Power generation power prediction method, device, equipment and medium

Abstract

The application discloses a power generation power prediction method, device, equipment and medium. The method comprises the following steps: acquiring weather data of a region to which a power station to be measured belongs within a to-be-predicted time length; inputting the weather data into a pre-trained power fluctuation mode identification model to obtain a predicted power fluctuation mode; inputting the weather data into a power prediction model corresponding to the predicted power fluctuation mode to obtain power generation power prediction data of the power station to be measured within the to-be-predicted time length. The technical scheme of the embodiment can improve the accuracy of power generation power prediction and guarantee the reliability of energy supply.

Description

Technical Field

[0001] This invention relates to the field of computer processing technology, and in particular to a method, apparatus, device and medium for predicting power generation. Background Technology

[0002] In recent years, with the promotion of new energy sources, the proportion of new energy power generation in the power system has been increasing. However, under the influence of atmospheric dynamics, the sparse density, volatility, and intermittency of wind and solar energy cause problems such as uncontrollable frequencies, voltage fluctuations, and harmonics, which lead to uncertainties in power generation and increased grid operation and maintenance costs, affecting the safety and stability of the power system. Therefore, in order to cope with this uncertainty, it is necessary to predict the power generation capacity of new energy power plants to ensure energy security and stability.

[0003] Currently, the method for predicting the power generation of new energy power plants is usually based on the mapping relationship between weather parameters and power generation in the future. However, for distributed new energy power plants, the geographical locations of the power plants are relatively scattered, and the deployed weather monitoring equipment is inadequate, making it difficult to accurately collect local weather parameters, resulting in poor power prediction accuracy. Summary of the Invention

[0004] This invention provides a method, apparatus, equipment, and medium for predicting power generation, thereby achieving the technical effect of improving the accuracy of power generation prediction.

[0005] According to one aspect of the present invention, a method for predicting power generation is provided, the method comprising:

[0006] Obtain weather data for the area where the power plant to be tested is located within the forecast period;

[0007] The weather data is input into a pre-trained power fluctuation pattern recognition model to obtain a predicted power fluctuation pattern. The power fluctuation pattern recognition model is trained based on historical power generation data and power fluctuation pattern labels corresponding to the historical power generation data. The power fluctuation pattern labels are determined based on the power fluctuation attributes of the historical power generation data under at least two power fluctuation evaluation indicators. The historical power generation data includes output power data and weather data.

[0008] The weather data is input into the power prediction model corresponding to the predicted power fluctuation pattern to obtain the power generation prediction data of the power generation station under test within the predicted time period; wherein, different power prediction models are trained based on the sample feature matrix and time-varying adjacency matrix of historical power generation data corresponding to different power fluctuation patterns.

[0009] According to another aspect of the present invention, a power generation prediction device is provided, the device comprising:

[0010] The weather data acquisition module is used to acquire weather data for the area where the power plant to be tested is located within the forecast period.

[0011] A power fluctuation pattern determination module is used to input the weather data into a pre-trained power fluctuation pattern recognition model to obtain a predicted power fluctuation pattern; wherein, the power fluctuation pattern recognition model is trained based on historical power generation data and power fluctuation pattern labels corresponding to the historical power generation data; the power fluctuation pattern labels are determined based on the power fluctuation attributes of the historical power generation data under at least two power fluctuation evaluation indicators; the historical power generation data includes output power data and weather data;

[0012] The power generation prediction data determination module is used to input the weather data into the power prediction model corresponding to the predicted power fluctuation pattern to obtain the power generation prediction data of the power station under test within the predicted time period; wherein, different power prediction models are trained based on the sample feature matrix and time-varying adjacency matrix of historical power generation data corresponding to different power fluctuation patterns.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the power generation prediction method according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the power generation prediction method according to any embodiment of the present invention.

[0018] The technical solution of this invention obtains weather data for the area of ​​the power plant under test within the predicted time period; inputs the weather data into a pre-trained power fluctuation pattern recognition model to obtain a predicted power fluctuation pattern; and inputs the weather data into a power prediction model corresponding to the predicted power fluctuation pattern to obtain the predicted power generation data of the power plant under test within the predicted time period. This solves the problem of poor power prediction accuracy caused by predicting power generation based on the mapping relationship between weather parameters and power generation in the prior art. It achieves the evaluation of power fluctuation attributes of historical power generation data based on at least two power fluctuation evaluation indicators, determines power fluctuation pattern labels for historical power generation data according to different power fluctuation attributes, improves the accuracy of power fluctuation pattern evaluation, and then determines the predicted power fluctuation pattern corresponding to the weather data based on the power fluctuation pattern recognition model trained on the historical power generation data and its corresponding power fluctuation pattern labels. This ensures that the predicted power fluctuation pattern can truly reflect the power fluctuation within the predicted time period, effectively addressing the volatility and uncertainty of power plant power. Furthermore, the predicted power generation data is determined based on the weather data by the power prediction model corresponding to the predicted power fluctuation pattern, improving the accuracy of power prediction. Energy dispatch is then carried out based on the predicted power generation, thereby ensuring the reliability of energy supply.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a power generation prediction method provided in Embodiment 1 of the present invention;

[0022] Figure 2 This is a flowchart of a power generation prediction method provided in Embodiment 2 of the present invention;

[0023] Figure 3 This is a schematic diagram of the power generation prediction method provided in Embodiment 2 of the present invention;

[0024] Figure 4 This is a flowchart of a power generation prediction method provided in Embodiment 3 of the present invention;

[0025] Figure 5 This is a schematic diagram of the structure of a power generation prediction device according to Embodiment 4 of the present invention;

[0026] Figure 6 This is a schematic diagram of the structure of an electronic device that implements the power generation prediction method of the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] Example 1

[0030] Figure 1 This is a flowchart of a power generation prediction method according to Embodiment 1 of the present invention. This embodiment is applicable to predicting the power generation of a power plant. The method can be executed by a power generation prediction device, which can be implemented in hardware and / or software and can be configured in a computing device. Figure 1 As shown, the method includes:

[0031] S110. Obtain weather data for the area where the power plant to be tested is located within the forecast period.

