31 results about "Wind power forecasting" patented technology

The invention discloses a water electrolysis hydrogen production control method and device based on wind and light power generation and electronic equipment. The method comprises the following steps: acquiring the operation cost, the electrolytic cell efficiency and the power grid interaction power variation of the wind-solar hybrid power generation hydrogen production system in a first sampling interval; constructing a target function based on the operation cost, the electrolytic cell efficiency and the power grid interaction power variation; determining the wind power generation predicted power of a wind power generation system in the wind-solar hybrid power generation hydrogen production system in the prediction time period and the photovoltaic power generation predicted power of a photovoltaic power generation system in the prediction time period according to a second sampling interval; and based on the wind power generation predicted power and the photovoltaic power generation predicted power, determining a water electrolysis hydrogen production control strategy (including an electrolytic cell power instruction value and an energy storage compensation power instruction value) by taking the minimum function value of the target function as a target. According to the invention, the technical problems of unbalanced power supply and demand and unstable power grid interaction in the wind-solar hybrid power generation hydrogen production system in the prior art are solved.

Disclosed are an ultra-short-term wind power forecasting method and system, relating to the technical field of artificial intelligence. The method comprises: obtaining an original dataset of a wind farm, processing the original dataset, and performing training on the basis of processed original data; decomposing wind speed data in the trained original data, calculating each decomposition component, and constructing a feature matrix on the basis of the calculation results; and introducing a residual attention mechanism to reconstruct the feature matrix, using the reconstructed result to establish a network model, performing secondary training, and forecasting ultra-short-term wind power. The present invention improves the accuracy and reliability of ultra-short-term wind power forecasting and achieves significant advances in algorithm optimization, thereby providing effective support for the stable power supply of renewable energy sources such as wind farms and for power grid operation.

The invention relates to the technical field of new energy power prediction, in particular to an extreme weather offshore wind power generation situation prediction method based on artificial intelligence, and the method comprises the steps: obtaining data under historical extreme weather, dividing the data into a training set and an optimization set, removing noise, extracting environment data, and carrying out the feature data fusion. The method comprises the following steps: determining a freezing proportion according to an extreme weather disaster grade, freezing partial layer parameters of a pre-trained conventional power generation prediction model, training an unfrozen layer, constructing an extreme weather power generation prediction model, periodically obtaining data in an optimization set through constructing a simulation time axis, carrying out automatic learning, and finally obtaining environmental data in real time for prediction. And judging the prediction accuracy according to the similarity between the prediction result and the optimization set data, and if the prediction accuracy is not accurate, analyzing an abnormal reason and correcting related parameters. The method provided by the invention effectively overcomes the difficulty of inaccurate offshore wind power generation prediction in extreme weather, and significantly improves the accuracy and reliability of wind power generation prediction under extreme weather conditions.

The invention relates to the technical field of energy storage coordination control, in particular to an energy storage coordination control method and system for wind power generation. The method comprises the following steps: acquiring historical power generation states and environmental data, and drawing a power generation power fluctuation state curve to obtain a fluctuation state filling curve; then, on the basis of a historical environment state, multivariable regression analysis and generated power fluctuation data association are utilized to obtain multivariable power fluctuation associated data, and power generation behavior learning induction is carried out to form generated power incremental behavior data; and finally, constructing a wind power generation prediction model by using the generated power increasing behavior learning data, predicting the future generating capacity based on the model, formulating an energy storage adaptive coordination control strategy, and sending the strategy to a control terminal to execute energy storage coordination control. According to the invention, the energy storage coordination control technology is optimized, so that the energy storage coordination control technology is more accurate.

The invention discloses a wind power prediction covariable offset adaptive method based on a conditional energy model. According to the method, the wind power generation power prediction model is constructed by utilizing the gated cycle unit network, and offline training of the wind power generation power prediction model is completed by adopting a sample weighting mechanism, so that the robustness of the wind power generation power prediction model to distribution change is enhanced. A conditional de-noising score matching strategy is adopted to learn the distribution difference of data in a training stage and a prediction stage through a conditional energy model, and a sample weight used for measuring the covariable offset degree is obtained based on the model. Data samples flowing in real time are stored in a replay buffer area, incremental learning is carried out on a condition energy model and a wind power generation prediction model through an online updating mechanism, and the prediction performance is kept stable. The method can effectively improve the power prediction precision and operation scheduling capability of the wind power plant under complex meteorological conditions, and has good engineering practical value and deployment flexibility.

