Wind farm and photovoltaic power station combined power prediction method and system
By using a joint prediction model and a dynamic weight allocation algorithm, the problem of coordination in power prediction between wind farms and photovoltaic power plants was solved, improving prediction accuracy and grid dispatch adaptability, and realizing stable grid connection and efficient dispatch of new energy sources.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TAUSHGAN DARYA HYDROPOWER BRANCH OF HUANENG XINJIANG ENERGY DEVELOPMENT CO LTD
- Filing Date
- 2026-01-28
- Publication Date
- 2026-06-19
AI Technical Summary
In existing technologies, power prediction for wind farms and photovoltaic power plants does not fully consider the synergistic effect, and the prediction results lack integrity and correlation. Traditional methods fail to effectively assess the confidence level of the prediction results, resulting in poor grid connection stability of new energy sources, insufficient multi-energy coordinated dispatch capability, and low grid operation efficiency.
By acquiring operational data from wind farms and photovoltaic power plants, as well as multi-source meteorological forecast data, and inputting them into a joint prediction model for forecasting, and combining confidence assessment and dynamic weight allocation algorithms for power integration, collaborative forecasting of wind power and photovoltaics is achieved, thereby improving forecast accuracy and adaptability to grid dispatch.
It improves the accuracy of joint power prediction for wind farms and photovoltaic power plants and the stability of new energy grid connection, supports multi-energy coordinated dispatch, optimizes grid operation efficiency, and adapts to the dynamic changes in grid dispatching needs.
Smart Images
Figure CN122246678A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of new energy power plant management technology, and in particular to a method and system for joint power prediction of wind farms and photovoltaic power plants. Background Technology
[0002] With the rapid development of the new energy industry, the installed capacity of wind farms and photovoltaic power stations is constantly expanding, and their proportion in the power grid is gradually increasing. However, the output of wind power and photovoltaic power is significantly affected by weather conditions, exhibiting strong randomness and volatility, which poses a huge challenge to the safe and stable operation and efficient dispatch of the power grid.
[0003] Currently, existing technologies for predicting wind and solar power output mostly perform these predictions separately, failing to fully consider the synergistic effects between the two, resulting in a lack of overall coherence and correlation in the prediction results. Furthermore, traditional prediction methods lack effective confidence assessments of the prediction results, making it difficult to determine the reliability of the predicted values and thus affecting the rationality of subsequent power integration. In addition, power integration methods often employ fixed weight allocation, which cannot adapt to the dynamic changes in real-time grid dispatch demands. This makes it difficult for joint prediction results to meet the actual operational requirements of grid peak shaving and absorption, ultimately leading to problems such as poor stability of new energy grid connection, insufficient multi-energy coordinated dispatch capabilities, and low grid operating efficiency.
[0004] Therefore, there is an urgent need for a technical solution that can achieve joint power prediction of wind farms and photovoltaic power plants, improve prediction accuracy, and adapt to grid dispatch requirements, so as to solve the defects of the existing technologies. Summary of the Invention
[0005] This invention provides a method and system for joint power prediction of wind farms and photovoltaic power plants, which can realize joint power prediction of wind farms and photovoltaic power plants, improve prediction accuracy, and adapt to grid dispatching needs.
[0006] On the one hand, the present invention provides a method for joint power prediction of wind farms and photovoltaic power plants, which includes: Acquire wind farm operation data, photovoltaic power station operation data, and multi-source meteorological forecast data covering the areas where the wind farms and photovoltaic power stations are located; The wind farm operation data, the photovoltaic power station operation data, and the multi-source meteorological forecast data are input into a pre-constructed joint prediction model for prediction, and the predicted values of wind power active power and photovoltaic active power within a set future time are output. The confidence levels of the predicted active power of wind power and the predicted active power of photovoltaic power are assessed to obtain the assessment results. Based on the evaluation results and real-time power grid dispatch requirements, a dynamic weight allocation algorithm is used for power integration to obtain joint power prediction results.
[0007] On the other hand, the present invention also provides a combined power prediction system for wind farms and photovoltaic power plants, comprising: The acquisition module is used to acquire wind farm operation data, photovoltaic power station operation data, and multi-source meteorological forecast data covering the areas where the wind farms and photovoltaic power stations are located; The prediction module is used to input the wind farm operation data, the photovoltaic power station operation data and the multi-source meteorological forecast data into a pre-constructed joint prediction model for prediction, and output the predicted values of wind power active power and photovoltaic active power within a set future time. The evaluation module is used to evaluate the confidence level of the predicted wind power active power and the predicted photovoltaic active power, and obtain the evaluation results. The integration module is used to integrate power based on the evaluation results and real-time power grid dispatch requirements, using a dynamic weight allocation algorithm to obtain joint power prediction results.
