A method and system for long-term wind power probability prediction based on causal enhancement
Patent Information
- Application Number
- CN202610939689.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-26
- Publication Date
- 2026-09-15
Smart Images
Figure CN122763335A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of new energy power generation prediction technology, and relates to a medium- and long-term wind power probability prediction method and system based on causal enhancement. Background Technology
[0002] Wind energy, being clean and renewable, has become a key energy source for carbon neutrality. However, wind power output is highly intermittent and volatile due to weather fluctuations. Accurate medium- and long-term probabilistic forecasting is crucial for ensuring grid security and stability and improving the economic benefits of wind farms. Existing medium- and long-term wind power forecasts largely rely on pure data-driven methods or traditional statistical models, which suffer from problems such as redundant meteorological characteristics, insufficient quantification of uncertainties, and poor robustness to extreme weather, making it difficult to meet the needs of medium- and long-term grid dispatching and wind farm operation and maintenance planning.
[0003] Short-term wind power forecasting: mainly based on deep learning, such as spatiotemporal graph convolution, attention LSTM, Transformer, etc., which has high accuracy in the short term, but weak generalization ability in the medium and long term.
[0004] Medium- to long-term wind power forecasting: Traditional statistical methods such as ARIMA and exponential smoothing struggle to fit nonlinear meteorological-power relationships, and their accuracy drops sharply under variable weather conditions. Pure data-driven deep learning methods such as LSTM, GRU, and CNN-LSTM are susceptible to autocorrelation interference from meteorological features, leading to overfitting and insufficient uncertainty quantification capabilities. Clustering / dimensionality reduction methods such as K-means, GA, and PSO suffer from high computational complexity, are prone to getting trapped in local optima, and lack causal basis for feature selection.
[0005] Probabilistic predictions are mostly based on Gaussian assumptions or simple interval estimations, which result in insufficient coverage and low reliability of probability intervals under extreme weather conditions.
[0006] In summary, existing technologies suffer from poor accuracy and reliability in prediction. Summary of the Invention
[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a medium- and long-term wind power probability prediction method and system based on causal reinforcement. This method and system can accurately and reliably predict medium- and long-term wind power.
[0008] To achieve the above objectives, this invention discloses a medium- to long-term wind power probability prediction method based on causal reinforcement, comprising: Acquire meteorological data and wind power data of the wind farm, and preprocess the meteorological data and wind power data of the wind farm; The preprocessed meteorological data and wind power data of the wind farm are input into the DAG-GNN model to obtain the causal effect matrix. Based on the causal effect matrix, the core features for power prediction and the weather state discrimination features are selected. The core features of power prediction are input into the ARIMA-LSTM model to obtain the prediction of meteorological elements and uncertainty intervals for the next few days. Determine future weather conditions based on the aforementioned weather condition discrimination features; Based on the future weather conditions, the medium- and long-term wind power probability is predicted using the multi-day meteorological element point forecasts and uncertainty interval forecasts.
[0009] Furthermore, the meteorological data of the wind farm includes wind speed, wind direction, temperature, air pressure, humidity, and radiation.
[0010] Furthermore, the meteorological data and wind power data of the wind farm are processed by removing missing / outlier values, aligning timestamps, and performing minimum-maximum normalization to obtain preprocessed meteorological data and wind power data of the wind farm.
[0011] Furthermore, the DAG-GNN model uses meteorological data and wind power as nodes, and uses directed causal edges between nodes to maximize ELBO, thereby achieving joint optimization of causal structure and parameters.
[0012] Furthermore, the weather state discrimination features are input into the WGAN discrimination model to calculate the Wasserstein distance, and the future weather state is determined accordingly. The weather state includes normal state and extreme state.
[0013] Furthermore, it also includes training a first LSTM power prediction model for normal conditions and training a second LSTM power prediction model for extreme conditions.
[0014] Furthermore, the predicted meteorological elements and uncertainty intervals for the next few days are input into the corresponding first LSTM power prediction model or second LSTM power prediction model to predict the probability of wind power for the next few days.