The invention discloses a regulation and control method and device for hydrogen production and methanol production and electronic equipment, and relates to the technical field of energy system regulation and control, and the method comprises the steps: firstly, obtaining weather forecast data in a preset time period, carrying out wind power generation prediction according to the weather forecast data, and obtaining a wind power prediction result; then, a reward function and a penalty function corresponding to each production link are constructed, and a comprehensive score corresponding to each production link is obtained based on the reward function and the penalty function; and finally, based on the wind power prediction result and the comprehensive score corresponding to each production link, performing wind power distribution and production scheduling for each production link. By means of the method, the reward function and the penalty function are designed, the fluctuation situation occurring in the production process can be effectively dealt with, the production process is adjusted in advance, high risks possibly existing in a short time when the demand list is currently met are actively avoided, the production smoothness is optimized, and the production stability and continuity are improved.

The invention discloses a day-ahead wind power prediction method under the combination of physics and data driving, and relates to the field of wind power prediction. The problem that the existing single-mode wind power prediction method has limitation and low prediction accuracy when coping with multi-time scale wind power prediction is solved. According to the method, a final prediction result is obtained through a physical driving mode and a data driving mode, the final prediction result is mainly composed of two parts, namely a first part (1-5-hour prediction) and a second part (6-24-hour prediction), a data driving result is directly adopted as a predicted value of wind power, the physical driving prediction result and the data driving prediction result are fused, and the prediction result of the wind power is obtained. The two parts are combined to obtain the wind power sequence of the target electric field in the next 24 hours in the combination mode so as to improve the accuracy of wind power prediction. The method is mainly used for predicting the day-ahead wind power of the wind power plant.

The invention relates to the technical field of wind power prediction, and particularly discloses a gradient synergy and double alignment fused small sample wind power prediction method, which comprises the following steps: designing a Fourier enhanced Transform shared feature extractor to extract general feature representation with periodic perception from wind power time sequence data of a multi-source domain and a target domain; then, constructing a hybrid domain adaptive module, and realizing implicit and explicit dual alignment of feature distribution of a source domain and a target domain through a plurality of adversarial domain classifiers arranged in parallel and multi-core maximum mean difference measurement; and finally, introducing a gradient projection algorithm, carrying out collaborative optimization on conflict gradients of the prediction task and the domain adaptation task in a back propagation process, and eliminating gradient conflicts in multi-task learning. The method effectively improves the feature extraction capability, domain adaptability and optimization stability of the model in a small sample scene, and remarkably improves the prediction precision and robustness in a cross-domain wind power prediction task.

The invention discloses a wind power generation predicted power correction method and device influenced by extreme weather, equipment and a medium, and the method comprises the steps: obtaining the current environment data of a place where a target wind power generator is located, inputting the current environment data into an environment recognition training network, and obtaining an environment recognition result; if the environment identification result simultaneously comprises an extreme weather identification result and an extreme weather identification type, determining the operation deterioration probability of the target wind driven generator; according to the extreme weather identification type, the operation power of the target wind driven generator is corrected for the first time, and first correction power is obtained; and according to the operation deterioration probability and the first correction power, carrying out second correction on the operation power of the target wind driven generator, and taking the second correction power as the target prediction power of the target wind driven generator under the current environment data. The invention belongs to the field of wind power generation power prediction. Wind power generation prediction under the influence of extreme weather can be realized.

The invention relates to the technical field of wind and light power generation prediction, in particular to a solar photovoltaic and wind power generation prediction method based on DTCN-FFT and a related device. Comprising the following steps: acquiring photovoltaic and wind power generation data of solar energy, and preprocessing the photovoltaic and wind power generation data of the solar energy to obtain processed data; spearman correlation analysis is carried out on the processed data, and input characteristics with high correlation with the generating capacity are screened out; inputting the input features into a pre-constructed DTCN-FFT model, and outputting prediction results of solar photovoltaic power generation and wind power generation; according to the method, frequency domain information and time domain information are fused, original time sequence data are converted into a frequency domain through fast Fourier transform, global features such as periodic components, trend terms and high-frequency noise are extracted, and the defect that a traditional model is insufficient in long-range dependence capture is overcome.