[0008] The present invention provides a method and system for joint power prediction of wind farms and photovoltaic power plants. This method acquires wind farm operation data, photovoltaic power plant operation data, and multi-source meteorological forecast data covering the areas where both are located. These data are then input into a pre-constructed joint prediction model, which outputs predicted active power values for both wind and photovoltaic power. The predicted values are evaluated for confidence level, and combined with real-time grid dispatch requirements, a dynamic weight allocation algorithm is used for power integration to obtain the joint power prediction result. Simultaneously, the model parameters can be adaptively corrected using actual power data. This achieves coordinated power prediction for wind farms and photovoltaic power plants, improving prediction accuracy and adaptability to grid dispatch requirements. It effectively enhances the stability of new energy grid connection, supports multi-energy coordinated dispatch, and optimizes grid operation efficiency. Attached Figure Description
[0009] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0010] Figure 1 This is a flowchart illustrating the method for a combined power prediction system for wind farms and photovoltaic power plants provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the combined power prediction system for wind farms and photovoltaic power plants provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation
[0011] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this 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 this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0012] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0013] Figure 1 This is a flowchart illustrating the method for joint power prediction of wind farms and photovoltaic power plants provided in an embodiment of the present invention.
[0014] like Figure 1 As shown, the joint power prediction method for wind farms and photovoltaic power plants provided in this embodiment of the invention mainly includes the following steps: 101. Obtain wind farm operation data, photovoltaic power station operation data, and multi-source meteorological forecast data covering the areas where the wind farms and photovoltaic power stations are located; Specifically, wind farm operation data should include historical power generation, real-time wind speed, wind turbine operating status (such as speed, pitch angle, and operating time), and topographic parameters; photovoltaic power station operation data should include historical power generation, real-time solar irradiance, photovoltaic module operating status (such as temperature, conversion efficiency, and fault records), and installation tilt angle; multi-source meteorological forecast data should cover the areas where wind farms and photovoltaic power stations are located, including short-term and medium-to-long-term forecasts of key meteorological elements such as wind speed, wind direction, solar irradiance, atmospheric transparency, temperature, and precipitation. The data can be obtained through various channels such as the station monitoring system, on-site sensors, and third-party meteorological service platforms, and should undergo preprocessing such as data cleaning and format standardization to remove outliers and redundant information.
[0015] 102. Input the wind farm operation data, the photovoltaic power station operation data and the multi-source meteorological forecast data into the pre-constructed joint prediction model for prediction, and output the predicted values of wind power active power and photovoltaic active power within a set future time. In a specific implementation process, the joint prediction model needs to be able to process multi-source heterogeneous data and mine the spatiotemporal correlation characteristics of wind power and photovoltaic power output. The model is input with preprocessed wind farm operation data, photovoltaic power station operation data and multi-source meteorological forecast data. The model extracts and analyzes the data features and outputs the predicted values of wind power active power and photovoltaic active power within a future set time period (such as 1 hour, 6 hours, 24 hours, etc., which can be flexibly set according to actual dispatching needs).
[0016] Specifically, the joint prediction model can employ a dual-channel deep neural network that includes a spatiotemporal attention mechanism. One channel is used to extract the spatiotemporal features of wind power data, and the other channel is used to extract the spatiotemporal features of photovoltaic data. The features of the two channels are interacted and integrated in the fusion layer to output the predicted values of wind power active power and photovoltaic active power.
[0017] In detail, the joint prediction model employed is a dual-channel deep neural network incorporating a spatiotemporal attention mechanism. This model structure is specifically designed for the characteristics of wind power and photovoltaic data. One channel is a wind power data processing channel, dedicated to processing wind farm operation data and corresponding multi-source weather forecast data. Through the spatiotemporal attention mechanism, it focuses on spatiotemporal features that significantly affect wind power output, such as the trend of wind speed changes over different time periods and the impact of spatial correlation between different wind turbine units on output, accurately extracting key spatiotemporal features of wind power data.
[0018] The other channel is the photovoltaic data processing channel, which is used to process photovoltaic power plant operation data and corresponding multi-source weather forecast data. It also uses a spatiotemporal attention mechanism to focus on key information such as the spatiotemporal distribution characteristics of light intensity and the impact of the spatial layout of photovoltaic modules on power output, effectively extracting the core spatiotemporal characteristics of photovoltaic data.
