A photovoltaic power station power generation prediction method based on multi-model fusion

Abstract

The application relates to the technical field of power generation prediction data processing, and discloses a photovoltaic power station power generation prediction method based on multi-model fusion, which fuses a photovoltaic physical model, a power generation statistical model and a power generation prediction model, considers physical laws, combines statistical learning and machine learning to improve the stability and accuracy of prediction, uses environmental meteorological data, photovoltaic power station electrical parameters and historical power generation data to realize the prediction of the power generation capacity of the photovoltaic power station. Accurate photovoltaic power generation power prediction helps the power grid to better arrange power generation plans and dispatch resources. The prediction accuracy and stability are improved through multi-model fusion, so that the power grid dispatching is more accurate and efficient, and the unstable factors of the power grid caused by photovoltaic power generation fluctuation are reduced. Photovoltaic power generation is more stably connected to the power grid and meets the power demand, thereby promoting the wide application and development of clean energy.

Description

Technical Field

[0001] This invention relates to the field of power generation forecasting data processing technology, and in particular to a method for forecasting photovoltaic power generation based on multi-model fusion. Background Technology

[0002] As a crucial component of clean energy, photovoltaic (PV) power generation forecasting is of paramount importance for the stable operation of power systems, grid dispatching, and the operation and maintenance of PV power plants. Accurate power forecasting helps the grid better plan its power generation, reduces grid instability caused by fluctuations in PV power generation, and also contributes to improving the economic efficiency of PV power plants. A PV power forecasting system utilizes modern digital communication technology, computer technology, real-time meteorological information monitoring technology, and numerical weather prediction technology to predict the power generation of PV power plants. Its main purpose is to comprehensively analyze and predict changes in the power generation of PV power plants in advance, providing crucial information for power dispatching, network optimization, resource management, and fault prevention.

[0003] Current technologies employ single-model prediction methods, which often utilize only a subset of useful information while neglecting other potentially equally important data. In photovoltaic (PV) power generation prediction, this can lead to models failing to comprehensively consider various factors influencing PV power generation, such as solar radiation, temperature, humidity, wind speed, and equipment status. Due to the lack of comprehensive information processing capabilities, the accuracy and reliability of prediction results may be compromised, and single models often struggle to achieve high precision in predicting PV power generation. Particularly under extreme weather conditions or when equipment malfunctions, single-model predictions may exhibit significant deviations. Single models also have limitations in terms of scalability. As PV power generation technology continues to develop and its applications expand, the demands on prediction models are increasing. However, single models often fail to meet these new requirements, necessitating significant improvements or redesigns. Summary of the Invention

[0004] To address the aforementioned problems, this invention integrates a photovoltaic physical model, a power generation statistical model, and a power generation prediction model, utilizing environmental meteorological data, photovoltaic power plant electrical parameters, and historical power generation data to predict the power generation of a photovoltaic power plant. It not only considers physical laws but also incorporates statistical learning and machine learning to improve the stability and accuracy of the prediction. Specifically, this invention provides a photovoltaic power plant power generation prediction method based on multi-model fusion, comprising the following steps:

[0005] S1. Data collection and preprocessing: Collect environmental meteorological data, photovoltaic power station electrical parameters, historical power generation data, and preprocess the data.

[0006] S2. Construct and fuse multiple models, including:

[0007] Photovoltaic physical model: Based on the photoelectric conversion principle of photovoltaic cells, a photovoltaic physical model is constructed according to factors such as temperature, light intensity, and module efficiency;

[0008] Power generation statistical model: Based on historical power generation data and environmental meteorological data, a time series model is constructed to statistically analyze power generation.

[0009] Power generation prediction model: A power generation prediction model is established based on machine learning methods, combined with environmental meteorological data, electrical parameters of photovoltaic power plants, and power generation statistical models.

[0010] Multiple models are fused using model ensemble methods;

[0011] S3. Train and validate the constructed multi-model, use historical power generation data to train the model, evaluate the predictive performance of the power generation prediction model and adjust and optimize the parameters to improve the prediction accuracy.

[0012] S4. Real-time forecasting and monitoring: Integrates real-time environmental and meteorological data interfaces to ensure the real-time performance and accuracy of the input data for the power generation forecasting model, updates forecast results in real time, monitors the power generation status of the power plant, adjusts the power plant operation strategy in a timely manner, and performs remote monitoring and early warning management.

[0013] S5. Feedback on prediction results and iterative optimization of multiple models: Compare the real-time prediction results of the power generation prediction model with the actual power generation data, analyze the causes of prediction errors, and continuously optimize multiple models.