[0015] This invention discloses a medium- to long-term wind power probability prediction system based on causal reinforcement, comprising: The acquisition module is used to acquire meteorological data and wind power data of the wind farm, and to preprocess the meteorological data and wind power data of the wind farm. The filtering module is used to input the preprocessed meteorological data and wind power data of the wind farm into the DAG-GNN model to obtain the causal effect matrix, and to filter out the core features for power prediction and weather state discrimination features based on the causal effect matrix. The first prediction module is used to input the core features of the power prediction into the ARIMA-LSTM model to obtain the prediction of meteorological element points and uncertainty intervals for the next few days. The judgment module is used to judge the future weather conditions based on the weather condition discrimination features; The second prediction module is used to predict the medium- and long-term wind power probability based on the predicted meteorological elements and uncertainty intervals for the future weather conditions.
[0016] The present invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the medium- and long-term wind power probability prediction method based on causal reinforcement.
[0017] The present invention discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the medium- and long-term wind power probability prediction method based on causal reinforcement.
[0018] The present invention has the following beneficial effects: In practical operation, the medium- and long-term wind power probability prediction method and system based on causal enhancement described in this invention inputs preprocessed meteorological data and wind power data of the wind farm into the DAG-GNN model to obtain the causal effect matrix. The core features of power prediction are then input into the ARIMA-LSTM model to obtain the prediction and uncertainty interval of meteorological elements for the next few days. Based on the prediction and uncertainty interval of meteorological elements for the next few days, the probability of medium- and long-term wind power is predicted, thereby accurately and reliably predicting medium- and long-term wind power, which is highly practical. Attached Figure Description
[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation
[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0022] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.
[0023] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0024] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.
[0025] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.
[0026] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."
[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0028] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.
[0029] As is well known, DAG-GNN (Directed Acyclic Graph-Graph Neural Network) is used to discover nonlinear causal relationships between variables and to achieve feature screening without prior assumptions.
[0030] ARIMA-LSTM: Autoregressive Integral Moving Average-Long Short-Term Memory Network, which integrates statistical time series fitting with the nonlinear capture capability of deep learning, is used for medium- and long-term autoregressive prediction of meteorological elements.
[0031] WGAN: Wasserstein Generative Adversarial Network, which uses Wasserstein distance to discriminate data distribution and is used for normal / extreme weather identification.
[0032] MAE: Mean Absolute Error, RMSE: Root Mean Square Error.
[0033] ELBO: Lower bound of evidence, the core objective function in probabilistic graphical models and variational inference, the training optimization objective of DAG-GNN, used for causal graph structure learning and parameter estimation.
[0034] PICP: Predicted Interval Coverage Probability, is the core evaluation metric for probabilistic prediction. It represents the proportion of actual measured values falling within a given prediction interval (e.g., a 90% confidence interval). A higher PICP indicates a more reliable probability prediction and a more trustworthy interval.
[0035] Example 1 refer to Figure 1 The medium- and long-term wind power probability prediction method based on causal enhancement described in this invention includes the following steps: 1) Data preprocessing module; Acquire meteorological data and wind power data of the wind farm, wherein the meteorological data of the wind farm includes wind speed, wind direction, temperature, air pressure, humidity and radiation; Missing / outlier removal, timestamp alignment, and min-max normalization are performed on the meteorological data and wind power data of the wind farm to obtain preprocessed data, and a dataset is constructed based on the preprocessed data; 2) Causal Feature Filtering Module; A DAG-GNN model is constructed, which uses meteorological data and wind power as nodes and directed causal edges between nodes to maximize ELBO and achieve joint optimization of causal structure and parameters. The preprocessed data from step 1) is input into the DAG-GNN model to obtain the causal effect matrix, which is used to quantify the direct causal contribution of each meteorological feature to wind power. Based on the causal effect matrix, the core features for power prediction and the features for weather state discrimination are obtained by screening. For example, wind speed, air pressure, and relative humidity at 70m are selected as core features for power prediction; wind speed, air pressure, relative humidity, and temperature at 70m are selected as weather condition discrimination features, redundant features are eliminated, and dimensionality and overfitting are reduced.
[0036] 3) Meteorological autoregressive prediction module; The core features of power prediction are input into the ARIMA-LSTM model to obtain the prediction of meteorological elements and uncertainty intervals for the next few days. The ARIMA-LSTM model performs differential stabilization on meteorological time series data to determine the ARIMA order (p, d, q). The ARIMA fitting residuals are input into the LSTM to learn nonlinear dynamic residual correction. The model outputs multi-day meteorological element point predictions and uncertainty intervals, replacing NWP data as input for subsequent power predictions and reducing external data errors. The weather state discrimination features are input into the WGAN discrimination model to calculate the Wasserstein distance, and the future weather is determined to be normal or extreme.