[0019] In the fusion layer of the model, the spatiotemporal features extracted from the wind power data processing channel and the photovoltaic data processing channel are interacted and integrated to explore the potential correlation between wind power and photovoltaic features (such as the joint influence of the same weather conditions on the output of both, the complementarity of output periods, etc.). The integrated features are further processed through network structures such as fully connected layers. Finally, the predicted values of wind power active power and photovoltaic active power are output through the output layer to ensure that the predicted values can accurately reflect the actual output of wind power and photovoltaic.
[0020] 103. Conduct a confidence assessment on the predicted values of wind power active power and photovoltaic active power to obtain the assessment results; In a specific implementation process, the reliability analysis of the output wind power active power prediction value and photovoltaic active power prediction value can be carried out by combining factors such as the uncertainty of multi-source meteorological forecast data and the statistical results of historical prediction errors. The degree of fit between the prediction results and the actual output can be judged, and the corresponding confidence assessment results can be obtained, providing a reliability basis for subsequent power integration.
[0021] Specifically, this step can be achieved in the following way: a1. Based on the set of forecast values of key meteorological elements for the wind farm in the multi-source meteorological forecast data, calculate the dispersion of the set of forecast values of key meteorological elements for the wind farm as the first forecast uncertainty; a2. Based on the set of forecast values of key meteorological elements for photovoltaic power plants in the multi-source meteorological forecast data, calculate the dispersion of the set of forecast values of key meteorological elements for photovoltaic power plants as the second forecast uncertainty; Specifically, all forecast values for key meteorological elements (such as wind speed and direction) for wind farms can be selected from multi-source meteorological forecast data to form a set of key meteorological element forecast values. The dispersion (such as standard deviation and coefficient of variation) of this set is used to characterize the degree of difference between data from different forecast sources. This dispersion is the first forecast uncertainty. The larger the dispersion, the worse the consistency of the meteorological forecast results, and the more uncertain the impact on wind power prediction. Similarly, the dispersion of the forecast value set for key meteorological elements (such as solar irradiance and atmospheric transparency) for photovoltaic power plants can be calculated as the second forecast uncertainty.
[0022] a3. The first forecast uncertainty is converted into an initial confidence level for wind power through a first preset nonlinear mapping function; a4. The second forecast uncertainty is converted into an initial confidence level for photovoltaic power through a second preset nonlinear mapping function; Specifically, a first preset nonlinear mapping function and a second preset nonlinear mapping function can be pre-constructed. These two functions are obtained based on a large amount of historical data statistics and can quantify the forecast uncertainty into the corresponding initial confidence level. The first forecast uncertainty is input into the first preset nonlinear mapping function to obtain the initial confidence level of wind power; the second forecast uncertainty is input into the second preset nonlinear mapping function to obtain the initial confidence level of photovoltaic power. The higher the forecast uncertainty, the lower the corresponding initial confidence level.
[0023] a5. Based on the statistical results of prediction errors in the same period of history, the initial confidence scores of wind power and photovoltaic power are corrected to form the prediction confidence scores of wind power and photovoltaic power.
[0024] Specifically, this step can be implemented as follows: a51. Determine the forecast weather type for a future set time period based on the multi-source meteorological forecast data; a52. Based on the predicted active power of wind power and the predicted active power of photovoltaic power, determine the power range to which it belongs; a53. From the preset historical error feature library, query the historical average prediction error that matches the forecast weather type, the current season, and the power range; a54. Convert the historical average prediction error into the corresponding confidence decay factor; a55. The initial confidence level of wind power is fused with the corresponding confidence level decay factor to obtain the wind power prediction confidence level; a56. The photovoltaic initial confidence level and the corresponding confidence level decay factor are fused together to obtain the photovoltaic power prediction confidence level.
[0025] Specifically, in the confidence correction process, the characteristics of the prediction scenario are first determined. Based on the meteorological element information in the multi-source meteorological forecast data, the forecast weather type for the future set time period is determined, such as sunny, cloudy, partly cloudy, windy, and rainy. At the same time, based on the wind power active power prediction value and photovoltaic active power prediction value obtained in claim 1, combined with the installed capacity of the power station and the historical output range, the power range to which it belongs is determined, such as the high power range (e.g., 80%-100% of rated output), the medium power range (e.g., 30%-80% of rated output), and the low power range (e.g., 0%-30% of rated output).
