Wind-solar power generation power combined prediction method and platform

Abstract

The invention discloses a wind-solar power generation power combined prediction method and platform, and relates to the technical field of power management, and the method comprises the steps: collecting multi-dimensional power generation data of a wind power plant and a photovoltaic power plant in real time based on a data sensing assembly, and extracting multi-dimensional prediction features after preprocessing; historical wind and light power generation data are collected, a combined prediction model is generated through training, the multi-dimensional prediction features are used as input for combined prediction, and predicted wind and light power generation power is obtained; and meanwhile, the deviation between the actual wind-solar power generation power and the predicted wind-solar power generation power is monitored in real time, and the combined prediction model is dynamically optimized. The method solves the technical problem that the existing method cannot effectively combine the synergistic effect of wind power and photovoltaic power generation and lacks a self-adaptive updating mechanism in the face of environment change, so that the prediction accuracy is insufficient, and achieves the purpose that the prediction accuracy is low through combined modeling of the wind-solar power generation power and introduction of a dynamic optimization mechanism. And the prediction precision and the adaptive capability are obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power management, and in particular to a method and platform for joint prediction of wind and solar power generation. Background Art

[0002] Wind power and photovoltaic power generation are currently the most widely used forms of renewable energy, but due to their high dependence on meteorological conditions (such as wind speed, temperature, radiation intensity, etc.), the power generation is highly volatile and uncertain. Therefore, accurately predicting wind and photovoltaic power generation has become a key challenge to optimize grid scheduling and improve energy efficiency.

[0003] At present, the power prediction of wind power generation and photovoltaic power generation has become a hot topic of research. Traditional single wind or photovoltaic power prediction methods usually only consider the influencing factors of a single energy source, ignoring the linkage between wind power and photovoltaic power generation in the same period and their joint effect on power output, and failing to effectively deal with the impact of environmental changes. Summary of the invention

[0004] The present application provides a wind and photovoltaic power joint prediction method and platform, which is used to solve the technical problems that existing methods cannot effectively combine the synergistic effects of wind and photovoltaic power generation, and lack an adaptive update mechanism when facing environmental changes, resulting in insufficient prediction accuracy.

[0005] In a first aspect of the present application, a method for joint prediction of wind and photovoltaic power generation is provided, the method comprising: based on a data sensing component, real-time collection of multi-dimensional power generation data of wind farms and photovoltaic power farms; data preprocessing of the multi-dimensional power generation data to generate benchmark power generation data; key feature extraction using the benchmark power generation data, and obtaining multi-dimensional prediction features through feature correlation analysis; collecting historical wind and photovoltaic power generation data, training and generating a joint prediction model; using the multi-dimensional prediction features as input, using the joint prediction model for joint prediction, and obtaining predicted wind and photovoltaic power generation; real-time monitoring of the deviation between the actual wind and photovoltaic power generation power and the predicted wind and photovoltaic power generation power, and dynamic optimization of the joint prediction model.