[0032] The power plant to be tested can refer to a power plant whose power generation needs to be predicted; for example, this power plant could be a new energy photovoltaic power plant. The prediction duration can be the time period corresponding to which the power generation prediction needs to be performed; for example, a specified future time period can be used as the prediction duration. Weather data includes, but is not limited to, numerical weather prediction data and actual anemometer data. For example, numerical weather prediction data includes solar irradiance, light intensity, wind speed, wind direction, temperature, humidity, air pressure, and solar radiation, etc.

[0033] In this embodiment, weather data can be predicted by a weather data prediction device to predict the power generation capacity of the power plant under test within the predicted time period based on the weather data of the area where the power plant under test is located.

[0034] S120. Input the weather data into the pre-trained power fluctuation pattern recognition model to obtain the predicted power fluctuation pattern.

[0035] The power fluctuation pattern recognition model is trained based on historical power generation data and corresponding power fluctuation pattern labels. Historical power generation data includes output power data and weather data. Power fluctuation pattern labels are used to identify power fluctuation patterns. A power fluctuation pattern can refer to different situations that cause fluctuations in output power, such as wind speed changes, cloud cover, and equipment failure. Power fluctuation patterns can also represent different degrees of power fluctuation. The power fluctuation pattern labels are determined based on the power fluctuation attributes of historical power generation data under at least two power fluctuation evaluation indicators. Power fluctuation attributes can be used to characterize the degree of fluctuation in historical power generation data.

[0036] In this embodiment, weather data can be used as input data in the power fluctuation pattern recognition model, and the predicted power fluctuation pattern can be determined based on the output of the power fluctuation pattern recognition model.

[0037] In this embodiment, the method of determining the predicted power fluctuation pattern may also include at least one of the following methods: one method may be to determine the predicted power fluctuation pattern based on the correlation between numerical weather forecast data and actual anemometer data in the weather data.

[0038] Numerical weather forecast data consists of atmospheric data collected from the atmosphere by meteorological instruments (such as satellites, aircraft, ships, weather stations, and weather balloons), including meteorological parameters such as wind speed, wind direction, temperature, and air pressure. Actual anemometer data consists of weather data measured by monitoring equipment such as anemometers and wind vanes installed on wind towers (masts).

[0039] Specifically, statistical methods (such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) can be used to analyze the correlation between numerical weather prediction data and actual wind measurement data. When the correlation between the two is low, it can be considered that the predicted power fluctuation pattern has large fluctuations or uncertainties.

[0040] Another approach could be to determine the predicted power fluctuation pattern based on the solar irradiance in weather data using a pre-defined data model; where the data model includes the correlation between irradiance and power.

[0041] Specifically, a data model reflecting the correlation between solar irradiance and power can be pre-established; for example, this data model can be a linear model. Solar irradiance from weather data is input into the data model, and the power fluctuations corresponding to the solar irradiance are analyzed to obtain predicted power fluctuation patterns.

[0042] Another approach is to determine the extreme values ​​of preset parameters in the weather data and then use these extreme values ​​to determine the predicted power fluctuation pattern. These preset parameters can include various weather parameters such as wind speed, temperature, light intensity, irradiance, and humidity.

[0043] Specifically, the extreme values ​​of preset parameters can be extracted from weather data, and adjacent extreme values ​​of the same preset parameter can be defined as "waves". By analyzing the fluctuation of these "waves", the predicted power fluctuation pattern can be determined. For example, the greater the fluctuation, the greater the fluctuation or uncertainty in the predicted power fluctuation pattern.

[0044] Another approach is to analyze time series data in weather data, such as the amplitude, frequency, and period of weather fluctuations, and treat the data series under a fixed time window as a fluctuation period, and determine the predicted power fluctuation pattern through different fluctuation periods.

[0045] The technical solution provided in this embodiment preliminarily determines the predicted power fluctuation pattern based on weather data through various methods, thereby improving the accuracy of power fluctuation pattern analysis. As a result, weather data is input into the corresponding power prediction model based on the predicted power fluctuation pattern, thus improving the accuracy of power prediction.

[0046] S130. Input the weather data into the power prediction model corresponding to the predicted power fluctuation mode to obtain the power generation prediction data of the power generation station under test within the predicted time period.

[0047] The different power prediction models are trained based on sample feature matrices and time-varying adjacency matrices of historical power generation data corresponding to different power fluctuation patterns. Each element in the sample feature matrix represents historical power generation data. Each element in the time-varying adjacency matrix represents the temporal relationship between two historical power generation data points. These different power prediction models can be used to predict the power generation of power plants under different weather conditions corresponding to different power fluctuation patterns.

[0048] In this embodiment, after determining the predicted power fluctuation pattern corresponding to the weather data, the weather data can be used as input data for the power prediction model corresponding to the predicted power fluctuation pattern. Based on the output of the power prediction model, the predicted power generation data of the power station under test within the predicted time period is obtained. For example, if the predicted power fluctuation pattern is determined to be A based on the weather data, then the power prediction model trained based on the sample feature matrix and time-varying adjacency matrix of the historical power generation data corresponding to power fluctuation pattern A is input, and the model outputs the predicted power generation.