[0026] Next, the historical average prediction error is queried. A historical error feature library is pre-built, which stores the historical average prediction errors corresponding to different combinations of forecast weather types, seasons, and power ranges. By inputting the current forecast weather type, the current season, and the determined power range, the corresponding historical average prediction error is obtained by matching and querying the feature library.
[0027] Next, a confidence decay factor is generated. The corresponding confidence decay factor is set based on the magnitude of the historical average prediction error; the larger the error, the smaller the decay factor. For example, when the historical average prediction error is within 5%, the decay factor is 1.0; when the error is between 5% and 10%, the decay factor is 0.9; when the error is between 10% and 20%, the decay factor is 0.8, and so on, ensuring that the decay factor accurately reflects the impact of historical errors on the confidence level.
[0028] Finally, confidence level fusion calculation is performed. The initial confidence level of wind power obtained in claim 2 is fused with the corresponding confidence level decay factor found in the query (e.g., multiplied) to obtain the corrected confidence level of wind power prediction; the initial confidence level of photovoltaic power is fused with the corresponding confidence level decay factor to obtain the corrected confidence level of photovoltaic power prediction, thereby achieving accurate confidence level correction based on specific prediction scenarios.
[0029] This embodiment refines forecast weather types, seasons, and power ranges, accurately matches historical average forecast errors, and converts them into attenuation factors to correct the initial confidence level. This makes the confidence level assessment more scenario-specific, effectively avoids assessment bias caused by statistical analysis of single historical errors, and further improves the accuracy and reliability of the confidence level assessment.
[0030] 104. Based on the evaluation results and real-time power grid dispatch requirements, a dynamic weight allocation algorithm is used to integrate power and obtain joint power prediction results.
[0031] In a specific implementation process, this step can be implemented in the following way: b1. Based on the evaluation results, generate the basic weights for wind power reliability and photovoltaic reliability; The obtained confidence assessment results (wind power prediction confidence and photovoltaic power prediction confidence) can be used as the core basis to generate wind power reliability base weights and photovoltaic reliability base weights. The confidence level and the reliability base weights are positively correlated, that is, the higher the prediction confidence of a certain power source, the greater its corresponding reliability base weight, so as to ensure that the weights can reflect the reliability of the prediction results.
[0032] b2. Generate wind power strategy weights and photovoltaic strategy weights based on the real-time grid dispatch requirements; The system can analyze real-time grid dispatch demands to identify key indicators such as the grid's current wind power absorption capacity, photovoltaic absorption capacity, peak-shaving demand, and power supply stability requirements. Based on these indicators, an appropriate weight allocation strategy can be formulated to generate wind power strategy weights and photovoltaic strategy weights. For example, when the grid's wind power absorption capacity is strong and peak-shaving demand is low, the wind power strategy weight can be appropriately increased; when there is sufficient photovoltaic absorption capacity and priority must be given to ensuring the absorption of clean energy, the photovoltaic strategy weight can be increased accordingly.
[0033] Specifically, this step can be implemented as follows: analyze the real-time grid dispatch demand to obtain the wind power consumption preference coefficient and the photovoltaic consumption preference coefficient; obtain the current peak-shaving demand intensity, and adjust the wind power consumption preference coefficient and the photovoltaic consumption preference coefficient in a contextualized manner based on the power output characteristics of wind power and photovoltaic; calculate the wind power strategy weight and the photovoltaic strategy weight based on the adjusted wind power consumption preference coefficient and photovoltaic consumption preference coefficient.
[0034] In detail, real-time grid dispatch instructions, dispatch plans and other related information can be analyzed to extract parameters that reflect the grid's priority for wind power and photovoltaic power consumption, and wind power consumption preference coefficient and photovoltaic power consumption preference coefficient can be obtained. The coefficient value range can be set to 0-1. The larger the coefficient, the higher the grid's preference for consuming the power source.
[0035] Next, the current peak-shaving demand intensity of the power grid is obtained through the power grid operation monitoring system. This demand intensity can be quantified (e.g., high, medium, and low levels) based on indicators such as grid load factor and peak-valley difference. Simultaneously, considering the power output characteristics of wind and solar power—wind power output is characterized by large fluctuations and relatively stable output at night, while solar power output is characterized by diurnal intermittency and significant influence from sunlight intensity—targeted adjustments can be made to the wind power absorption preference coefficient and the solar power absorption preference coefficient. For example, during periods of high peak-shaving demand, the wind power absorption preference coefficient can be appropriately reduced because fluctuations in wind power output may exacerbate grid peak-shaving pressure. During periods of abundant sunlight and low grid load, excessive solar power output may lead to curtailment, so the solar power absorption preference coefficient can be appropriately reduced. During periods when the grid needs to increase clean energy supply, the absorption preference coefficients of both can be increased simultaneously.
