characteristics and patterns of high-impact weather in a power grid. In order to verify its beneficial effects, comparison results of two solutions are provided.
[0110] The meteorological data for this implementation case comes from a private meteorological service company, and it cooperates with the power grid operation center to obtain power grid performance and operation data.
[0111] In the operation and maintenance of power grids, it is crucial to accurately identify the key meteorological factors that affect the stability of the power grid. The LASSO (Least Absolute Shrinkage and Selection Operator) algorithm is used for feature selection, aiming to select the features most relevant to the stability of the power grid from a large number of meteorological features and calculate the correlation between them;
[0112] The LASSO algorithm is a regression analysis method that compresses the model coefficients by introducing penalty terms, thereby achieving feature selection and regression prediction at the same time. The LASSO algorithm is used to select features of meteorological data, and the correlation threshold is set to 0.75; if the correlation between a meteorological feature and power grid stability is lower than 0.75, it will be regarded as an irrelevant feature and excluded from the feature subset;
[0113] After calculation by the LASSO algorithm, a feature subset containing key meteorological features is obtained, and the correlation between them and power grid stability is calculated, as shown in Table 1;
[0114] Table 1 Feature selection
[0115] feature Original eigenvalue range Relevance Selected temperature -10℃-40℃ 0.85 yes humidity 20%-100% 0.76 no Wind speed 0km / h-50km / h 0.90 yes wind direction N,S,E,W,... 0.55 no Rainfall 0mm-200mm 0.80 yes
[0116] As shown in Table 1, the correlations of the four features of temperature, humidity, wind speed and rainfall all exceeded the set threshold of 0.75, and were therefore successfully selected as feature subsets.
[0117] After feature selection and extraction, deep learning models are used for training and prediction, such as Figure 2 As shown in the figure, the present invention significantly reduces the training loss and greatly improves the accuracy of the validation set by training the deep learning model at different learning rates. Figure 2 It demonstrated the rapid progress of model optimization effect and proved the model's good generalization ability for new data. During the entire training process, the model showed a high degree of stability and adaptability to different parameters, which reflected the advancement and practicality of the invention and proved its excellent performance in practical applications.
[0118] Based on the feature selection results, the meteorological and power grid operation data collected at different dates and time points are recorded, as well as the abnormal probability predicted by the model and the results of whether to trigger an early warning, as shown in Table 2;
[0119] Table 2 Real-time prediction and warning results data table
[0120]
[0121] In the early warning system, a reasonable early warning threshold is first set. When the predicted abnormal probability exceeds the threshold, the system will automatically trigger the early warning mechanism to remind relevant managers to take necessary measures. In this experiment, the early warning threshold is set to 0.7;
[0122] The test results are shown in Table 2. From the data in Table 1, it can be seen that when the predicted abnormal probability is lower than the warning threshold, the system does not trigger the warning, indicating that the power grid is operating normally. However, at several time points with higher abnormal probability (such as 14:30 on October 20, 21:00 on October 25, 19:30 on November 10, and 02:00 on November 15), the system successfully triggered the warning, prompting managers that there may be potential risks in the power grid;
[0123] In summary, the present invention integrates meteorological data and power grid operation data, combines the LASSO algorithm for accurate feature selection and correlation calculation, and effectively identifies key meteorological factors that affect power grid stability; on this basis, the deep learning model is used for training and prediction, which not only significantly reduces training losses, but also greatly improves the accuracy of the validation set, demonstrating the model's powerful generalization ability for new data. Through real-time collection of meteorological and power grid operation data, combined with model prediction results, the present invention can accurately predict power grid abnormalities and trigger early warning mechanisms, providing strong guarantees for the safe and stable operation of the power grid.
[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A method for identifying characteristics and patterns of high-impact weather in power grids, characterized in that: include: Obtain meteorological data and power grid operation data; Performing feature selection on the meteorological data and the power grid operation data to obtain a feature subset, and performing feature extraction on the feature subset using PCA to obtain a high-level feature set; Based on the advanced feature set, a weather event data set that exceeds the threshold of the power grid response index is screened, the weather event data set is input into a deep learning model for training, the weather pattern that causes abnormal operation of the power system is obtained, and the deep learning model is optimized using a multi-class cross entropy loss function; Real-time meteorological data and power grid operation data are input into the trained deep learning model for prediction to obtain prediction results. If the prediction results exceed the warning threshold, the warning mechanism is automatically triggered.