[0049] The technical solution of this embodiment obtains weather data for the area of ​​the power plant under test within the predicted time period; inputs the weather data into a pre-trained power fluctuation pattern recognition model to obtain a predicted power fluctuation pattern; and inputs the weather data into a power prediction model corresponding to the predicted power fluctuation pattern to obtain the predicted power generation data of the power plant under test within the predicted time period. This solves the problem of poor power prediction accuracy caused by predicting power generation based on the mapping relationship between weather parameters and power generation in the prior art. It realizes the evaluation of power fluctuation attributes of historical power generation data based on at least two power fluctuation evaluation indicators, and determines power fluctuation pattern labels for historical power generation data according to different power fluctuation attributes, thereby improving the accuracy of power fluctuation pattern evaluation. Furthermore, based on the power fluctuation pattern recognition model trained on the historical power generation data and its corresponding power fluctuation pattern labels, it determines the predicted power fluctuation pattern corresponding to the weather data, so that the predicted power fluctuation pattern can truly reflect the power fluctuation within the predicted time period, effectively dealing with the volatility and uncertainty of power plant power. Finally, the power prediction model corresponding to the predicted power fluctuation pattern determines the predicted power generation data based on the weather data, improving the accuracy of power prediction, and using the predicted power generation for energy dispatch, thereby ensuring the reliability of energy supply.

[0050] Example 2

[0051] Figure 2This is a flowchart of a power generation prediction method according to Embodiment 2 of the present invention. Based on the foregoing embodiments, a power fluctuation pattern recognition model and a power prediction model can be pre-trained. Specific implementation methods can be found in the technical solution of this embodiment. Technical terms that are the same as or corresponding to those in the above embodiments will not be repeated here.

[0052] like Figure 2 As shown, the method specifically includes the following steps:

[0053] S210. Obtain historical power generation datasets from at least two power plants.

[0054] To improve the accuracy of model predictions, it is advisable to acquire as much and comprehensive a historical power generation dataset as possible. Furthermore, to ensure model accuracy, the historical power generation data can be preprocessed, such as by performing wavelet decomposition to filter out noise, resulting in a preprocessed historical power generation dataset. This improves data quality and thus enhances prediction accuracy.

[0055] S220. Divide the historical power generation dataset into multiple time slices of historical power generation data.

[0056] In this embodiment, each time slice can be a time period. The historical power generation dataset can be periodically divided based on the time slices, resulting in historical power generation data divided into multiple time slices, so as to determine the power fluctuation pattern of the historical power generation data within each time slice.

[0057] S230. For each historical power generation data, determine the power fluctuation attribute of the historical power generation data under each power fluctuation evaluation index, and determine the power fluctuation mode label corresponding to the historical power generation data based on each power fluctuation attribute.

[0058] Among them, power fluctuation evaluation indicators include at least range indicators, flux indicators, entropy indicators, and variance indicators.

[0059] In this embodiment, the power fluctuation attributes of historical power generation data under each power fluctuation evaluation index can be determined separately, and the determination method includes at least one of the following:

[0060] One approach is to determine the maximum and minimum power values ​​in historical power generation data, and then, based on the difference between the maximum and minimum power values, determine the power fluctuation attributes of the historical power generation data under the range index.

[0061] Specifically, the maximum and minimum power values ​​in historical power generation data can be interpolated to obtain a difference value, which can be used to reflect the amplitude of power fluctuations. This difference value can be used to measure the power fluctuation attributes of historical power generation data under a range indicator. For example, the smaller the difference value, the smaller the fluctuation amplitude; conversely, the larger the difference value, the larger the fluctuation amplitude, meaning the fluctuation represented by the power fluctuation attribute is more unstable.

[0062] Another approach is to determine the total change in power fluctuations in historical power generation data, and based on the total change in power fluctuations, determine the power fluctuation attributes of historical power generation data under the flux index.

[0063] Specifically, the sum of the absolute values ​​of power changes in historical power generation data can be calculated to obtain the total change in power fluctuations. This total change in power fluctuations can be used to reflect the cumulative effect of power fluctuations over a period of time, i.e., flux. The power fluctuation attributes of historical power generation data under the flux index can be measured based on the total change in power fluctuations. For example, the larger the total change in power fluctuations, the stronger the cumulative effect of power fluctuations, which may have a greater impact on system stability, and the more unstable the fluctuations represented by the power fluctuation attributes.

[0064] Another approach is to process historical power generation data using an entropy calculation method based on probability distribution to obtain entropy values, and then determine the power fluctuation attributes of historical power generation data under the entropy index based on the entropy values.

[0065] Specifically, an entropy calculation method based on probability distribution can be used to statistically analyze the probability distribution of power fluctuations in historical power generation data across different intervals, thus calculating the entropy value. This entropy value can be used to describe the irregularity or complexity of power fluctuations. The power fluctuation attributes of historical power generation data under entropy indices can be measured based on the entropy value. For example, a larger entropy value indicates higher irregularity or complexity of power fluctuations, which may be difficult to predict and control; in other words, the fluctuations represented by the power fluctuation attributes are more unstable.

[0066] Another approach is to determine the average power value in the historical power generation data, and based on each power value and the average power value in the historical power generation data, determine the power fluctuation attribute of the historical power generation data under the variance index.