[0036] Then, the adjusted wind power consumption preference coefficient and photovoltaic consumption preference coefficient are normalized to ensure that the sum of the two is 1. The processed coefficients are the wind power strategy weight and photovoltaic strategy weight, so that the strategy weight can accurately match the current grid operation situation and dispatch requirements.
[0037] This embodiment obtains the absorption preference coefficient by analyzing the grid dispatch demand, and makes contextual adjustments based on the peak-shaving demand intensity and power output characteristics. The generated strategy weights can closely match the actual operating state and dispatch demand of the grid, avoiding the problem of insufficient adaptability caused by fixed strategy weights, improving the rationality and flexibility of the comprehensive weights, and thus optimizing the adaptability of the joint power prediction results to grid dispatch.
[0038] b3. The wind power reliability base weight and the wind power strategy weight are combined to obtain the wind power comprehensive weight, and the photovoltaic reliability base weight and the photovoltaic strategy weight are combined to obtain the photovoltaic comprehensive weight; Specifically, this step can be performed as follows: The first integrated intermediate value of wind power is calculated as the product of the basic weight of wind power reliability and the weight of wind power strategy, and the first integrated intermediate value of photovoltaic power is calculated as the product of the basic weight of photovoltaic reliability and the weight of photovoltaic strategy. The first integrated intermediate values of wind power and photovoltaic power are normalized to obtain the comprehensive weight of wind power and the comprehensive weight of photovoltaic power.
[0039] In detail, the basic weight of wind power reliability is multiplied by the weight of wind power strategy to obtain the first integrated intermediate value of wind power. This intermediate value reflects both the reliability of wind power forecasting and the grid's dispatch strategy requirements for wind power. Similarly, the basic weight of photovoltaic reliability is multiplied by the weight of photovoltaic strategy to obtain the first integrated intermediate value of photovoltaic, thus achieving the initial integration of photovoltaic reliability and dispatch strategy.
[0040] The sum of the first convergence median values for wind power and photovoltaics is calculated. Then, each of the first convergence median values for wind power and photovoltaics is divided by this sum to obtain the normalized comprehensive weights for wind power and photovoltaics. This ensures that the sum of the comprehensive weights for wind power and photovoltaics is 1, conforming to the basic logic of weight allocation. Through normalization, the comprehensive weights can reflect the combined impact of reliability and strategy requirements on a unified dimension, providing a standardized and reasonable weighting basis for subsequent power integration.
[0041] b4. Based on the wind power comprehensive weight, the photovoltaic comprehensive weight, the wind power active power prediction value, and the photovoltaic active power prediction value, generate the joint power prediction result.
[0042] Specifically, this step can be achieved as follows: The product of the wind power comprehensive weight and the wind power active power prediction value is calculated to obtain the wind power weighted power; the product of the photovoltaic comprehensive weight and the photovoltaic active power prediction value is calculated to obtain the photovoltaic weighted power; the wind power weighted power and the photovoltaic weighted power are summed to generate the joint power prediction result.
[0043] In detail, the obtained wind power comprehensive weight can be multiplied by the obtained wind power active power prediction value to obtain the wind power weighted power. This value reflects the degree of contribution of wind power in the joint prediction results. The higher the comprehensive weight, the greater the contribution. Similarly, the photovoltaic comprehensive weight can be multiplied by the photovoltaic active power prediction value to obtain the photovoltaic weighted power, which reflects the degree of contribution of photovoltaic.
[0044] After obtaining the wind power weighted power, the wind power weighted power and the photovoltaic power weighted power can be added together, and the sum is the joint power prediction result. In this way, the predicted power of wind power and photovoltaic power can be quickly integrated. At the same time, by adjusting the comprehensive weights, the contribution ratio of wind power and photovoltaic power can be dynamically adjusted according to reliability and grid dispatch requirements, ensuring the rationality and practicality of the results.
[0045] This embodiment of the joint power prediction method for wind farms and photovoltaic power plants acquires wind farm operation data, photovoltaic power plant operation data, and multi-source meteorological forecast data covering the areas where both are located. These data are then input into a pre-constructed joint prediction model, which outputs predicted active power values for both wind and photovoltaic power. The predicted values are then evaluated for confidence level, and combined with real-time grid dispatch requirements, a dynamic weight allocation algorithm is used for power integration to obtain the joint power prediction result. Simultaneously, the model parameters can be adaptively corrected using actual power data. This application achieves coordinated power prediction for wind farms and photovoltaic power plants, improving prediction accuracy and adaptability to grid dispatch requirements, effectively enhancing the stability of new energy grid connection, supporting multi-energy coordinated dispatch, and optimizing grid operation efficiency.