[0037] 4) Wind power probability prediction module; The first LSTM power prediction model is obtained by training under normal conditions; the second LSTM power prediction model is obtained by training under extreme conditions. The predicted meteorological elements and uncertainty intervals for the next few days are input into the corresponding first LSTM power prediction model or second LSTM power prediction model to predict the probability of wind power for the next few days.
[0038] Example 2 This embodiment uses annual measured data from an onshore wind farm in northern China as an example, and specifically includes the following steps: 1) Data preparation; Dataset: January 1, 2025 - December 31, 2025, 15-minute resolution, 21 types of meteorological features + power data, total of approximately 35,040 samples; 26,537 effective training samples and 5,920 test samples after preprocessing.
[0039] Feature list: 10 / 30 / 50 / 70m / 100m / 120m wind speed and direction, air temperature, air pressure, relative humidity, total precipitation, downward shortwave radiation from the surface, downward longwave radiation from the surface, upward shortwave radiation from the surface, upward longwave radiation from the surface, surface roughness, a total of 21 dimensions.
[0040] 2) Implementation of causal feature screening; Construct a DAG-GNN network: Input layer → Graph convolutional layer → Adjacency matrix learning layer → Output layer, trained with ELBO as the loss function to obtain the causal graph and effect matrix.
[0041] Results: 70m wind speed, air pressure, and relative humidity have the most significant causal effect on power and are identified as core input features; 70m wind speed, air pressure, relative humidity, and temperature are weather discrimination features.
[0042] 3) Implementation of meteorological autoregressive forecasting; ARIMA-LSTM models were constructed for 70m wind speed, air pressure, relative humidity, and temperature, and predictions were made every 15 minutes the following day, outputting the predicted values and 90% confidence intervals.
[0043] Comparative verification: ARIMA-LSTM has lower MAE and RMSE than Transformer, and PICP ≥ 87.5%, making predictions more accurate and intervals more reliable.
[0044] 4) Implementation of weather condition assessment; The training set was labeled with normal / extreme samples using the 3σ rule, and the WGAN generator and discriminator were trained. The weather was judged as normal if the Wasserstein distance was <0.2.
[0045] Input the predicted weather sequence and output the weather status labels for consecutive dates, such as determining August 31 as continuous normal weather.
[0046] 5) Implementation of power probability prediction; Data is categorized by weather label; normal data is input into a normal LSTM, and extreme data is input into an extreme LSTM. The output is a power prediction value and a multi-level probability range.
[0047] Results: The predicted values have a high degree of fit with the measured values, covering the vast majority of samples in 90% of the intervals, and the medium- and long-term prediction accuracy is ≥3.51% higher than that of traditional methods.
[0048] 6) Incremental calibration implementation; New data is added every 7 days, the backbone network is frozen and only the fully connected layers are fine-tuned, the model is updated online, catastrophic amnesia is avoided and the model is kept stable in the medium and long term.
[0049] Example 3 The medium- and long-term wind power probability prediction system based on causal enhancement described in this invention includes: The acquisition module is used to acquire meteorological data and wind power data of the wind farm, and to preprocess the meteorological data and wind power data of the wind farm. The filtering module is used to input the preprocessed meteorological data and wind power data of the wind farm into the DAG-GNN model to obtain the causal effect matrix, and to filter out the core features for power prediction and weather state discrimination features based on the causal effect matrix. The first prediction module is used to input the core features of the power prediction into the ARIMA-LSTM model to obtain the prediction of meteorological element points and uncertainty intervals for the next few days. The judgment module is used to judge the future weather conditions based on the weather condition discrimination features; The second prediction module is used to predict the medium- and long-term wind power probability based on the predicted meteorological elements and uncertainty intervals for the future weather conditions.
[0050] The module division in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in each embodiment of this application can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.
[0051] Example 4 A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the causal reinforcement-based medium- and long-term wind power probability prediction method. For example, the steps include: acquiring meteorological data and wind power data of a wind farm; preprocessing the meteorological data and wind power data of the wind farm; inputting the preprocessed meteorological data and wind power data of the wind farm into the DAG-GNN model to obtain a causal effect matrix; filtering core features for power prediction and weather state discrimination features based on the causal effect matrix; inputting the core features for power prediction into an ARIMA-LSTM model to obtain predictions and uncertainty intervals for meteorological elements over several days; determining future weather states based on the weather state discrimination features; and predicting the medium- and long-term wind power probability based on the predictions and uncertainty intervals for meteorological elements over several days according to the future weather states. The memory may include main memory, such as high-speed random access memory (RAM), or non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry-standard architecture bus, a peripheral component interconnection standard bus, or an extended industry-standard architecture bus. The bus can be categorized as an address bus, data bus, or control bus. The memory stores programs; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.