[0067] Specifically, the average power value is calculated from all power values ​​in the historical power generation data. Then, the average of the squares of the differences between each power value and the average power value is calculated to obtain the variance. This variance can be used to describe the degree to which power fluctuations deviate from their average value, reflecting the dispersion of power fluctuations. Furthermore, the variance can be used to measure the power fluctuation attributes of historical power generation data under a variance index. For example, a larger variance indicates more dispersed power fluctuations and poorer power fluctuation stability.

[0068] Furthermore, by comprehensively evaluating the power fluctuation patterns corresponding to historical power generation data using some or all of the power fluctuation attributes, a power fluctuation pattern label can be obtained and annotated for that historical power generation data. For example, different weights can be set for different indicators, and then the weighted average or total score of different power fluctuation attributes can be calculated to assess the overall impact of power fluctuations and obtain the power fluctuation pattern.

[0069] For example, see Figure 3 After acquiring power generation data from new energy power plants (i.e., historical power generation datasets) and filtering out noise through wavelet decomposition, the power generation data can be evaluated based on multi-dimensional indicators (range, flux, entropy, and variance). Power fluctuations can be categorized to obtain historical power generation data corresponding to multiple power fluctuation patterns. Based on the historical power generation data and power fluctuation pattern labels, a power fluctuation pattern recognition model and a power prediction model can be trained. The advantage of this approach is that by fully considering the fluctuation time scale and amplitude range of the categorized power data and using multi-dimensional indicators to evaluate which power fluctuation pattern the historical power generation data belongs to, the single classification dimension of fluctuation patterns is expanded to multiple dimensions, improving the accuracy of power fluctuation pattern assessment.

[0070] S240. Based on each historical power generation data and its corresponding power fluctuation pattern label, train the model to be trained to obtain a trained model.

[0071] The models to be trained include a power fluctuation pattern recognition model and a power prediction model.

[0072] In this embodiment, a training sample set can be determined using each historical power generation data point and its corresponding power fluctuation pattern label. The power fluctuation pattern recognition model is then pre-trained using this sample set. The model learns the sample features corresponding to historical power generation data, which includes output power data and weather data, thus enabling it to predict power fluctuation patterns. Optionally, see [link to relevant documentation]. Figure 3 The power fluctuation pattern recognition model can be a support vector machine. Support vector machines can be used to identify clustered power, correctly divide the training dataset and reach the separating hyperplane with the largest geometric interval, and build a power fluctuation pattern recognition model. After obtaining the weather data of the area where the power plant to be tested belongs within the forecast period, the weather data is input into the pre-trained power fluctuation pattern recognition model, so that the power fluctuation pattern recognition model can predict the predicted power fluctuation pattern.

[0073] In this embodiment, after determining each historical power generation data and its corresponding power fluctuation pattern label, all historical power generation data corresponding to the same power fluctuation pattern label can be integrated to obtain the historical power generation data corresponding to each power fluctuation pattern. The corresponding power prediction model is then pre-trained using the historical power generation data corresponding to each power fluctuation pattern to obtain the power prediction model corresponding to different power fluctuation patterns.

[0074] See also Figure 3 It should be noted that, in order to ensure data consistency, before training the model based on each historical power generation data and its corresponding power fluctuation pattern label, the historical power generation data can be subjected to maximum-minimum normalization. By subtracting the minimum value from the historical power generation data and dividing by the range of the data (the range is the maximum value minus the minimum value), the data is mapped to the range [0,1]. The historical power generation data is then updated, and the model is trained based on the updated historical power generation data and its corresponding power fluctuation pattern label. This avoids certain features having too much influence on the model results, maps the value range of the features to the same interval, makes the features comparable, and thus improves the accuracy and stability of the model.

[0075] It should also be noted that, in order to further improve the model's prediction accuracy, multiple sources of data, such as meteorological data, environmental data, equipment status data, electricity market data, user demand data, and power grid dispatch data, can be used in conjunction with historical power generation data. By comprehensively considering the fusion of multiple sources of data, the model to be trained can be trained to build a more accurate model.

[0076] The technical solution of this embodiment divides the historical power generation dataset into multiple time slices of historical power generation data. Then, based on the power fluctuation attributes of the historical power generation data under each power fluctuation evaluation index, it comprehensively determines the power fluctuation pattern label corresponding to the historical power generation data, thereby improving the accuracy of power fluctuation pattern determination. This improves the accuracy of the power fluctuation pattern recognition model in predicting power fluctuation patterns, and enables the power prediction model to accurately predict power generation based on the weather data corresponding to the power fluctuation pattern.

[0077] Example 3

[0078] Figure 4This is a flowchart of a power generation prediction method according to Embodiment 3 of the present invention. Based on the foregoing embodiments, a power prediction model can be trained based on each historical power generation data and its corresponding power fluctuation pattern label to obtain a trained power prediction model. Specific implementation details can be found in the technical solution of this embodiment. Technical terms that are the same as or corresponding to those in the above embodiments will not be repeated here.

[0079] like Figure 4 As shown, the method specifically includes the following steps:

[0080] S310. Based on all historical power generation data corresponding to the same power fluctuation mode label, determine sample datasets corresponding to at least three spatial levels respectively.

[0081] The spatial hierarchy includes a single-point power generation hierarchy, a first-cluster power generation hierarchy, and a second-category power generation hierarchy. The granularity of the data differs within each spatial hierarchy. It should be noted that the method for determining the power prediction model corresponding to each power fluctuation pattern is the same; the method for determining the power prediction model corresponding to any given power fluctuation pattern can be used as an example for illustration.