[0014] Furthermore, the environmental meteorological data collected in S1 includes, but is not limited to, light intensity, temperature, humidity, wind speed, cloud cover, etc.; the electrical parameters of the photovoltaic power station include, but are not limited to, the installation angle, tilt angle, component type, component efficiency, array area, inverter efficiency, etc. of the photovoltaic array; and the historical power generation data is the power generation data of the power station over a period of time.

[0015] Furthermore, the data preprocessing in S1 specifically involves: firstly cleaning the data, performing data noise reduction, handling missing values ​​and outliers, and then standardizing or normalizing the data to ensure that the data can be effectively compared and analyzed.

[0016] Furthermore, the model integration method in S2 includes, but is not limited to, the weighted average method and the stacking method.

[0017] Furthermore, in S3, the predictive performance of the power generation prediction model is evaluated using indicators such as cross-validation, AUC, and RMSE.

[0018] Furthermore, S3 also includes:

[0019] S31. Model Training: Use the preprocessed data to train the photovoltaic physical model, power generation statistical model, and power generation prediction model. During the training process, monitor the performance of each model and make adjustments as needed.

[0020] S32. Hyperparameter optimization: Adjust the hyperparameters of each model, including learning rate, number of iterations, hidden layer dimension, etc., through Bayesian optimization methods to find the optimal model configuration;

[0021] S33. Regularization and overfitting prevention: Add regularization terms during training to prevent model overfitting, monitor performance on the validation set through an early stopping strategy, and stop training when performance no longer improves.

[0022] Furthermore, the continuous optimization of multiple models in S5 specifically includes:

[0023] Adjusting model parameters: Adjust the hyperparameters of the power generation prediction model based on the validation results, and optimize the parameter settings of the photovoltaic physical model and the power generation statistical model;

[0024] Data quality improvement: Identify and improve deficiencies in data collection and processing processes, and reduce data noise and missing values;

[0025] Introducing new features: Based on the analysis of prediction results, new environmental meteorological data or electrical parameters are introduced as input features for the model.

[0026] Furthermore, the multi-model fusion also includes integration with the power grid system, integrating the prediction results of the power generation prediction model with the power grid dispatch system, optimizing the power grid dispatch strategy, adjusting the power output of the power plant according to the prediction results, optimizing the power generation plan, and providing early warning of possible faults based on the power generation prediction model.

[0027] Furthermore, the real-time prediction and monitoring in S4 also includes: regularly generating prediction reports, including power generation prediction results, prediction error analysis, model optimization suggestions, etc., and using data visualization tools to display and monitor power generation prediction results, environmental meteorological data, photovoltaic power station electrical parameters, etc. in the form of charts, dashboards, etc.

[0028] The beneficial effects of this invention are:

[0029] 1. Improve Prediction Accuracy: Multi-model fusion fully leverages the advantages of different models in information processing, achieving information complementarity. Photovoltaic physical models can be modeled based on the physical principles of photovoltaic cells, considering the impact of physical factors such as solar radiation and temperature on photovoltaic power generation efficiency; power generation statistical models can explore the periodicity and regularity of photovoltaic power generation through statistical analysis of historical data. Fusing these two models comprehensively considers physical factors and statistical laws, thereby improving the prediction accuracy of the power generation forecasting model. Error Correction: Multi-model fusion can improve the accuracy of prediction results through error correction between different models. Each model may have certain errors during the prediction process, but the errors of different models may have different characteristics and distribution patterns. Through multi-model fusion, the complementarity of these errors can be used for correction, thereby reducing the overall prediction error.

[0030] 2. Enhanced Prediction Stability: A single model may experience significant fluctuations during prediction due to specific factors, while multi-model fusion can enhance prediction stability by diversifying risk. When one model is affected by abnormal data or external disturbances, other models can provide relatively stable prediction results, thus avoiding drastic fluctuations in the overall prediction outcome. Multi-model fusion can automatically adjust the weights and parameters of each model according to different environmental conditions and data characteristics to adapt to different prediction scenarios and needs. This strong adaptability makes multi-model fusion methods more reliable and stable in practical applications.

[0031] 3. Enhanced Generalization Ability: Multi-model fusion can transcend different disciplines and technical fields, organically combining models and methods from various domains. This cross-domain integration fully leverages the expertise and technical advantages of each field, enhancing the generalization ability of the prediction model. For example, photovoltaic physics models can be fused with power generation prediction machine learning models, utilizing the physical meaning of the physics model and the powerful computational capabilities of the machine learning model to jointly improve prediction accuracy. Photovoltaic power generation is influenced by a variety of complex factors, including weather, temperature, humidity, wind speed, and equipment status. Multi-model fusion can comprehensively consider the interactions between these complex factors, thereby more accurately predicting the changing trends of photovoltaic power generation. This ability to cope with complex environments makes multi-model fusion methods more advantageous in practical applications.