The invention provides a short-term wind power prediction method based on GWO-VMD-FE and TCN-BiGRU. The short-term wind power prediction method is used for improving the accuracy and prediction efficiency of short-term wind power prediction. The method comprises the following steps: firstly, carrying out automatic optimization on a penalty factor alpha and a modal decomposition number K of VMD by utilizing GWO, and decomposing wind power data into an intrinsic mode function (IMF) component through the VMD; secondly, calculating a fuzzy entropy value of the IMF component by adopting an FE algorithm, reconstructing the IMF component into a trend component and a random component according to the entropy value, fusing the trend component and the random component with an optimal feature screened by a Spearman correlation coefficient to obtain a trend sample and a random sample, taking the trend sample and the random sample as input of a TCN-BiGRU model for prediction, and superposing prediction results of the two samples to obtain a TCN-BiGRU model; and obtaining a wind power prediction result. According to the method, the time sequence characteristics and the front-and-back dependency relationship of the data can be more sensitively captured, and the accuracy of wind power prediction can be effectively improved.

The application discloses a method for fine-tuning a pre-trained large model for wind power generation prediction, which comprises: obtaining a historical multi-dimensional time series dataset, generating a wind speed trend sequence through space-time feature fusion by a sliding window; generating a step adjustment parameter by a normalization function based on the wind speed gradient of adjacent time windows; dynamically adjusting the pre-trained time series prediction model based on the step adjustment parameter, combining a hierarchical transfer learning strategy, and adjusting the top network parameters of the time series prediction model through an adaptive optimization algorithm to obtain an optimized time series prediction model; and processing real-time multi-dimensional time series data sets by using the optimized model to generate wind power prediction results. Through dynamic adjustment of model parameters and the hierarchical transfer learning strategy, the accuracy and model adaptation capability of wind power prediction are effectively improved, and the method is suitable for time series prediction scenarios in the field of wind power generation.

The application discloses a kind of with self-attention mechanism and gate image-based wind power short-term prediction method and system, belong to wind power generation prediction technical field, wherein, the method includes: obtaining NWP meteorological data and SCADA fan data is preprocessed, and the data after preprocessing is divided into training set and test set;Training set is decomposed into continuous multiple historical time subsequences, and is reconstructed into two-dimensional image;Residual-based deep convolutional neural network is established, and gate convolutional neural network layer and self-attention mechanism are added, to obtain residual-based deep convolutional neural network with self-attention mechanism and gate;Two-dimensional image is input into the neural network training;Test set is input into the neural network trained and is predicted, to obtain short-term wind power.The method comprehensively uses historical time series data and meteorological grid data, extracts features and converts into images, uses advanced image processing technology to predict wind, greatly improves the prediction accuracy.

The invention belongs to the technical field of energy prediction, and particularly relates to a long historical data-oriented periodic sensing wind power prediction method, which comprises the following steps of: obtaining wind power generation historical data; determining a data cycle value based on autocorrelation function analysis, and performing fragmentation processing and sequence decomposition on historical data according to a cycle length to obtain a seasonal component and a trend component; performing multi-scale feature enhancement on the fragmented data, and extracting rich time sequence dynamic information; and based on the enhanced fragment representation, outputting a plurality of fragment prediction results through a prediction model, and performing weighted fusion according to cosine similarity to obtain a final wind power generation power prediction result. And carrying out optimization training on the prediction model by adopting a time-frequency domain difference loss function. According to the method, periodic information and time sequence characteristics in the long historical window can be effectively utilized, the accuracy and calculation efficiency of wind power generation power prediction are remarkably improved, and reliable support is provided for power grid dispatching and energy management.

The present invention relates to a method for predicting wind power generation suitable for complex terrain, comprising: a weather information collection step for collecting weather information including forecasts for wind components predicted for a plurality of eta layers arranged vertically from the ground; a vertical characteristic integration step for deriving additional input characteristics using a characteristic extraction method that considers the physical characteristics of wind components for the wind components predicted for the plurality of eta layers; and a power generation prediction step for predicting power generation by applying the weather information and the additional input characteristics to a machine-learned algorithm for predicting power generation, wherein the algorithm is characterized by being learned using weather information for wind components, additional input characteristics derived using a characteristic extraction method that considers the physical characteristics of wind components for the wind components of the plurality of eta layers, and power generation. The present invention has the effect of improving the accuracy of wind power generation prediction for complex terrain by applying input characteristics derived by applying a characteristic extraction method that considers the physical characteristics of wind components predicted in a plurality of vertically arranged eta layers together with a conventional method for predicting wind power generation using weather information.