[0046] In some embodiments, after obtaining the joint power prediction result, the method further includes: c1. Collect the actual active power of wind farms and photovoltaic power plants within a future set time period; Specifically, actual power data can be collected. After the set time period corresponding to the joint forecast results ends, the actual active power of the wind farm and the actual active power of the photovoltaic power station during that time period are collected through the operation monitoring system of the wind farm and the photovoltaic power station to ensure the authenticity and completeness of the data.
[0047] c2. Calculate the real-time error sequence of wind power prediction based on the actual active power of the wind farm and the predicted active power of the wind power; calculate the real-time error sequence of photovoltaic power prediction based on the actual active power of the photovoltaic power station and the predicted active power of the photovoltaic power station. Specifically, the actual active power of the wind farm collected can be compared with the corresponding predicted active power of the wind power in each time period, and the prediction error (such as absolute error, relative error, etc.) in each time period can be calculated to form a real-time error sequence for wind power prediction. Similarly, the actual active power of the photovoltaic power station can be compared with the predicted active power of the photovoltaic power station to calculate the real-time error sequence for photovoltaic power prediction.
[0048] c3. Perform joint analysis on the real-time error sequence of the wind power prediction and the real-time error sequence of the photovoltaic power prediction to determine the statistical characteristics and spatiotemporal correlation pattern of the joint prediction error. Specifically, joint analysis can be performed on the real-time error sequences of wind power and photovoltaic power prediction to uncover the statistical characteristics of the joint prediction error, such as the mean, variance, maximum and minimum values of the error. At the same time, the spatiotemporal correlation patterns of the error can be analyzed, such as the degree of correlation of errors in different time periods and the mutual influence relationship between errors of wind farms and photovoltaic power plants, so as to fully grasp the distribution law and changing trend of prediction error.
[0049] c4. When the statistical characteristics of the joint prediction error exceed the preset stability threshold, a dynamic adjustment command for the model parameters is triggered. Specifically, a stability threshold for the joint prediction error (such as the maximum allowable value of the error variance) can be preset. The statistical characteristics of the joint prediction error obtained from the analysis are compared with this threshold. When the statistical characteristics exceed the threshold, it indicates that the model prediction accuracy has decreased and cannot meet the actual needs. At this time, the dynamic adjustment command of the model parameters is triggered. If the threshold is not exceeded, the model continues to run with the current parameters.
[0050] c5. Based on the dynamic adjustment instructions for the model parameters and the spatiotemporal correlation mode, adaptively correct the internal parameters of the pre-constructed joint prediction model.
[0051] Specifically, the instructions can be dynamically adjusted based on the model parameters. Combined with the previously analyzed spatiotemporal correlation patterns of errors, the internal parameters of the pre-constructed joint prediction model (such as the weights, biases, activation function parameters, etc. of the neural network) can be adaptively corrected. The adjustment direction should be aimed at the cause of the error. For example, if the error is large in a certain period and is related to changes in meteorological elements, the parameters of the corresponding meteorological feature extraction part in the model can be adjusted to improve the prediction accuracy of the model for this type of scenario.
[0052] Based on the same general inventive concept, this invention also protects a combined power prediction system for wind farms and photovoltaic power plants. The combined power prediction system for wind farms and photovoltaic power plants provided by this invention will be described below. The combined power prediction system for wind farms and photovoltaic power plants described below can be referred to in correspondence with the combined power prediction method for wind farms and photovoltaic power plants described above.
[0053] Figure 2 This is a schematic diagram of the structure of the combined power prediction system for wind farms and photovoltaic power plants provided in an embodiment of the present invention, as shown below. Figure 2 As shown, the wind farm and photovoltaic power station joint power prediction system of this embodiment includes an acquisition module 21, a prediction module 22, an evaluation module 23 and an integration module 24.
[0054] The acquisition module 21 is used to acquire wind farm operation data, photovoltaic power station operation data, and multi-source meteorological forecast data covering the areas where the wind farm and photovoltaic power station are located. Prediction module 22 is used to input the wind farm operation data, the photovoltaic power station operation data and the multi-source meteorological forecast data into a pre-constructed joint prediction model for prediction, and output the predicted values of wind power active power and photovoltaic active power within a set future time. Evaluation module 23 is used to evaluate the confidence level of the predicted wind power active power and the predicted photovoltaic active power, and obtain the evaluation result; The integration module 24 is used to integrate power based on the evaluation results and real-time power grid dispatch requirements, using a dynamic weight allocation algorithm to obtain joint power prediction results.