[0006] The second aspect of the present application provides a wind-solar power joint prediction platform, the platform comprising: a data acquisition module, the data acquisition module is used to collect multi-dimensional power generation data of wind farms and photovoltaic power plants in real time based on data sensing components; a data preprocessing module, the data preprocessing module is used to perform data preprocessing on the multi-dimensional power generation data to generate benchmark power generation data; a feature extraction module, the feature extraction module is used to use the benchmark power generation data to extract key features, and obtain multi-dimensional prediction features through feature correlation analysis; a prediction model training module, the prediction model training module is used to collect historical wind-solar power generation data, train and generate a joint prediction model; a joint prediction module, the joint prediction module is used to use the multi-dimensional prediction features as input, use the joint prediction model to perform joint prediction, and obtain predicted wind-solar power generation; a dynamic optimization module, the dynamic optimization module is used to monitor the deviation between the actual wind-solar power generation and the predicted wind-solar power generation in real time, and dynamically optimize the joint prediction model.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The present application provides a wind-solar power joint prediction method and platform, which relate to the field of power management technology. The multi-dimensional power generation data of wind and photovoltaic power plants are collected in real time, key feature extraction and correlation analysis are performed after pre-processing to obtain multi-dimensional prediction features, and a joint prediction model is trained based on historical data. The extracted features are used for joint prediction to obtain wind-solar power prediction results. At the same time, the deviation between the actual power generation and the predicted value is monitored in real time, and the model parameters are dynamically optimized to improve the prediction accuracy and adaptability. The technical problems that the existing methods cannot effectively combine the synergistic effects of wind and photovoltaic power generation and lack an adaptive update mechanism in the face of environmental changes, resulting in insufficient prediction accuracy, are solved. The technical effect of significantly improving the prediction accuracy and adaptability is achieved by jointly modeling wind and photovoltaic power and introducing a dynamic optimization mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A schematic diagram of a wind-solar power generation joint prediction method provided in an embodiment of the present application;

[0011] Figure 2A schematic diagram of the structure of a wind-solar power generation joint prediction platform provided in an embodiment of the present application.

[0012] Explanation of the accompanying drawings: data acquisition module 11, data preprocessing module 12, feature extraction module 13, prediction model training module 14, joint prediction module 15, dynamic optimization module 16. DETAILED DESCRIPTION

[0013] The present application provides a method and platform for joint prediction of wind and photovoltaic power generation, which is used to solve the technical problems that existing methods in the prior art cannot effectively combine the synergistic effects of wind and photovoltaic power generation, and lack an adaptive update mechanism when facing environmental changes, resulting in insufficient prediction accuracy.

[0014] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0015] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices.

[0016] Embodiment 1, as Figure 1 As shown, the present application provides a method for jointly predicting wind and solar power generation, the method comprising:

[0017] P10: Based on data sensing components, multi-dimensional power generation data of wind farms and photovoltaic power plants are collected in real time.

[0018] Specifically, by deploying a series of high-precision data sensor components, multi-dimensional data of wind farms and photovoltaic power plants are collected in real time. These data include key power generation parameters, environmental factors, and equipment status information of wind and photovoltaic power generation.

[0019] First, wind speed sensors (such as ultrasonic anemometers) and wind direction sensors are deployed at wind farms. These sensors are used to measure the strength and direction of the wind, and wind speed is one of the main factors affecting wind power generation. There is a direct linear relationship between wind speed and power generation - the greater the wind speed, the higher the power generation is usually, so real-time monitoring of wind speed is crucial. Other environmental data, such as temperature, air pressure, and humidity, are also collected through weather stations because these meteorological factors will indirectly affect the efficiency and stability of wind power generation.

[0020] Secondly, the photovoltaic power plant deploys radiation sensors and temperature sensors to monitor the solar radiation intensity and the temperature of the photovoltaic panels in real time. Radiation intensity is the core parameter that affects the power of photovoltaic power generation. The stronger the light intensity, the higher the power generation capacity of the photovoltaic cell. In addition, the temperature sensor is used to monitor the operating temperature of the photovoltaic panel. Excessive temperature will reduce the efficiency of the photovoltaic cell. Therefore, timely understanding of the working status of the photovoltaic panel is also very important for optimizing the power generation efficiency.

[0021] In addition to meteorological and environmental data, the operating status and performance parameters of power generation equipment are monitored simultaneously, such as the operating speed of wind turbines, the operating voltage and current of photovoltaic panels, etc. These equipment status data can reflect the healthy status of system operation, help to timely discover potential failure risks, and thus ensure the reliability of the system.

[0022] The above collected data can be uploaded to the cloud platform or local data storage unit in real time through the data acquisition system (such as SCADA system or IoT sensor network). These data will serve as the basis for subsequent data processing, feature extraction and model training, ensuring that the prediction system can obtain high-quality, real-time updated input data.