[0032] 4. Optimizing resource allocation: Accurate photovoltaic (PV) power generation forecasting helps the power grid better plan power generation and dispatch resources. Improving forecast accuracy and stability through multi-model fusion allows for more precise and efficient grid dispatching, reducing grid instability caused by fluctuations in PV power generation. This enables more stable grid connection of PV power generation to meet electricity demand, thereby promoting the widespread application and development of clean energy. Attached Figure Description

[0033] Figure 1 This is a flowchart of a photovoltaic power plant power generation prediction method based on multi-model fusion. Detailed Implementation

[0034] The following is in conjunction with the appendix Figure 1 The preferred embodiments of the present invention will be described in detail so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more explicit definition of the scope of protection of the present invention.

[0035] This invention provides a photovoltaic power generation prediction method based on multi-model fusion. By combining the advantages of physical models, statistical models, and machine learning models, it achieves high-precision prediction of photovoltaic power generation. Furthermore, through real-time monitoring, iterative optimization, and grid system integration, the stability and practicality of the prediction are further improved. Specifically:

[0036] First, data collection and preprocessing are carried out, including collecting environmental meteorological data, electrical parameters of photovoltaic power plants, and historical power generation data, and preprocessing the data. The collected environmental meteorological data includes, but is not limited to, light intensity, temperature, humidity, wind speed, cloud cover, etc., covering key factors affecting photovoltaic power generation, which have a direct impact on the power generation efficiency of photovoltaic power plants.

[0037] The electrical parameters of a photovoltaic power station include, but are not limited to, the installation angle, tilt angle, module type, module efficiency, array area, and inverter efficiency of the photovoltaic array. These parameters reflect the physical characteristics and technical configuration of the power station.

[0038] Historical power generation data refers to the power generation data of a power plant over a past period. It serves as the basis for evaluating the performance of predictive models and allows us to understand the power generation performance of a power plant under different conditions.

[0039] The data preprocessing process involves: first, cleaning the data, reducing noise, and handling missing and outlier values; then, standardizing or normalizing the data to ensure the consistency and comparability of the model inputs and to ensure that the data can be effectively compared and analyzed.

[0040] Then, multiple models are constructed and fused, including:

[0041] Photovoltaic physical model: Based on the photoelectric conversion principle of photovoltaic cells, a photovoltaic physical model is constructed according to factors such as temperature, light intensity, and module efficiency. Based on the photoelectric conversion principle of photovoltaic cells, the power generation process of photovoltaic power plants is directly simulated, providing a physical basis for prediction.

[0042] Power generation statistical model: Based on historical power generation data and environmental meteorological data, a time series model is constructed to statistically analyze power generation. By using time series analysis techniques, statistical patterns of power generation are extracted from historical data to provide data support for forecasting.

[0043] Power generation prediction model: Based on machine learning methods, combined with environmental meteorological data, electrical parameters of photovoltaic power plants, and power generation statistical models, a power generation prediction model is established to predict power generation with high accuracy.

[0044] Model ensemble methods are used to fuse multiple models; these methods include, but are not limited to, weighted averaging and stacking. Each model has its unique advantages and limitations; combining them can improve the accuracy and robustness of predictions.

[0045] Multiple models can be integrated using a weighted average method. Specifically, a weight is assigned to the prediction results of each model, and then these weighted prediction results are summed to obtain the final prediction value. The weight allocation is usually based on the model's performance on historical data, with better-performing models receiving higher weights. This method is simple to implement, computationally efficient, and can adjust weights according to model performance, fully utilizing the advantages of photovoltaic physical models, power generation statistical models, and power generation prediction models.

[0046] A stacking approach can also be used for multi-model ensemble. First, multiple base models are trained using the collected raw datasets (environmental and meteorological data, photovoltaic power plant electrical parameters, and historical power generation data). Then, new features are generated: the base models are used to predict power generation on the training set, and the predicted results are added as new features to the training set. Next, a meta-model is trained using the training set containing the new features from the power generation predictions. The meta-model's input is the prediction results from the base models, and its output is the final predicted value. Finally, the trained meta-model is used to predict power generation on the test set, and the prediction results are input into the trained meta-model to obtain the final prediction result.