The invention provides a wind power generation capacity prediction method and system considering the operation state of fan equipment, and relates to the technical field of wind power generation prediction, and the method comprises the following steps: periodically collecting a plurality of wind power data and a plurality of fan operation state data in a time period; constructing fan operation state characteristics according to the fan operation state data; quantifying the running state characteristics of the fan to obtain a running state coefficient of the fan; calculating the theoretical power generation power of the fan in the next time period according to the wind power data; and correcting the theoretical generated power of the fan through the running state coefficient to obtain predicted generated power considering the influence of the running state of the equipment. The method has the advantage that the accuracy of a prediction result can be improved.

This invention discloses a wind power generation prediction method based on a variable-priority adaptive grey model. The method involves collecting annual wind power generation data for a target area and establishing an original sequence; introducing weight parameters and an adaptive variable weight function to construct a variable-priority accumulation operator; calculating a variable-priority accumulation sequence using the original sequence and the variable-priority accumulation operator; introducing the variable-priority accumulation sequence and a double nonlinear term to improve the grey discrete model, thus constructing a variable-priority adaptive grey discrete model; optimizing the nonlinear parameters of the model using a hybrid algorithm of social spider and differential evolution to obtain optimal nonlinear parameter values; and then solving the parameters of the model's time response equation using the least squares method to obtain the optimal model; finally, using the optimal model to predict future wind power generation. The introduction of a variable-priority accumulation operator adaptively adjusts the data priority to reduce the impact of uncertainty in new wind power data, thereby improving the prediction accuracy of wind power generation.

This invention belongs to the field of wind power generation prediction technology and discloses a method for predicting wind power output in complex terrain based on high-precision meteorological and terrain coupling. The method includes: acquiring digital elevation model (DEM) data and global reanalysis meteorological data; constructing a multi-scale nested meteorological simulation system, embedding the DEM into the reanalysis data, enabling large eddy simulation at the innermost layer, and parameterizing the wind turbine in the form of a momentum sink to generate three-dimensional wind field data; extracting static terrain feature vectors; calculating transient air density based on real-time air pressure, temperature, and humidity to correct the theoretical power of the wind turbine; and inputting the three-dimensional wind field data, static terrain feature vectors, and corrected theoretical power into a hybrid neural network prediction model to output predicted active power. This invention solves the problems of low prediction accuracy and lack of multi-physics coupling in wind power generation in complex terrain, significantly improving prediction accuracy and applicable to wind farm planning and operation and maintenance decisions.

Method for hierarchically coordinated control of a wind-hydrogen coupling system based on model predictive control (MPC), wherein the wind-hydrogen coupling system is divided into a higher-level grid coupling control and a lower-level electrolysis cell control; the procedure has the following features: (1) The higher-level grid coupling control uses a model predictive control algorithm to control a grid-connected power and follow a wind power forecast curve, while a power control variable of an electrolysis cell is obtained for the lower-level electrolysis cell control, where the model predictive control is a model forecast control; (2) The operating state of the electrolysis cell is divided into four operating states, namely: nominal power operation, fluctuating power operation, overload power operation and shutdown; (3) According to the power control parameter of the electrolysis cell, a dual-track time-power control strategy with rotation cycle is applied to determine the operating state of each electrolysis cell, so that the electrolysis cell is operated alternately in one of the four operating states.

The application provides a unified space-time wind power prediction method based on a Fourier generative adversarial network, comprising: obtaining a hyper-variable graph based on unified space-time, the hyper-variable graph being constructed based on historical power sequence data sets of multiple wind generators; constructing a multi-wind farm power prediction model; the multi-wind farm power prediction model is a deep regret analysis generative adversarial network (DRAGAN), which is an adversarial network with a Fourier GNN as a generator and a convolutional neural network (CNN) as a discriminator; the hyper-variable graph is used to iteratively train the multi-wind farm power prediction model to obtain a multi-wind farm power prediction model meeting the prediction requirements. The method provided by the application optimizes the parameters of the generator through the mutual game training of the generator Fourier GNN and the discriminator CNN, improves the quality of the generated samples, and further improves the prediction accuracy.