[0052] Example 5 A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the causal reinforcement-based medium- and long-term wind power probability prediction method. For example, the method includes: acquiring meteorological data and wind power data of a wind farm; preprocessing the meteorological data and wind power data of the wind farm; inputting the preprocessed meteorological data and wind power data of the wind farm into the DAG-GNN model to obtain a causal effect matrix; filtering the core features for power prediction and weather state discrimination features based on the causal effect matrix; inputting the core features for power prediction into an ARIMA-LSTM model to obtain predictions and uncertainty intervals for meteorological elements over several days; determining the future weather state based on the weather state discrimination features; and predicting the medium- and long-term wind power probability based on the predictions and uncertainty intervals for meteorological elements over several days. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.
[0053] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0054] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0055] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0056] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0057] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and disclosure of the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.
[0058] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.
[0059] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A medium- to long-term wind power probability prediction method based on causal reinforcement, characterized in that, include: Acquire meteorological data and wind power data of the wind farm, and preprocess the meteorological data and wind power data of the wind farm; The preprocessed meteorological data and wind power data of the wind farm are input into the DAG-GNN model to obtain the causal effect matrix. Based on the causal effect matrix, the core features for power prediction and the weather state discrimination features are selected. The core features of power prediction are input into the ARIMA-LSTM model to obtain the prediction of meteorological elements and uncertainty intervals for the next few days. Determine future weather conditions based on the aforementioned weather condition discrimination features; Based on the future weather conditions, the medium- and long-term wind power probability is predicted using the multi-day meteorological element point forecasts and uncertainty interval forecasts.
2. The medium- and long-term wind power probability prediction method based on causal reinforcement as described in claim 1, characterized in that, The meteorological data for the wind farm includes wind speed, wind direction, temperature, air pressure, humidity, and radiation.
3. The medium- and long-term wind power probability prediction method based on causal reinforcement as described in claim 1, characterized in that, Missing / outlier values are removed, timestamps are aligned, and minimum-maximum normalization is performed on the meteorological and wind power data of the wind farm to obtain preprocessed meteorological and wind power data of the wind farm.
4. The medium- and long-term wind power probability prediction method based on causal reinforcement as described in claim 1, characterized in that, The DAG-GNN model uses meteorological data and wind power as nodes, and uses directed causal edges between nodes to maximize ELBO, thereby achieving joint optimization of causal structure and parameters.
5. The medium- and long-term wind power probability prediction method based on causal reinforcement as described in claim 1, characterized in that, The weather state discrimination features are input into the WGAN discrimination model to calculate the Wasserstein distance, and the future weather state is determined accordingly. The weather state includes normal state and extreme state.
6. The medium- and long-term wind power probability prediction method based on causal reinforcement as described in claim 5, characterized in that, It also includes training a first LSTM power prediction model for normal conditions and training a second LSTM power prediction model for extreme conditions.
7. The medium- and long-term wind power probability prediction method based on causal reinforcement as described in claim 6, characterized in that, The predicted meteorological elements and uncertainty intervals for the next few days are input into the corresponding first LSTM power prediction model or second LSTM power prediction model to predict the probability of wind power for the next few days.
8. A medium- to long-term wind power probability prediction system based on causal reinforcement, characterized in that, include: The acquisition module is used to acquire meteorological data and wind power data of the wind farm, and to preprocess the meteorological data and wind power data of the wind farm. The filtering module is used to input the preprocessed meteorological data and wind power data of the wind farm into the DAG-GNN model to obtain the causal effect matrix, and to filter out the core features for power prediction and weather state discrimination features based on the causal effect matrix. The first prediction module is used to input the core features of the power prediction into the ARIMA-LSTM model to obtain the prediction of meteorological element points and uncertainty intervals for the next few days. The judgment module is used to judge the future weather conditions based on the weather condition discrimination features; The second prediction module is used to predict the medium- and long-term wind power probability based on the predicted meteorological elements and uncertainty intervals for the future weather conditions.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the medium- and long-term wind power probability prediction method based on causal enhancement as described in any one of claims 1-7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the medium- and long-term wind power probability prediction method based on causal enhancement as described in any one of claims 1-7.