[0047] Then, the constructed multi-model is trained and validated. Historical power generation data is used to train the model, evaluate the predictive performance of the power generation prediction model, and adjust and optimize the parameters to improve the prediction accuracy.

[0048] The predictive performance of the power generation prediction model is evaluated using metrics such as cross-validation, AUC, and RMSE. These metrics help understand the model's prediction accuracy and stability, providing a basis for subsequent optimization. Cross-validation is used to train and test each base model, obtaining the prediction error of each model, and weights are assigned to each model based on the magnitude of the prediction error. Weight allocation typically employs optimization algorithms such as least squares to minimize the prediction error of the fusion model.

[0049] Model training: The photovoltaic physical model, power generation statistical model, and power generation prediction model are trained using preprocessed data. During the training process, the performance of each model is monitored and adjusted as needed.

[0050] Hyperparameter optimization: Adjust the hyperparameters of each model, including learning rate, number of iterations, hidden layer dimension, etc., using Bayesian optimization methods to find the optimal model configuration;

[0051] Regularization and overfitting prevention: Add regularization terms (such as L1 or L2 regularization, Dropout, etc.) during training to prevent model overfitting. Use early stopping strategies to monitor performance on the validation set and stop training when performance no longer improves.

[0052] Then, real-time forecasting and monitoring are implemented, integrating real-time environmental and meteorological data interfaces to ensure the real-time nature and accuracy of input data for the power generation forecasting model. Data is collected and transmitted instantly through IoT technology. Forecast results are updated in real-time, monitoring of power plant generation allows for timely adjustments to operating strategies, and remote monitoring and early warning management are conducted to promptly identify and address potential problems. Regular forecast reports are generated, including power generation forecast results, forecast error analysis, and model optimization suggestions. Data visualization tools (such as Tableau, Power BI, and Matplotlib) are used to display and monitor power generation forecast results, environmental and meteorological data, and photovoltaic power plant electrical parameters in charts, dashboards, and other formats. Regularly generated forecast reports, including power generation forecast results, forecast error analysis, and model optimization suggestions, facilitate management decision-making. This helps to quickly understand the power plant's operating status, monitor forecast accuracy, and identify potential problems or trends.

[0053] Finally, the prediction results are fed back and multiple models are optimized iteratively. The real-time prediction results of the power generation prediction model are compared with the actual power generation data to analyze the causes of prediction errors. These errors may stem from data quality, limitations of model assumptions, uncertainties in environmental changes, etc. Continuous optimization of multiple models is then performed.

[0054] Adjusting model parameters: Adjust the hyperparameters of the power generation prediction model based on the validation results, and optimize the parameter settings of the photovoltaic physical model and the power generation statistical model;

[0055] Data quality improvement: Identify and improve deficiencies in data collection and processing processes, and reduce data noise and missing values;

[0056] Introducing new features: Based on the analysis of prediction results, new environmental meteorological data or electrical parameters are introduced as input features of the model to improve the model's generalization ability and prediction ability.

[0057] Multi-model fusion also includes integration with the power grid system. This involves integrating the prediction results of power generation forecasting models with the power grid dispatching system to optimize dispatching strategies and improve the grid's capacity to accommodate renewable energy. Deep integration of photovoltaic (PV) power plant power generation forecasts with the power grid dispatching system enables coordinated dispatching of PV power generation and grid load, improving grid stability and economic efficiency. Power plant power output is adjusted based on forecast results to optimize power generation plans, reduce fluctuations, and improve grid stability. Furthermore, based on power generation forecasting models, early warnings of potential faults are provided, optimizing equipment maintenance plans and reducing downtime. Based on forecast results and data analysis, decision support is provided for the operation, maintenance, expansion, and renovation of power plants.

[0058] With technological advancements and changes in the operating environment of photovoltaic power plants, the technology and algorithms of the prediction system should be regularly evaluated and updated. As new machine learning algorithms or data processing technologies develop, the potential improvement of prediction performance by these new technologies can be assessed, and technological upgrades can be carried out in a timely manner.

[0059] Any embodiment of the present invention can be used as an independent technical solution or in combination with other embodiments. All patents and publications mentioned in this specification represent publicly available technologies that can be used with the present invention. All patents and publications cited herein are also listed in the references as if each publication were individually referenced. The present invention can be implemented in the absence of any one or more elements, or one or more limitations, which are not specifically stated herein. The terminology and expressions used herein are descriptive methods and are not intended to be limiting, nor is there any intention to exclude any equivalent features from the terms and interpretations described herein; however, it is understood that any suitable changes or modifications can be made within the scope of the invention and the claims. It is understood that the embodiments described herein are embodiments and features in some examples, and any modifications and variations can be made by those skilled in the art based on the spirit of the description, and such modifications and variations are also considered to fall within the scope of the invention and the limitations of the independent and appended claims.