[0023] P20: Perform data preprocessing on the multi-dimensional power generation data to generate reference power generation data.

[0024] Furthermore, step P20 of the embodiment of the present application further includes:

[0025] P21: The multi-dimensional power generation data includes real-time power output data, meteorological data and geographic information data of wind farms and photovoltaic power plants; P22: The real-time power output data is cleaned, denoised, and missing values ​​are filled, the meteorological data is standardized, and the geographic information data is position calibrated and normalized to generate the benchmark power generation data.

[0026] Optionally, a series of data preprocessing operations are performed on the multi-dimensional data collected in real time from the wind farm and photovoltaic farm to improve the quality and accuracy of the data, thereby providing a reliable basis for subsequent analysis and prediction.

[0027] The multi-dimensional data collected include real-time power output data, meteorological data and geographic information data of wind farms and photovoltaic farms. Real-time power output data reflects the actual power generation capacity of each power farm, while meteorological data (such as wind speed, temperature, humidity, radiation intensity, etc.) and geographic information data (such as the installation location of photovoltaic panels, the geographic coordinates of the power farm, etc.) provide environmental factors and site characteristics that affect power generation efficiency and power output. Since these data come from different sources and their formats and accuracy vary greatly, they must be uniformly processed and optimized.

[0028] First, the real-time power output data is cleaned and denoised. The power output data may have abnormal values ​​or noise data due to equipment failure, communication delay or sensor error, which will seriously affect the subsequent analysis and modeling results. To this end, a reasonable threshold range is set (for example, the point where the power value exceeds the maximum or minimum value is abnormal data) to remove or correct it. At the same time, filtering algorithms (such as median filtering, Kalman filtering, etc.) are used to further smooth the power data, remove high-frequency noise, and ensure the stability and reliability of the power output data.

[0029] In practical applications, there may be missing values ​​in the collected data, especially in meteorological data or geographic information data. In order to avoid the impact of missing data on model training or prediction, interpolation methods (such as linear interpolation, spline interpolation, etc.) are used to fill in the missing data to ensure the continuity and integrity of the data. Meteorological data (such as wind speed, temperature, radiation intensity, etc.) often have different dimensions and distributions. In order to eliminate these differences and make the data comparable under the same dimension, standardization is performed. For example, the z-score standardization method is used to convert the data into a standard normal distribution with zero mean and unit variance, so that different types of data are comparable and avoid a certain feature dominating the model training process due to its large scale.

[0030] Geographic information data usually involves information such as the location coordinates of the power plant. This information may lead to deviations in the location data due to factors such as equipment installation errors and differences in sensor accuracy. Therefore, during the preprocessing process, the geographic information data needs to be calibrated to ensure the accuracy of the geographic coordinates. In addition, normalization is performed to standardize the range of geographic data, such as normalizing the longitude and latitude coordinates to the [0,1] interval, to avoid the location data from having too much or too little impact on the model training process.

[0031] Through the above preprocessing steps, the final generated benchmark power generation data not only has high data quality, but also is unified in structure, and can provide accurate and reliable input data for subsequent feature extraction, modeling and prediction steps.

[0032] P30: The benchmark power generation data is used to extract key features, and multi-dimensional prediction features are obtained through feature correlation analysis.

[0033] Furthermore, step P30 of the embodiment of the present application further includes:

[0034] P31: Perform correlation analysis on the benchmark power generation data to screen multiple power output characteristics, multiple meteorological characteristics and multiple geographical characteristics that are closely related to the power output of wind power and photovoltaic power generation; P32: Use principal component analysis to reduce the dimension of the multiple power output characteristics, multiple meteorological characteristics and multiple geographical characteristics, remove redundant characteristics, and generate multi-dimensional prediction features.

[0035] It should be understood that by extracting key features from the preprocessed benchmark power generation data, features that have a significant impact on wind power generation and photovoltaic power generation output are further screened out. These features will provide high-quality input data for the subsequent joint prediction model.