2. The method for identifying characteristics and patterns of high-impact weather in a power grid according to claim 1, characterized in that: The meteorological data and the power grid operation data are subjected to feature selection to obtain a feature subset, and the feature subset is subjected to feature extraction using PCA to obtain a high-level feature set including: Using a machine learning algorithm to calculate the correlation between each weather feature in the meteorological data and the stability of the power grid, and obtaining a feature subset based on the correlation; PCA converts the feature subset into a set of linearly unrelated principal components through orthogonal transformation, and obtains a high-level feature set after dimensionality reduction by selecting and projecting the meteorological data and power grid operation data onto the principal components.
3. The method for identifying the characteristics and patterns of high-impact weather in a power grid according to claim 1 or 2, characterized in that: Based on the advanced feature set, the weather event datasets that exceed the threshold of the power grid response index are screened, including: The weather event data set is labeled, and the labeled weather event data set is aligned with the meteorological data.
4. The method for identifying characteristics and patterns of high-impact weather in a power grid according to claim 3, characterized in that: The weather event dataset is fed into the deep learning model to obtain the weather patterns that cause abnormal operation of the power system, including: If the data volume of the weather event data set does not reach the data volume threshold, the advanced feature set is processed using transfer learning and data enhancement technology to increase sample diversity, a small sample data set is obtained, and the small sample data set is input into a deep learning model to obtain the weather pattern that causes abnormal operation of the power system; If the data volume of the weather event data set exceeds the data volume threshold, a convolutional neural network, a recurrent neural network, and a long short-term memory network are used for integrated prediction.
5. The method for identifying characteristics and patterns of high-impact weather in a power grid according to claim 4, characterized in that: Ensemble predictions using convolutional neural networks, recurrent neural networks, and long short-term memory networks include: Set the number of network layers, number of neurons, and learning rate for convolutional neural networks, recurrent neural networks, and long short-term memory networks; When using the convolutional neural network for prediction, the weather data set is converted into a matrix form, and then the matrix data is converted into a heat map, key features are extracted according to the heat map, and a mapping relationship between the key features and the abnormal operation of the power system is established to obtain the convolutional neural network prediction result; Using the recurrent neural network and the long short-term memory network to extract features from the time series data of the weather data set, respectively, calculate the probability that each weather event will cause abnormal operation of the power system, and obtain the prediction results of the recurrent neural network and the long short-term memory network; The stacking method is used to combine the convolutional neural network prediction results, the recurrent neural network and the long short-term memory network prediction results to obtain the final prediction result.
6. The method for identifying characteristics and patterns of high-impact weather in a power grid according to claim 5, characterized in that: Optimizing the deep learning model using a multi-class cross entropy loss function includes: The multi-category cross entropy loss function is used to calculate the difference between the deep learning model prediction and the true label. The multi-category cross entropy loss function is expressed as: Where N is the number of weather event samples, M is the number of weather event categories, and y ic is a binary indicator indicating whether the sample weather event i belongs to the weather event category c, p ic is the probability that the model predicts that sample i belongs to category c.
7. The method for identifying characteristics and patterns of high-impact weather in a power grid according to claim 4, characterized in that: If the prediction result exceeds the warning threshold, the warning mechanism is automatically triggered including: Based on the final prediction results of the deep learning model, the probability values of different weather events are output, and the warning thresholds of different types of weather events and power system parameters are set; The automatic early warning system is used to monitor the probability values of different weather events in real time. If the probability values of different weather events exceed the preset probability threshold, the early warning mechanism is automatically triggered.
8. A system for identifying characteristics and patterns of high-impact weather in power grids, characterized in that: include, Data acquisition module, used to acquire meteorological data and power grid operation data; A feature selection and extraction module, used for performing feature selection on the meteorological data and the power grid operation data to obtain a feature subset, wherein the feature subset is subjected to feature extraction using PCA to obtain an advanced feature set; A model training module, for screening weather event data sets that exceed the threshold of power grid response indicators based on the advanced feature set, inputting the weather event data sets into the deep learning model for training, obtaining the weather patterns that cause abnormal operation of the power system, and optimizing the deep learning model using a multi-class cross entropy loss function; The model prediction module is used to input real-time meteorological data and power grid operation data into the trained deep learning model to obtain prediction results. If the prediction results exceed the warning threshold, the warning mechanism is automatically triggered.
9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method for identifying high-impact weather characteristics and patterns of power grids as described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for identifying characteristics and patterns of high-impact weather in a power grid as described in any one of claims 1 to 7.
Citation Information
Cited By
Data processing method for predicting occurrence time of flight turnover key event
CN120317461A
A data processing method for predicting the occurrence time of key events in flight turnaround
CN120317461B