Claims

1. A photovoltaic power station power generation prediction method based on multi-model fusion, characterized in that, Specifically comprising the following steps: S1, data collection and preprocessing, collecting environmental meteorological data, photovoltaic power station electrical parameters, historical power generation data and preprocessing the data; S2, constructing multiple models and fusion, including: Photovoltaic physical model: according to the photoelectric conversion principle of photovoltaic cells, according to temperature, light intensity, component efficiency factors to construct photovoltaic physical model; Power generation statistical model: according to historical power generation data, environmental meteorological data, construct time series model for power generation statistics; Power generation prediction model: according to machine learning method, combined with environmental meteorological data, photovoltaic power station electrical parameters, combined with power generation statistical model to establish power generation prediction model; Through model integration method to fuse multiple models; S3, training and verification of the constructed multiple models, using historical power generation data to train the model, evaluate the prediction performance of the power generation prediction model and adjust the optimization parameters to improve the prediction accuracy; S4, real-time prediction and monitoring, integrating real-time environmental meteorological data interface to ensure the real-time and accuracy of the input data of the power generation prediction model, real-time updating the prediction results, monitoring the power station power generation, timely adjusting the power station operation strategy, and performing remote monitoring and early warning management; S5, prediction result feedback and multiple model optimization iteration, comparing the real-time prediction results of the power generation prediction model with the actual power generation data, analyzing the reasons for the prediction error, and continuously optimizing the multiple models; S31, model training: using the preprocessed data to train the photovoltaic physical model, power generation statistical model and power generation prediction model, monitoring the performance of each model during training, and adjusting as needed; S32, hyperparameter optimization: adjusting the hyperparameters of each model, including learning rate, iteration number, hidden layer dimension, through Bayesian optimization method to find the optimal model configuration; S33, regularization and prevention of overfitting: adding regularization term to prevent model overfitting during training, monitoring the performance on the validation set through early stopping strategy, and stopping training when the performance no longer improves; The S5 continuously optimizes the multiple models, specifically including: Adjusting model parameters: adjusting the hyperparameters of the power generation prediction model according to the verification results, optimizing the parameter settings of the photovoltaic physical model and power generation statistical model; Data quality improvement: identifying and improving the shortcomings in the data collection and processing process, reducing data noise and missing values; Introducing new features: according to the prediction effect analysis, introducing new environmental meteorological data or electrical parameters as model input features. The environmental meteorological data collected in S1 includes but is not limited to light intensity, temperature, humidity, wind speed, cloud cover; photovoltaic power station electrical parameters include but are not limited to photovoltaic array installation angle, inclination angle, component type, component efficiency, array area, inverter efficiency; historical power generation data is the power generation data of the power station in the past period of time. 2.The photovoltaic power station power generation prediction method based on multi-model fusion according to claim 1, characterized in that, The data preprocessing in S1 specifically includes: first cleaning the data, performing data denoising, handling missing values and outliers, then performing data standardization or normalization to ensure that the data can be effectively compared and analyzed. 3.The photovoltaic power station power generation prediction method based on multi-model fusion according to claim 1, characterized in that, The model integration method in S2 includes but is not limited to weighted average method, stacking method. 4.The photovoltaic power station power generation prediction method based on multi-model fusion of claim 1, wherein, ​ 5. The photovoltaic power station power generation prediction method based on multi-model fusion according to claim 1, characterized in that, The prediction performance of the power generation prediction model in the S3 is evaluated by cross-validation, AUC, and RMSE indicators. 6.The photovoltaic power station power generation prediction method based on multi-model fusion according to claim 1, characterized in that, The multi-model fusion further includes integration with the power grid system, integration of the prediction results of the power generation prediction model with the power grid dispatching system, optimization of the power grid dispatching strategy, adjustment of the power output of the power station according to the prediction results, optimization of the power generation plan, and early warning of possible faults according to the power generation prediction model.

7. The photovoltaic power station power generation prediction method based on multi-model fusion according to claim 1, characterized in that, The real-time prediction and monitoring in the S4 further includes: periodically generating a prediction report including power generation prediction results, prediction error analysis, and model optimization suggestion content, and using a data visualization tool to display and monitor the power generation prediction results, environmental meteorological data, and photovoltaic power station electrical parameters in the form of charts and dashboards.

Images