[0036] First, based on the benchmark power generation data, correlation analysis is performed to identify and screen multiple features that are closely related to wind and photovoltaic power output. Correlation analysis is to find factors that are closely related to power output changes by calculating the correlation coefficient between each feature and the target variable (i.e., power output), including multiple power output features, multiple meteorological features, and multiple geographical features.

[0037] Specifically, the power output features mainly include real-time wind power and photovoltaic power data. These features directly reflect the actual power generation capacity of the power plant and are the core data of the prediction model. Meteorological features include environmental factors such as wind speed, temperature, humidity, and radiation intensity, which have a certain relationship with power generation. For example, when the wind speed is high, the wind power generation is usually high; the solar radiation intensity is positively correlated with the photovoltaic power generation. Through correlation analysis, the system can automatically identify which meteorological factors have the greatest impact on the power generation, so as to select the most representative meteorological features for the model. Geographical features include the geographical location of wind farms and photovoltaic farms, the layout angle of photovoltaic panels, the height of wind turbines, etc. These geographical factors will affect the distribution of wind speed and the intensity of light, so they also need to be considered in feature screening. Through correlation analysis, the system screens out geographical features that have a greater impact on power generation.

[0038] After the feature screening is completed, the principal component analysis (PCA) method is used to reduce the dimension of the selected multiple power output features, meteorological features, and geographical features. Principal component analysis is a commonly used dimensionality reduction technique that can merge multiple features into a few principal components through linear transformation while retaining most of the information. The purpose of dimensionality reduction is to remove redundant features, reduce computational complexity, and improve model performance.

[0039] Through principal component analysis, several important principal components are extracted from the original multiple features. These principal components are comprehensive features that have the greatest explanatory power for power output changes. This not only simplifies the data, but also improves the expressive power of the features, allowing subsequent prediction models to be trained and predicted more efficiently and accurately.

[0040] P40: Collect historical wind and solar power generation data and train and generate a joint prediction model.

[0041] Furthermore, step P40 of the embodiment of the present application also includes:

[0042] P41: Based on historical wind and solar power generation records, historical wind and solar power generation data are collected to construct a model training data set, which includes a meteorological feature data set, a geographic feature data set and a corresponding power output data set; P42: Combined with a deep learning model, the model training data set is used as training data to perform model training and generate the joint prediction model.

[0043] Optionally, historical wind and solar power generation data is collected and combined with deep learning methods to train and generate a model for joint prediction.

[0044] First, based on the historical wind and solar power generation records, multi-dimensional data of historical wind farms and photovoltaic power plants are collected, including meteorological feature data sets, geographical feature data sets, and corresponding power output data sets. Among them, the meteorological feature data set includes historical meteorological factors such as wind speed, temperature, humidity, and radiation intensity. These data reflect the impact of external environmental changes on wind and photovoltaic power generation. The geographical feature data set includes geographical factors that affect power generation efficiency, such as the geographical location, terrain, and photovoltaic panel installation angle of the power plant. The power output data set includes historical wind power generation and photovoltaic power generation, which are the key target variables for training the model. By using these historical data as training samples, the intrinsic relationship between meteorological conditions, geographical features, and power generation can be learned. After unified standardization, these data sets form a complete model training data set.

[0045] Furthermore, based on the training data set, the system uses a deep learning model for training. Here, deep learning methods such as long short-term memory networks (LSTM) and convolutional neural networks (CNN) can be used to capture the spatiotemporal variation of wind and solar power generation through learning from historical data, so as to effectively perform joint predictions. During the training process, the model continuously optimizes its weights and parameters to minimize the prediction error. Through the back propagation algorithm, the weights of each layer of the model are adjusted according to the error, so that the prediction results are closer and closer to the actual wind and solar power generation power output. After the training is completed, the generated joint prediction model can accurately predict the power output of wind and solar power generation under different meteorological and geographical conditions.