This invention relates to the field of wind power generation technology and discloses a wind power generation prediction method, system, electronic device, and storage medium. The method includes: acquiring a high-dimensional feature representation, which is obtained by mapping multi-source heterogeneous spatiotemporal sequence data to a decoupled latent space through a feature variational autoencoder; based on the high-dimensional feature representation, performing physical simulation and data-driven prediction in parallel to generate a power prediction sequence; constructing a digital twin environment for wind farm operation based on real-time wind farm operation data and the power prediction sequence; and in the digital twin environment for wind farm operation, solving the conflicts between multiple optimization objectives through multi-round iterative game to obtain a power generation plan and bidding strategy. This invention achieves autonomous collaborative optimization of the power generation plan and bidding strategy, improving the accuracy, continuity, and collaborative optimization capability of prediction.

The invention provides a wind power generation prediction method and system, and belongs to the technical field of wind power generation, and the method comprises the steps: S1, obtaining wind power generation data; s2, decomposing the wind power generation data based on an ensemble empirical mode decomposition algorithm to obtain a plurality of intrinsic mode function components and residual terms; s3, inputting the plurality of intrinsic mode function components into a pre-trained wind power generation prediction model to obtain prediction results corresponding to the plurality of intrinsic mode function components; and S4, superposing the prediction results of the plurality of intrinsic mode function components to obtain a wind power generation prediction result. The method has the beneficial effects that the wind power generation data is decomposed by adopting the ensemble empirical mode decomposition algorithm, the randomness of an original sequence is reduced, more regular input characteristics are provided for a subsequent prediction model, the mode aliasing problem of a traditional method is solved, and the prediction precision of the model is improved.

The application belongs to the technical field of wind power prediction, and particularly relates to a double-branch space-time wind power prediction method and system based on multi-modal fusion. The method comprises the following steps: S1, performing maximum minimum normalization on input original wind power sequence data to obtain standardized wind power sequence data; S2, performing multi-mode embedding processing on the standardized wind power sequence to obtain time sequence features integrating space-time information and feature correlation; meanwhile, performing mode conversion processing on the standardized wind power sequence to obtain image features; S3, respectively performing time sequence feature extraction and image feature extraction on the time sequence features and the image features obtained in S2 to obtain enhanced time sequence features and enhanced image features; S4, performing multi-modal feature fusion on the enhanced time sequence features and the enhanced image features obtained in S3 to obtain fusion features comprising time sequence modes and image modes; and S5, generating multi-step wind power prediction by using a prediction architecture based on GRU to provide prediction results for multiple future time periods.

The invention discloses a medium and short term wind power generation prediction method, relates to a power grid power generation prediction method, and aims to solve the problems of low prediction efficiency and poor prediction accuracy of an existing wind power generation prediction method. The method comprises the steps of obtaining historical data; preprocessing the acquired historical data to generate processed historical data; performing time sequence data analysis on the processed historical data to generate a historical data segmentation sequence; variational mode decomposition is carried out on the historical data segmentation sequence, and a mode component is extracted; and inputting the extracted modal component into a GA-BP model, and predicting the medium and short term wind power generation. The beneficial effects are that prediction precision and prediction efficiency are improved.

The invention, which relates to the technical field of new energy power generation, discloses a multi-source data-based new energy power generation prediction system comprising a data acquisition module, a data preprocessing module, a feature extraction module, a wind power generation prediction unit, a photovoltaic power generation prediction unit, a comprehensive prediction module and an output module. The data acquisition module acquires wind speed, wind direction, air density, solar radiation intensity, illumination duration and environment temperature data, the data preprocessing module performs abnormal value elimination, missing value filling and normalization processing, and the feature extraction module extracts wind power generation feature vectors and photovoltaic power generation feature vectors. The wind power generation prediction unit adopts a long-short-term memory neural network model to perform prediction, the photovoltaic power generation prediction unit adopts a gated circulation unit neural network model to perform prediction, the comprehensive prediction module performs weighted fusion on two prediction results, and the output module outputs a comprehensive power generation power prediction value.