[0082] In this embodiment, all historical power generation data corresponding to the same power fluctuation pattern label can be used as a sample dataset for a single-point power generation level. The data within this sample dataset represents power generation data for each power generation station. Simultaneously, a clustering method can be used to cluster all historical power generation data corresponding to the power fluctuation pattern label to obtain a first-clustered power generation level sample dataset. This dataset represents power generation data clustered at the level of multiple power generation stations. Alternatively, another clustering method can be used to cluster all historical power generation data corresponding to the power fluctuation pattern label to obtain a second-category power generation level sample dataset. This dataset represents power generation data clustered at the level of multiple power generation stations. The power prediction model corresponding to the power fluctuation pattern label can be trained using sample datasets corresponding to different spatial levels.

[0083] In this embodiment, the spatial hierarchy includes a single-point power generation hierarchy. Based on all historical power generation data corresponding to the same power fluctuation pattern label, sample datasets corresponding to at least three spatial hierarchies are determined, including: performing time series analysis on all historical power generation data to obtain sample datasets for the single-point power generation hierarchy, so that the sample datasets are power generation data with time series, so that power prediction can be performed subsequently through the power generation data with time series.

[0084] In this embodiment, the spatial hierarchy includes a first clustered power generation hierarchy. Based on all historical power generation data corresponding to the same power fluctuation pattern label, sample datasets corresponding to at least three spatial hierarchies are determined, including: performing cluster analysis on all historical power generation data to obtain at least one first power generation dataset corresponding to a first cluster; for each first power generation dataset, fusing historical power generation data of the same time dimension in the first power generation dataset to obtain second power generation data of the time dimension; and determining the sample dataset of the first clustered power generation hierarchy based on the second power generation data of each time dimension.

[0085] The first cluster includes at least two power plants.

[0086] In this embodiment, a clustering algorithm can be used to perform cluster analysis on all historical power generation data, grouping historical power generation data of power plants with high similarity together, thereby dividing all historical power generation data into multiple clusters. Each cluster is a first cluster, and all historical power generation data within a cluster is the first power generation dataset corresponding to its first cluster. Optionally, clustering algorithms include, but are not limited to, partition-based clustering algorithms (such as K-means), hierarchical clustering algorithms (such as bottom-up and top-down algorithms), density-based clustering algorithms (such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise), grid-based clustering algorithms (STING, Statistical Information Grid), and model-based clustering algorithms (such as Gaussian mixture models), etc. Furthermore, historical power generation data within the same time dimension of the first power generation dataset can be fused. For example, the historical power generation data within the same moment (or time period) can be averaged, and the average can be used as the second power generation data for that time dimension. This allows the second power generation data to be integrated with historical power generation data from multiple power stations within the first cluster, ensuring data authenticity while improving data richness.

[0087] In this embodiment, the spatial hierarchy includes a second-category power generation hierarchy. Based on all historical power generation data corresponding to the same power fluctuation pattern label, sample datasets corresponding to at least three spatial hierarchies are determined, including: performing fuzzy logic analysis on all historical power generation data to obtain at least one third power generation dataset corresponding to the second category; for each third power generation dataset, fusing historical power generation data of the same time dimension in the third power generation dataset to obtain fourth power generation data of the time dimension; and determining the sample dataset of the second-category power generation hierarchy based on the fourth power generation data of each time dimension.

[0088] The second category includes at least two power plants.

[0089] In this embodiment, a fuzzy logic analysis algorithm can be used to fuzzify all historical power generation data, mapping each data feature to a fuzzy set. A membership function is used to represent the membership degree of the data feature to the fuzzy set, and inference is performed on the fuzzy set according to predefined fuzzy rules (e.g., the membership degree of 60,000 lx of light intensity to "high light intensity" can be defined as 0.7). These fuzzy rules can be defined based on domain knowledge or historical data. By applying fuzzy rules, historical power generation data can be divided into different classes, each class being a second category. All historical power generation data within a class constitutes the third power generation dataset corresponding to that second category. Optionally, the fuzzy logic analysis algorithm includes, but is not limited to, the center method and the maximum membership method. Furthermore, historical power generation data within the same time dimension in the third power generation dataset can be fused. For example, the historical power generation data within the same moment (or time period) can be averaged, and the average can be used as the fourth power generation data for that time dimension. This allows the fourth power generation data to be fused with historical power generation data from multiple power stations within the second category, ensuring data authenticity while improving data richness.

[0090] For example, see [link to previous article] Figure 3 Time series analysis, cluster analysis, and fuzzy logic were performed on historical power generation data to obtain datasets for point-level (i.e., single-point power generation level), group-level (first cluster power generation level), and domain-level (i.e., second classification power generation level) predictions. These three datasets were then used to train the power prediction model, which can effectively improve the accuracy and stability of power prediction for different levels of distributed new energy.

[0091] S320. Determine the target sample set based on sample datasets at at least three spatial levels.

[0092] In this embodiment, the data in the sample datasets at different spatial levels are all data with temporal characteristics, and different sample datasets can be used as target sample sets.

[0093] S330. Perform time-series partitioning on the historical power generation data in the target sample set to obtain the sample feature matrix and time-varying adjacency matrix.

[0094] In this matrix, each element in the sample feature matrix represents historical power generation data; and each element in the time-varying adjacency matrix represents the temporal relationship between two historical power generation data.

[0095] Specifically, based on the temporal characteristics of historical power generation data in the target sample set, the time-series data within the historical power generation data can be sliced ​​using a fixed-width sliding window to obtain historical power generation data corresponding to different time nodes. Feature processing can then be performed on the historical power generation data corresponding to different time nodes to obtain their feature representations, constructing a sample feature matrix. Furthermore, a time-varying adjacency matrix can be constructed based on the temporal relationships between different historical power generation data. For example, if the historical power generation data corresponding to time node t has a temporal relationship with the historical power generation data corresponding to the next time node t+1, its element in the time-varying adjacency matrix is ​​represented as 1; otherwise, it is represented as 0. Alternatively, elements representing temporal relationships can be assigned based on the length of the time interval.