[0046] P50: Taking the multi-dimensional prediction features as input, using the joint prediction model to perform joint prediction to obtain the predicted wind and solar power generation power.

[0047] Furthermore, step P50 of the embodiment of the present application also includes:

[0048] P51: The joint prediction model includes a wind power prediction unit and a photovoltaic prediction unit, and is embedded with a feature extractor; P52: The multi-dimensional prediction features are taken as input, and the feature extractor extracts wind power related features and photovoltaic related features respectively, and inputs them into the corresponding wind power prediction unit and photovoltaic prediction unit for joint prediction of power generation, thereby generating the predicted wind and photovoltaic power generation.

[0049] In a possible embodiment of the present application, the previously extracted and processed multidimensional prediction features are used as input data and input into the joint prediction model for prediction. The design of the joint prediction model includes multiple modules, which are intended to process the relevant features of wind power generation and photovoltaic power generation respectively, and combine the prediction results of the two to generate an accurate comprehensive prediction.

[0050] Exemplarily, the joint prediction model consists of two core prediction units, namely a wind power prediction unit and a photovoltaic prediction unit. These two units are responsible for independent prediction based on the characteristics of wind power generation and photovoltaic power generation. The wind power prediction unit mainly processes characteristic data related to wind power, while the photovoltaic prediction unit processes characteristic data related to photovoltaic power generation. Each unit performs independent prediction of power generation through a deep learning model (such as LSTM, CNN, etc.).

[0051] In addition, the model also has a built-in feature extractor, which is used to perform preliminary processing and analysis of the input multi-dimensional prediction features. The feature extractor can automatically identify the key features related to wind power generation and photovoltaic power generation power output, and perform necessary conversion and mapping on these features for subsequent prediction unit use.

[0052] After the multi-dimensional prediction features are input into the joint prediction model, they are first processed by the feature extractor. The feature extractor extracts features related to wind power and photovoltaic power according to different power generation modes. This includes extracting features related to meteorological factors such as wind speed, wind direction, temperature, and air pressure from the input data as wind power related features. These features are crucial for predicting wind power generation and are the most direct factors affecting wind power generation. And features related to light intensity, radiation intensity, photovoltaic panel temperature, etc. are extracted from the input data as photovoltaic related features. These features are crucial for the accurate prediction of photovoltaic power generation, especially since light intensity directly affects the power generation efficiency of photovoltaic cells.

[0053] After extracting relevant features, the feature extractor inputs these features into the wind power prediction unit and the photovoltaic prediction unit respectively. Each unit independently predicts based on the received features and outputs the predicted values ​​of wind power generation and photovoltaic power generation respectively. Finally, by merging and optimizing the prediction results of the two units, a comprehensive predicted wind and photovoltaic power generation is finally generated to provide decision support for subsequent grid dispatch and energy management.

[0054] P60: Monitor in real time the deviation between the actual wind and solar power generation power and the predicted wind and solar power generation power, and dynamically optimize the joint prediction model.

[0055] Furthermore, step P60 of the embodiment of the present application further includes:

[0056] P61: monitor the actual output power of the wind and solar power plants in real time, compare it with the predicted wind and solar power generation power, and calculate the deviation value; P62: based on the deviation value, update the weights and parameters of the joint prediction model through an adaptive adjustment mechanism to optimize the model performance.

[0057] Specifically, through real-time monitoring and feedback mechanisms, the prediction effect of the joint prediction model is continuously evaluated, and dynamic optimization is performed based on the deviation between the prediction result and the actual power generation. This ensures that the prediction model can continuously adjust and improve itself in actual operation, thereby maintaining high-precision prediction capabilities.

[0058] First, the actual output power of wind farms and photovoltaic power plants is continuously monitored. These actual output powers are obtained through on-site sensors and real-time monitoring equipment, and compared with the wind and solar power generation power predicted by the system as actual results. By calculating the deviation between the predicted value and the actual value (usually using metrics such as mean square error or absolute error), the accuracy of the prediction and the shortcomings of the model can be clearly seen. In this process, not only the deviation at a single moment is focused on, but the deviation within a certain time window is evaluated to eliminate the interference of a single abnormal data point on the optimization effect and ensure the stability and accuracy of the deviation calculation.