The invention relates to the technical field of wind power generation, discloses a wind power generation prediction method, device and equipment, a medium and a program product, and aims to meet the second-level prediction basic data requirement by performing data acquisition according to a preset acquisition frequency. Furthermore, POD dimension reduction processing is carried out on the real-time target wind power generation data set, so that the real-time data dimension is reduced, redundant information is reduced, the subsequent model calculation burden is remarkably reduced, the high efficiency of real-time data processing by the model is ensured, and the real-time requirement of second-level prediction is met. Furthermore, by adopting a target wind power generation prediction model of a multi-layer Transform encoder, a time sequence dependency relationship of multi-dimensional characteristics in real-time data can be captured, and accurate prediction of short-term power is realized. And finally, restoring a low-dimensional prediction result to an original power data dimension and obtaining a power prediction value conforming to an actual application scene, thereby providing directly available quantized data for wind power plant regulation and control.

The invention provides a unified space-time wind power prediction method based on a Fourier generative adversarial network, and the method comprises the steps: obtaining a hypervariable graph based on a unified space-time, and the hypervariable graph is constructed based on a historical power sequence data set of a plurality of wind driven generators; constructing a multi-wind power plant power prediction model; the multi-wind power plant power prediction model is a deep regret analysis generative adversarial network DRAGAN, and the deep regret analysis generative adversarial network is an adversarial network taking a Fourier GNN as a generator and a convolutional neural network CNN as a discriminator; and performing iterative training on the multi-wind-power-plant power prediction model by adopting the hypervariable graph to obtain a multi-wind-power-plant power prediction model meeting prediction requirements. According to the method provided by the invention, through the mutual game training of the generator Fourier GNN and the discriminator CNN, the parameters of the generator are optimized, the quality of the generated sample is improved, and the prediction precision is further improved.

The present application relates to the field of short-term wind power generation prediction, and discloses a few-sample wind power prediction method fusing mechanism migration modeling, step 1) wind power generation data preprocessing and dataset division; step 2) MIC characteristic variable screening; step 3) defining characteristic variables and labels on the source domain and the target domain; step 4) establishing a KAN mechanism fusion model on the source domain; step 5) source domain pre-training and physical guided migration to the target domain; step 6) model training, prediction and model performance evaluation on the target domain; the present application introduces a migration strategy of freezing the physical layer, migrates the trained physical network layer in the source domain to the target domain, and only fine-tunes the prediction module, so as to realize wind power modeling under the condition of low samples; the strategy fully retains the physical feature expression ability in the source domain, significantly improves the prediction accuracy and stability under the condition of small sample learning of the target domain, and shows good migration generalization ability.

The invention discloses an interpretable short-term wind power generation prediction method based on a Kolmogorov-Arnold network, and relates to the field of wind power generation power prediction and frequency domain modeling analysis of a power system, and the method comprises the steps: carrying out the multi-source data collection of target power generation equipment in a historical time range, constructing a prediction input feature set, and carrying out the calculation of a prediction input feature set; acquiring a generation power time sequence and an environment correlation characteristic thereof; and according to the constructed input and output samples, constructing an initial KAN network and carrying out full-amount training. The method can provide mathematical model support for modeling, prediction and characteristic analysis of the power generation process of the wind power station. By introducing a symbolized activation function structure and a network pruning mechanism, a function relationship between input characteristics and power generation output can be accurately identified under a limited sample condition, and a prediction model with a clear mathematical analysis form is extracted, so that quantitative modeling and interpretable analysis of new energy power prediction are realized, and the prediction efficiency is improved. And the expression accuracy and engineering applicability of the model are improved.

The invention discloses a wind power prediction method and system based on stationary wavelet analysis and an attention mechanism. The method comprises the steps that long-term wind power generation observation data and meteorological element field data are collected and preprocessed; extracting periodic signals of different scales of the wind power generation observation data based on stationary wavelet analysis; training features of the meteorological element field extracted based on an auto-encoder and obtaining a feature sequence; building a neural network model based on a self-attention mechanism and a cross attention mechanism, and achieving the prediction of wind power generation. The method has the advantages that wind power prediction is introduced through stationary wavelet analysis, the problem that multi-scale periodic signals are difficult to capture in a traditional method is solved, and the signals are enhanced at the same time; the features of a meteorological element field in the same period are extracted through an attention mechanism, and extra forecasting factors are provided for wind power generation; and finally, based on a self-attention mechanism and a cross attention mechanism, external force of meteorological elements and a multi-scale periodic variation rate of wind power generation are effectively fused, and a wind power generation prediction skill is improved.