[0096] S340. The power prediction model is trained based on the sample feature matrix and the time-varying adjacency matrix to obtain the trained power prediction model corresponding to the power fluctuation mode label.

[0097] In this embodiment, the feature information of a node can be propagated to neighboring nodes by multiplying the time-varying adjacency matrix and the sample feature matrix. Using two graph convolutional blocks in the power prediction model, a bottleneck strategy is used in each graph convolutional block to complete scale compression and feature compression through a sandwich structure of gated convolution-graph convolution-gated convolution. The power prediction model is then trained to obtain a trained power prediction model corresponding to the power fluctuation mode label.

[0098] For example, see [link to previous article] Figure 3The power prediction model can include at least two graph convolutional blocks. Each graph convolutional block comprises a first gated linear unit (GLU), a graph convolutional layer, and a second gated linear unit (GLU). The first gated linear unit, the graph convolutional layer, and the second gated linear unit are connected in series. The trained power prediction model can be used to predict power generation data. Weather data can be input into the power prediction model corresponding to the predicted power fluctuation pattern to obtain the predicted power generation data of the power station under test within the predicted time period.

[0099] The technical solution of this embodiment determines sample datasets corresponding to at least three spatial levels based on all historical power generation data corresponding to the same power fluctuation pattern label. Considering the impact of sample datasets at different spatial levels on the model, the target sample set determined based on the sample datasets at least three spatial levels is time-series partitioned to obtain a sample feature matrix and a time-varying adjacency matrix. The power prediction model is trained using the sample feature matrix and the time-varying adjacency matrix to obtain a trained power prediction model corresponding to the power fluctuation pattern label, thereby improving the accuracy and stability of the power prediction model for power prediction of distributed new energy power plants at different spatial levels.

[0100] Example 4

[0101] Figure 5 This is a schematic diagram of a power generation prediction device according to Embodiment 4 of the present invention. Figure 5 As shown, the device includes: a weather data acquisition module 410, a power fluctuation pattern determination module 420, and a power generation prediction data determination module 430.

[0102] The system includes a weather data acquisition module 410, used to acquire weather data for the area of ​​the power plant under test within the predicted time period; a power fluctuation pattern determination module 420, used to input the weather data into a pre-trained power fluctuation pattern recognition model to obtain a predicted power fluctuation pattern; wherein the power fluctuation pattern recognition model is trained based on historical power generation data and power fluctuation pattern labels corresponding to the historical power generation data; the power fluctuation pattern labels are determined based on the power fluctuation attributes of the historical power generation data under at least two power fluctuation evaluation indicators; the historical power generation data includes output power data and weather data; and a power generation prediction data determination module 430, used to input the weather data into a power prediction model corresponding to the predicted power fluctuation pattern to obtain the predicted power generation data of the power plant under test within the predicted time period; wherein different power prediction models are trained based on the sample feature matrix and time-varying adjacency matrix of historical power generation data corresponding to different power fluctuation patterns.

[0103] The technical solution of this embodiment obtains weather data for the area of ​​the power plant under test within the predicted time period; inputs the weather data into a pre-trained power fluctuation pattern recognition model to obtain a predicted power fluctuation pattern; and inputs the weather data into a power prediction model corresponding to the predicted power fluctuation pattern to obtain the predicted power generation data of the power plant under test within the predicted time period. This solves the problem of poor power prediction accuracy caused by predicting power generation based on the mapping relationship between weather parameters and power generation in the prior art. It realizes the evaluation of power fluctuation attributes of historical power generation data based on at least two power fluctuation evaluation indicators, and determines power fluctuation pattern labels for historical power generation data according to different power fluctuation attributes, thereby improving the accuracy of power fluctuation pattern evaluation. Furthermore, based on the power fluctuation pattern recognition model trained on the historical power generation data and its corresponding power fluctuation pattern labels, it determines the predicted power fluctuation pattern corresponding to the weather data, so that the predicted power fluctuation pattern can truly reflect the power fluctuation within the predicted time period, effectively dealing with the volatility and uncertainty of power plant power. Finally, the power prediction model corresponding to the predicted power fluctuation pattern determines the predicted power generation data based on the weather data, improving the accuracy of power prediction, and using the predicted power generation for energy dispatch, thereby ensuring the reliability of energy supply.

[0104] Optionally, based on the above-described apparatus, the apparatus further includes: a historical power generation dataset determination unit, used to acquire historical power generation datasets from at least two power plants; a dataset partitioning unit, used to partition the historical power generation dataset into historical power generation data in multiple time slices; a pattern label determination unit, used to determine the power fluctuation attribute of each historical power generation data under each power fluctuation evaluation index, and to determine the power fluctuation pattern label corresponding to the historical power generation data based on each power fluctuation attribute; wherein the power fluctuation evaluation index includes at least a range index, a flux index, an entropy index, and a variance index; the power fluctuation pattern label is used to identify the power fluctuation pattern; and a model training unit, used to train a model to be trained based on each historical power generation data and its corresponding power fluctuation pattern label to obtain a trained model; wherein the model to be trained includes a power fluctuation pattern recognition model and a power prediction model.