[0055] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. The electronic device may include: a processor 310, a communication interface 320, a memory 330, and a communication bus 340. The processor 310, communication interface 320, and memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions stored in the memory 330 to execute a joint power prediction method for wind farms and photovoltaic power plants.
[0056] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0057] It should be noted that all relevant information that may be involved in the various embodiments of the present invention is processed in strict accordance with the requirements of laws and regulations, following the principles of legality, legitimacy, and necessity, based on the reasonable purpose of the business scenario, and is information that users actively provide or generate during the use of the product / service, as well as information obtained with user authorization.
[0058] The information processed by this invention may vary depending on the specific product / service scenario and should be based on the specific scenario in which the user uses the product / service. This may involve user account information, device information, or other related information. This invention will treat the relevant information and its processing with the utmost diligence.
[0059] This invention places great importance on the security of related information and has adopted reasonable and feasible security protection measures that comply with industry standards to protect related information and prevent unauthorized access, public disclosure, use, modification, damage or loss of related information.
[0060] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0061] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for joint power prediction of wind farms and photovoltaic power plants, characterized in that, include: Acquire wind farm operation data, photovoltaic power station operation data, and multi-source meteorological forecast data covering the areas where the wind farms and photovoltaic power stations are located; The wind farm operation data, the photovoltaic power station operation data, and the multi-source meteorological forecast data are input into a pre-constructed joint prediction model for prediction, and the predicted values of wind power active power and photovoltaic active power within a set future time are output. The confidence levels of the predicted active power of wind power and the predicted active power of photovoltaic power are assessed to obtain the assessment results. Based on the evaluation results and real-time power grid dispatch requirements, a dynamic weight allocation algorithm is used for power integration to obtain joint power prediction results.
2. The method for joint power prediction of wind farms and photovoltaic power plants according to claim 1, characterized in that, The confidence levels of the predicted active power of wind power and photovoltaic power are assessed to obtain the assessment results, including: Based on the set of forecast values of key meteorological elements for the wind farm in the multi-source meteorological forecast data, the dispersion of the set of forecast values of key meteorological elements for the wind farm is calculated as the first forecast uncertainty. Based on the set of forecast values for key meteorological elements for photovoltaic power plants in the multi-source meteorological forecast data, the dispersion of the set of forecast values for key meteorological elements for photovoltaic power plants is calculated as the second forecast uncertainty; The first forecast uncertainty is converted into an initial confidence level for wind power through a first preset nonlinear mapping function; The second forecast uncertainty is converted into an initial confidence level for photovoltaics using a second preset nonlinear mapping function; Based on the statistical results of prediction errors in the same period of history, the initial confidence scores of wind power and photovoltaic power are corrected to form the prediction confidence scores of wind power and photovoltaic power.
3. The method for joint power prediction of wind farms and photovoltaic power plants according to claim 2, characterized in that, Based on the statistical results of prediction errors from the same historical period, the initial confidence levels of wind power and photovoltaic power are corrected to form the predicted confidence levels of wind power and photovoltaic power, including: The forecast weather type for a future set time period is determined based on the multi-source meteorological forecast data. Based on the predicted active power values of wind power and photovoltaic power, determine the power range to which it belongs; From the preset historical error feature database, query the historical average prediction error that matches the forecast weather type, the current season, and the power range; The historical average prediction error is converted into a corresponding confidence decay factor. The wind power prediction confidence level is obtained by fusing the initial confidence level of the wind power with the corresponding confidence level decay factor. The photovoltaic power prediction confidence level is obtained by fusing the initial confidence level of the photovoltaic power with the corresponding confidence level decay factor.
4. The method for joint power prediction of wind farms and photovoltaic power plants according to claim 1, characterized in that, Based on the evaluation results and real-time power grid dispatch requirements, a dynamic weight allocation algorithm is used for power integration to obtain joint power prediction results, including: Based on the evaluation results, wind power reliability base weights and photovoltaic reliability base weights are generated. Based on the real-time grid dispatch requirements, wind power strategy weights and photovoltaic strategy weights are generated. The wind power reliability base weight and the wind power strategy weight are combined to obtain the wind power comprehensive weight, and the photovoltaic reliability base weight and the photovoltaic strategy weight are combined to obtain the photovoltaic comprehensive weight; The joint power prediction result is generated based on the wind power comprehensive weight, the photovoltaic comprehensive weight, the wind power active power prediction value, and the photovoltaic active power prediction value.