[0059] After the deviation value is calculated, the joint prediction model will be optimized through an adaptive adjustment mechanism. The core goal of the adaptive adjustment mechanism is to dynamically update the weights and parameters of the model according to the deviation value, thereby improving the accuracy of the model in future predictions. Specifically, the deviation value can be used to adjust the values ​​of each parameter in the model through back propagation (such as gradient descent). By continuously adjusting the weights of the model, the prediction error is reduced and the learning and generalization capabilities of the model are enhanced. Over time, the model will continue to self-learn and optimize as new data is added, thereby gradually improving the accuracy and reliability of predicting wind and solar power generation.

[0060] Furthermore, the embodiment of the present application further includes step P70, and step P70 further includes:

[0061] P71: Based on a preset update cycle, regularly collect updated multi-dimensional power generation data of wind and solar power plants; P72: Use the updated multi-dimensional power generation data to perform incremental learning updates on the joint prediction model.

[0062] Optionally, in order to ensure that the joint prediction model can adapt to the long-term changes in wind power generation and photovoltaic power generation, and make accurate predictions based on new environmental conditions, the model needs to be updated and incrementally learned regularly. By periodically acquiring new power generation data and incrementally updating the model, it is ensured that the model can maintain efficient and accurate prediction capabilities in the face of a changing external environment.

[0063] In this step, updated multi-dimensional power generation data of wind farms and photovoltaic power plants can be collected regularly based on a preset update cycle (such as every hour, every day, or every month). These data include the latest meteorological data, geographic data, and real-time power output data, reflecting the current environmental conditions and performance of the power plant. The collection of regularly updated data is a prerequisite for ensuring that the prediction model can adapt to new environmental changes, especially under the influence of factors such as climate change and seasonal fluctuations, the power generation efficiency may change, so the update of real-time data is very critical. By comparing with historical data, the newly collected updated data can provide new information for the model, thereby helping the model identify long-term trends and short-term fluctuations, and providing support for incremental learning updates.

[0064] Use the collected update data to perform incremental learning updates on the existing joint prediction model. The purpose of incremental learning is to continuously optimize the prediction performance of the model by introducing new data without losing the original knowledge. Unlike traditional full-scale training, incremental learning can improve the adaptability and accuracy of the model in a relatively short period of time by updating model parameters in a small range. It is especially suitable for scenarios such as wind and solar power generation that are greatly affected by external environment and seasonal changes.

[0065] Specifically, the new multi-dimensional power generation data can be integrated with the existing data, and the weights and parameters in the model can be gradually updated through model training algorithms (such as online learning, incremental training, etc.). The updated model can still maintain high-accuracy predictions of wind power generation and photovoltaic power generation in the face of new environmental changes, thereby further improving the accuracy and stability of the prediction results.

[0066] In summary, the embodiments of the present application have at least the following technical effects:

[0067] The present application collects multi-dimensional power generation data of wind farms and photovoltaic power farms in real time based on data sensing components, extracts multi-dimensional prediction features after pre-processing; collects historical wind and solar power generation data, trains and generates a joint prediction model, uses the multi-dimensional prediction features as input for joint prediction, and obtains predicted wind and solar power generation power; at the same time, monitors the deviation between the actual wind and solar power generation power and the predicted wind and solar power generation power in real time, and dynamically optimizes the joint prediction model.

[0068] The technical effect of significantly improving prediction accuracy and adaptability has been achieved by jointly modeling wind and solar power generation and introducing a dynamic optimization mechanism.