[0105] Optionally, based on the above-mentioned device, the pattern label determination unit includes: a fluctuation attribute determination first unit, used to determine the maximum power value and minimum power value in the historical power generation data, and determine the power fluctuation attribute of the historical power generation data under a range index based on the difference between the maximum power value and the minimum power value; a fluctuation attribute determination second unit, used to determine the total change in power fluctuation in the historical power generation data, and determine the power fluctuation attribute of the historical power generation data under a flux index based on the total change in power fluctuation; a fluctuation attribute determination third unit, used to process the historical power generation data based on an entropy calculation method of probability distribution to obtain an entropy value, and determine the power fluctuation attribute of the historical power generation data under an entropy index based on the entropy value; and a fluctuation attribute determination fourth unit, used to determine the average power value in the historical power generation data, and determine the power fluctuation attribute of the historical power generation data under a variance index based on each power value in the historical power generation data and the average power value.

[0106] Based on the above-mentioned device, optionally, the model to be trained includes a power prediction model, and the model training unit includes: a spatial hierarchy determination unit, used to determine sample datasets corresponding to at least three spatial hierarchies based on all historical power generation data corresponding to the same power fluctuation pattern label; the spatial hierarchy includes a single-point power generation hierarchy, a first cluster power generation hierarchy, and a second classification power generation hierarchy; a target sample set determination unit, used to determine a target sample set based on the sample datasets of the at least three spatial hierarchies; a matrix determination unit, used to perform time-series partitioning on the historical power generation data in the target sample set to obtain a sample feature matrix and a time-varying adjacency matrix; wherein, each element in the sample feature matrix represents historical power generation data; each element in the time-varying adjacency matrix represents the time-series relationship between two historical power generation data; and a training unit, used to train the power prediction model based on the sample feature matrix and the time-varying adjacency matrix to obtain a trained power prediction model corresponding to the power fluctuation pattern label.

[0107] Based on the above-mentioned device, optionally, the spatial hierarchy includes a first clustered power generation hierarchy, and the spatial hierarchy determining unit includes: a first clustering determining unit, used to perform cluster analysis on all historical power generation data to obtain at least one first cluster corresponding to a first power generation dataset; the first cluster includes at least two power stations; a second power generation data determining unit, used to fuse historical power generation data of the same time dimension in each first power generation dataset to obtain second power generation data of the time dimension; and a sample dataset determining unit, used to determine a sample dataset of the first clustered power generation hierarchy based on the second power generation data of each time dimension.

[0108] Based on the above-mentioned device, optionally, the spatial hierarchy includes a second-category power generation hierarchy, and the spatial hierarchy determination unit includes: a second-category determination unit, used to perform fuzzy logic analysis on all the historical power generation data to obtain at least one third power generation dataset corresponding to a second category; the second category includes at least two power plants; a fourth power generation data determination unit, used to fuse the historical power generation data of the same time dimension in each third power generation dataset to obtain the fourth power generation data of the time dimension; and a dataset determination unit, used to determine the sample dataset of the second-category power generation hierarchy based on the fourth power generation data of each time dimension.

[0109] The power generation prediction device provided in the embodiments of the present invention can execute the power generation prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0110] Example 5

[0111] Figure 6 This is a schematic diagram of the structure of an electronic device implementing the power generation prediction method of embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0112] like Figure 6As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0113] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0114] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as power generation prediction methods.

[0115] In some embodiments, the power generation prediction method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the power generation prediction method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the power generation prediction method by any other suitable means (e.g., by means of firmware).

[0116] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0117] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0118] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0119] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0120] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0121] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0122] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0123] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method of power generation forecast, characterized by, The method comprises: acquiring weather data of a region to which a to-be-tested power station belongs within a to-be-predicted time length; inputting the weather data into a pre-trained power fluctuation pattern recognition model to obtain a predicted power fluctuation pattern; wherein the power fluctuation pattern recognition model is trained based on historical power generation data and power fluctuation pattern labels corresponding to the historical power generation data; the power fluctuation pattern label is determined based on power fluctuation attributes of the historical power generation data under at least two power fluctuation evaluation indexes; the historical power generation data includes output power data and weather data; inputting the weather data into a power prediction model corresponding to the predicted power fluctuation pattern to obtain power generation prediction data of the to-be-tested power station within the to-be-predicted time length; wherein different power prediction models are trained based on sample feature matrices and time-varying adjacency matrices of historical power generation data corresponding to different power fluctuation patterns; The method further comprises training a to-be-trained model based on each historical power generation data and the power fluctuation pattern label corresponding thereto to obtain a trained model, comprising: determining sample data sets corresponding to at least three spatial levels based on all historical power generation data corresponding to a same power fluctuation pattern label; the spatial levels include a single-point power generation level, a first clustering power generation level, and a second classification power generation level; determining a target sample set based on sample data sets of the at least three spatial levels; performing time sequence division on historical power generation data in the target sample set to obtain a sample feature matrix and a time-varying adjacency matrix; wherein each element in the sample feature matrix represents historical power generation data; each element in the time-varying adjacency matrix represents a time sequence relationship between two historical power generation data; training a power prediction model in the to-be-trained model based on the sample feature matrix and the time-varying adjacency matrix to obtain a trained power prediction model corresponding to the power fluctuation pattern label; The spatial levels include a first clustering power generation level, and the determining of the sample data sets corresponding to the at least three spatial levels based on all historical power generation data corresponding to a same power fluctuation pattern label comprises: performing clustering analysis on the all historical power generation data to obtain a first power generation data set corresponding to at least one first cluster; the first cluster includes at least two power stations; for each first power generation data set, performing fusion processing on historical power generation data of a same time dimension in the first power generation data set to obtain second power generation data of the time dimension; determining a sample data set of the first clustering power generation level based on the second power generation data of each time dimension; The spatial levels include a second classification power generation level, and the determining of the sample data sets corresponding to the at least three spatial levels based on all historical power generation data corresponding to a same power fluctuation pattern label comprises: performing fuzzy logic analysis on all the historical power generation data to obtain a third power generation data set corresponding to at least one second classification; the second classification includes at least two power plants; for each third power generation data set, performing fusion processing on historical power generation data of the same time dimension in the third power generation data set to obtain fourth power generation data of the time dimension; based on the fourth power generation data of each time dimension, determining a sample data set of the second classification power generation level.