5. The method for joint power prediction of wind farms and photovoltaic power plants according to claim 4, characterized in that, Based on the real-time grid dispatch requirements, wind power strategy weights and photovoltaic strategy weights are generated, including: The real-time grid dispatch demand is analyzed to obtain the wind power consumption preference coefficient and the photovoltaic consumption preference coefficient; The current peak-shaving demand intensity is obtained, and the wind power absorption preference coefficient and the photovoltaic absorption preference coefficient are adjusted in a contextualized manner based on the power output characteristics of wind power and photovoltaic power. Based on the adjusted wind power consumption preference coefficient and photovoltaic consumption preference coefficient, the wind power strategy weight and the photovoltaic strategy weight are calculated.
6. The method for joint power prediction of wind farms and photovoltaic power plants according to claim 4, characterized in that, The process of fusing the basic wind power reliability weights and the wind power strategy weights to obtain a comprehensive wind power weight, and fusing the basic photovoltaic reliability weights and the photovoltaic strategy weights to obtain a comprehensive photovoltaic weight, includes: The first integrated intermediate value for wind power is calculated as the product of the basic weight of wind power reliability and the weight of wind power strategy, and the first integrated intermediate value for photovoltaic power is calculated as the product of the basic weight of photovoltaic reliability and the weight of photovoltaic strategy. The first integrated intermediate value of wind power and the first integrated intermediate value of photovoltaic power are normalized to obtain the comprehensive weight of wind power and the comprehensive weight of photovoltaic power.
7. The method for joint power prediction of wind farms and photovoltaic power plants according to claim 4, characterized in that, Based on the wind power comprehensive weight, the photovoltaic comprehensive weight, the predicted wind power active power value, and the predicted photovoltaic active power value, the joint power prediction result is generated, including: The weighted power of wind power is obtained by multiplying the comprehensive weight of wind power with the predicted value of wind power active power. The photovoltaic weighted power is obtained by multiplying the photovoltaic comprehensive weight by the photovoltaic active power prediction value. The weighted power of wind power and the weighted power of photovoltaic power are summed to generate the joint power prediction result.
8. The method for joint power prediction of wind farms and photovoltaic power plants according to any one of claims 1-7, characterized in that, After obtaining the joint power prediction results, the method further includes: Collect the actual active power of wind farms and photovoltaic power plants within a future set time period; Based on the actual active power of the wind farm and the predicted active power of the wind power, calculate the real-time error sequence of wind power prediction; based on the actual active power of the photovoltaic power station and the predicted active power of the photovoltaic power station, calculate the real-time error sequence of photovoltaic power prediction. The real-time error sequences of the wind power prediction and the photovoltaic power prediction are jointly analyzed to determine the statistical characteristics and spatiotemporal correlation patterns of the joint prediction error. When the statistical characteristics of the joint prediction error exceed a preset stability threshold, a dynamic adjustment command for the model parameters is triggered. Based on the dynamic adjustment instructions for model parameters and the spatiotemporal correlation pattern, the internal parameters of the pre-constructed joint prediction model are adaptively corrected.
9. The method for joint power prediction of wind farms and photovoltaic power plants according to any one of claims 1-7, characterized in that, The joint prediction model employs a dual-channel deep neural network with a spatiotemporal attention mechanism. One channel is used to extract the spatiotemporal features of wind power data, and the other channel is used to extract the spatiotemporal features of photovoltaic data. The features of the two channels are interacted and integrated in the fusion layer to output the predicted values of wind power active power and photovoltaic active power.
10. A combined power prediction system for wind farms and photovoltaic power plants, characterized in that, include: The acquisition module is used to acquire wind farm operation data, photovoltaic power station operation data, and multi-source meteorological forecast data covering the areas where the wind farms and photovoltaic power stations are located; The prediction module is used to input the wind farm operation data, the photovoltaic power station operation data and the multi-source meteorological forecast data into a pre-constructed joint prediction model for prediction, and output the predicted values of wind power active power and photovoltaic active power within a set future time. The evaluation module is used to evaluate the confidence level of the predicted wind power active power and the predicted photovoltaic active power, and obtain the evaluation results. The integration module is used to integrate power based on the evaluation results and real-time power grid dispatch requirements, using a dynamic weight allocation algorithm to obtain joint power prediction results.