[0069] Embodiment 2, based on the same inventive concept as a wind-solar power generation combined prediction method in the above embodiment, Figure 2 As shown, the present application provides a wind and solar power generation joint prediction platform, and the platform and method embodiments in the present application embodiments are based on the same inventive concept. The platform includes:

[0070] The data acquisition module 11 is used to collect multi-dimensional power generation data of the wind farm and the photovoltaic farm in real time based on the data sensor component.

[0071] The data preprocessing module 12 is used to perform data preprocessing on the multi-dimensional power generation data to generate reference power generation data.

[0072] The feature extraction module 13 is used to extract key features using the reference power generation data and obtain multi-dimensional prediction features through feature correlation analysis.

[0073] The prediction model training module 14 is used to collect historical wind and solar power generation data and train and generate a joint prediction model.

[0074] The joint prediction module 15 is used to take the multi-dimensional prediction features as input, use the joint prediction model to perform joint prediction, and obtain the predicted wind and solar power generation power.

[0075] The dynamic optimization module 16 is used to monitor the deviation between the actual wind-solar power generation and the predicted wind-solar power generation in real time, and to dynamically optimize the joint prediction model.

[0076] Furthermore, the data preprocessing module 12 is further configured to perform the following steps:

[0077] The multi-dimensional power generation data includes real-time power output data, meteorological data and geographic information data of wind farms and photovoltaic power plants; the real-time power output data is cleaned, denoised, and missing values ​​are filled, the meteorological data is standardized, and the geographic information data is position calibrated and normalized to generate the benchmark power generation data.

[0078] Furthermore, the feature extraction module 13 is also used to perform the following steps:

[0079] A correlation analysis is performed on the benchmark power generation data to screen multiple power output characteristics, multiple meteorological characteristics, and multiple geographical characteristics that are closely related to wind power and photovoltaic power generation power output; principal component analysis is used to reduce the dimensions of the multiple power output characteristics, multiple meteorological characteristics, and multiple geographical characteristics, remove redundant characteristics, and generate multidimensional prediction features.

[0080] Furthermore, the prediction model training module 14 is also used to perform the following steps:

[0081] Based on historical wind and solar power generation records, historical wind and solar power generation data are collected to construct a model training data set, which includes a meteorological feature data set, a geographical feature data set and a corresponding power output data set; combined with a deep learning model, the model training data set is used as training data to perform model training to generate the joint prediction model.

[0082] Furthermore, the joint prediction module 15 is further configured to perform the following steps:

[0083] The joint prediction model includes a wind power prediction unit and a photovoltaic prediction unit, and is embedded with a feature extractor; the multidimensional prediction features are taken as input, and the feature extractor extracts wind power-related features and photovoltaic-related features respectively, and inputs them into the corresponding wind power prediction unit and photovoltaic prediction unit for joint prediction of power generation to generate the predicted wind and photovoltaic power generation.

[0084] Furthermore, the dynamic optimization module 16 is also used to perform the following steps:

[0085] The actual output power of the wind and solar power plants is monitored in real time, compared with the predicted wind and solar power generation power, and a deviation value is calculated; based on the deviation value, the weights and parameters of the joint prediction model are updated through an adaptive adjustment mechanism to optimize the model performance.

[0086] Furthermore, the platform also includes:

[0087] An update data acquisition module, which is used to regularly collect updated multi-dimensional power generation data of wind and solar power plants based on a preset update cycle; an incremental learning module, which is used to use the updated multi-dimensional power generation data to perform incremental learning updates on the joint prediction model.

[0088] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0089] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0090] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A wind and solar power generation joint prediction method, characterized in that: The method comprises: Based on data sensing components, multi-dimensional power generation data of wind farms and photovoltaic power plants are collected in real time; Performing data preprocessing on the multi-dimensional power generation data to generate reference power generation data; Extracting key features using the benchmark power generation data, and obtaining multi-dimensional prediction features through feature correlation analysis; Collect historical wind and solar power generation data and train and generate a joint prediction model; Taking the multi-dimensional prediction features as input, using the joint prediction model to perform joint prediction to obtain predicted wind and solar power generation power; The deviation between the actual wind-solar power generation and the predicted wind-solar power generation is monitored in real time, and the joint prediction model is dynamically optimized.