2. The method of claim 1, wherein, Further comprising: obtaining a historical power generation data set of at least two power plants; dividing the historical power generation data set into a plurality of time-sliced historical power generation data; for each historical power generation data, determining a power fluctuation attribute of the historical power generation data under each power fluctuation evaluation index, and based on each power fluctuation attribute, determining a power fluctuation mode label corresponding to the historical power generation data; wherein the power fluctuation evaluation index at least includes a range index, a flux index, an entropy index and a variance index; the power fluctuation mode label is used to identify the power fluctuation mode; based on each historical power generation data and the power fluctuation mode label corresponding thereto, training a to-be-trained model to obtain a trained model; wherein the to-be-trained model includes a power fluctuation mode recognition model and a power prediction model.

3. The method of claim 2, wherein, The determination of the power fluctuation attribute of the historical power generation data under each power fluctuation evaluation index comprises: determining the maximum power value and the minimum power value in the historical power generation data, and based on the difference between the maximum power value and the minimum power value, determining the power fluctuation attribute of the historical power generation data under the range index; determining the total power fluctuation change in the historical power generation data, and based on the total power fluctuation change, determining the power fluctuation attribute of the historical power generation data under the flux index.

4. The method of claim 2, wherein, The determination of the power fluctuation attribute of the historical power generation data under each power fluctuation evaluation index comprises: processing the historical power generation data based on an entropy calculation method of probability distribution to obtain an entropy value, and based on the entropy value, determining the power fluctuation attribute of the historical power generation data under the entropy index; determining the power average value in the historical power generation data, and based on each power value in the historical power generation data and the power average value, determining the power fluctuation attribute of the historical power generation data under the variance index.

5. A power generation power prediction device characterized by comprising: comprising: a weather data acquisition module configured to acquire weather data of a region to which a to-be-tested power plant belongs within a to-be-predicted time length; The power fluctuation mode determination module is configured to input the weather data into a pre-trained power fluctuation mode recognition model to obtain a predicted power fluctuation mode, wherein the power fluctuation mode recognition model is trained based on historical power generation data and power fluctuation mode labels corresponding to the historical power generation data, the power fluctuation mode labels are determined based on power fluctuation attributes of the historical power generation data under at least two power fluctuation evaluation indexes, and the historical power generation data includes output power data and weather data. The power generation power prediction data determination module is configured to input the weather data into a power prediction model corresponding to the predicted power fluctuation mode to obtain power generation power prediction data of the power station to be tested within the to-be-predicted time length, wherein different power prediction models are trained based on sample feature matrices and time-varying adjacency matrices of historical power generation data corresponding to different power fluctuation modes. The device further includes a model training unit configured to train a to-be-trained model based on each historical power generation data and a power fluctuation mode label corresponding thereto to obtain a trained model, the model training unit includes a spatial level determination unit configured to determine sample data sets corresponding to at least three spatial levels based on all historical power generation data corresponding to a same power fluctuation mode label, the spatial levels include a single-point power generation level, a first clustering power generation level, and a second clustering power generation level, a target sample set determination unit configured to determine a target sample set based on the sample data sets of the at least three spatial levels, a matrix determination unit configured to perform time sequence division on the historical power generation data in the target sample set to obtain a sample feature matrix and a time-varying adjacency matrix, wherein each element in the sample feature matrix represents historical power generation data, and each element in the time-varying adjacency matrix represents a time sequence relationship between two historical power generation data, and a training unit configured to train a power prediction model in the to-be-trained model based on the sample feature matrix and the time-varying adjacency matrix to obtain a trained power prediction model corresponding to the power fluctuation mode label; The spatial level includes the first clustering power generation level, and the spatial level determination unit includes a first clustering determination unit configured to perform clustering analysis on the all historical power generation data to obtain a first power generation data set corresponding to at least one first cluster, the first cluster includes at least two power stations, a second power generation data determination unit configured to, for each of the first power generation data sets, fuse historical power generation data of a same time dimension in the first power generation data set to obtain second power generation data of the time dimension, and a sample data set determination unit configured to determine a sample data set of the first clustering power generation level based on the second power generation data of each time dimension. The space level includes a second classification power generation level, and the space level determination unit includes a second classification determination unit configured to perform fuzzy logic analysis on the all historical power generation data to obtain a third power generation data set corresponding to at least one second classification; the second classification includes at least two power stations; a fourth power generation data determination unit configured to, for each of the third power generation data sets, perform fusion processing on historical power generation data of a same time dimension in the third power generation data set to obtain fourth power generation data of the time dimension; and a data set determination unit configured to determine a sample data set of the second classification power generation level based on the fourth power generation data of each time dimension.

6. An electronic device, comprising: The electronic device includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the power generation power prediction method in any one of claims 1-4.

7. A computer readable storage medium characterized by The computer readable storage medium stores computer instructions for causing the processor to implement the power generation power prediction method in any one of claims 1-4 when executed. The computer readable storage medium stores computer instructions for causing the processor to implement the power generation power prediction method in any one of claims 1-4 when executed.

Images