2. A wind and solar power generation joint prediction method as claimed in claim 1, characterized in that: Performing data preprocessing on the multi-dimensional power generation data to generate benchmark power generation data includes: The multi-dimensional power generation data includes real-time power output data of wind farms and photovoltaic farms, meteorological data and geographic information data; The real-time power output data is cleaned, denoised, and missing values ​​are supplemented, the meteorological data is standardized, and the geographic information data is position calibrated and normalized to generate the benchmark power generation data.

3. A wind and solar power generation joint prediction method as claimed in claim 2, characterized in that: The benchmark power generation data is used to extract key features, and through feature correlation analysis, multi-dimensional prediction features are obtained, including: performing correlation analysis on the benchmark power generation data to screen a plurality of power output characteristics, a plurality of meteorological characteristics, and a plurality of geographical characteristics that are closely related to wind power and photovoltaic power generation power output; The principal component analysis is used to reduce the dimension of the multiple power output features, multiple meteorological features and multiple geographical features, remove redundant features, and generate multidimensional prediction features.

4. A wind and solar power generation joint prediction method as claimed in claim 1, characterized in that: Collect historical wind and solar power generation data and train and generate a joint prediction model, including: Based on historical wind and solar power generation records, historical wind and solar power generation data are collected to construct a model training data set, wherein the model training data set includes a meteorological feature data set, a geographical feature data set and a corresponding power output data set; In combination with the deep learning model, model training is performed using the model training data set as training data to generate the joint prediction model.

5. The method for joint prediction of wind and solar power generation according to claim 1, characterized in that: The multi-dimensional prediction feature is used as input, and the joint prediction model is used to perform joint prediction to obtain the predicted wind and solar power generation power, including: The joint prediction model includes a wind power prediction unit and a photovoltaic prediction unit, and is embedded with a feature extractor; The multidimensional prediction features are taken as input, and the feature extractor extracts wind power related features and photovoltaic related features respectively, and inputs them into corresponding wind power prediction units and photovoltaic prediction units for joint prediction of power generation to generate the predicted wind and photovoltaic power generation.

6. A wind and solar power generation joint prediction method as claimed in claim 1, characterized in that: Real-time monitoring of the deviation between the actual wind and solar power generation power and the predicted wind and solar power generation power, and dynamic optimization of the joint prediction model, including: Real-time monitoring of the actual output power of the wind and solar power plant, and comparison with the predicted wind and solar power generation power, to calculate the deviation value; According to the deviation value, the weights and parameters of the joint prediction model are updated through an adaptive adjustment mechanism to optimize the model performance.

7. A wind and solar power generation joint prediction method as claimed in claim 1, characterized in that: include: Based on the preset update cycle, the updated multi-dimensional power generation data of wind and solar power plants are collected regularly; The updated multi-dimensional power generation data is used to perform incremental learning updates on the joint prediction model.

8. A wind and solar power generation joint prediction platform, characterized in that: The platform includes: A data acquisition module, which is used to collect multi-dimensional power generation data of wind farms and photovoltaic power plants in real time based on data sensing components; A data preprocessing module, the data preprocessing module is used to perform data preprocessing on the multi-dimensional power generation data to generate reference power generation data; A feature extraction module, the feature extraction module is used to extract key features using the benchmark power generation data, and obtain multi-dimensional prediction features through feature correlation analysis; A prediction model training module, which is used to collect historical wind and solar power generation data and train and generate a joint prediction model; A joint prediction module, the joint prediction module is used to take the multi-dimensional prediction feature as input, use the joint prediction model to perform joint prediction, and obtain predicted wind and solar power generation power; A dynamic optimization module is used to monitor in real time the deviation between the actual wind-solar power generation power and the predicted wind-solar power generation power, and to dynamically optimize the joint prediction model.

Images