New energy power prediction method and system based on multi-source meteorological data fusion

Abstract

The invention belongs to the technical field of power prediction, and discloses a new energy power prediction method and system based on multi-source meteorological data fusion, and the method comprises the steps: obtaining meteorological data from different meteorological data sources in real time or at regular time through a data obtaining module; carrying out preliminary screening and preprocessing on the obtained data by utilizing a data processing module; the data fusion module is used for integrating meteorological data from different sources by adopting an advanced data fusion algorithm to eliminate data redundancy; a mathematical model or a machine learning model suitable for new energy power prediction is constructed by using the prediction model module based on meteorological data of multi-source fusion; and the result output and evaluation module is used for outputting the prediction result to a user in a visual form, and carrying out real-time or regular evaluation on the prediction result. The method has the characteristics of diversified data sources, advanced data fusion technology, flexible prediction model construction, high real-time performance and accuracy, wide application scenarios and the like.

Description

Technical Field

[0001] The present invention belongs to but is not limited to the field of power prediction technology, and in particular relates to a new energy power prediction method and system based on multi-source meteorological data fusion. Background Art

[0002] The new energy power prediction method based on multi-source meteorological data fusion is mainly due to the complexity of new energy power generation and the significant impact of meteorological factors on its output. Multi-source meteorological data fusion technology can make up for the shortcomings of a single data source and improve the accuracy and comprehensiveness of meteorological data by integrating meteorological data from different sources and different types (such as ground observation data, satellite remote sensing data, and multiple numerical weather forecasting systems). Furthermore, the new energy power prediction method based on multi-source meteorological data fusion can significantly improve the accuracy and reliability of the prediction.

[0003] With the rapid development of technologies such as big data and artificial intelligence, the new energy power prediction method based on multi-source meteorological data fusion is gradually becoming a research hotspot and application trend. By building advanced data processing and analysis models, the changing laws of new energy output can be captured more accurately, providing strong support for the optimized dispatch and stable operation of the power system.

[0004] However, traditional renewable energy power forecasting methods often rely on a single meteorological data source, such as ground observation stations or a single numerical weather forecast system. However, these single data sources often have large errors and limited coverage, making it difficult to fully and accurately reflect the actual meteorological conditions of renewable energy sites. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a new energy power prediction method and system based on multi-source meteorological data fusion.

[0006] The present invention is implemented as follows: a new energy power prediction method based on multi-source meteorological data fusion, the method comprising:

[0007] 1: Acquisition of meteorological data

[0008] The data acquisition module is used to obtain meteorological data from multiple meteorological data sources in real time or at regular intervals, including:

[0009] Weather station: obtain data such as temperature, humidity, wind speed, wind direction, and solar radiation intensity;

[0010] Satellite remote sensing data: obtain large-scale meteorological change data, including cloud coverage and radiation intensity distribution;

[0011] Online API interface: Get real-time weather forecast data from weather service providers;

[0012] Local environmental sensors: monitor data such as wind speed, wind direction, humidity, and light intensity.

[0013] S2: Data preprocessing

[0014] The data processing module is used to perform preliminary screening and preprocessing on the acquired data, including:

[0015] Format conversion: unify data formats from different sources (such as JSON, CSV, XML) into a standardized format that can be used for modeling;

[0016] Outlier elimination: Detect and eliminate outliers in meteorological data through algorithms (such as the Z-score method) to ensure data accuracy;

[0017] Time series synchronization: Time alignment of data from different sources, using methods such as linear interpolation to handle incomplete time series data;

[0018] Noise filtering: Use filters (such as Kalman filters) to remove high-frequency noise from the data.

[0019] S3: Data Fusion

[0020] The data fusion module uses advanced data fusion algorithms to integrate meteorological data from different sources, including:

[0021] Weighted average method: Different weights are assigned to different data sources based on their precision and historical accuracy, and data at the same time point are merged;

[0022] Kalman filter algorithm: Based on the dynamic characteristics of different meteorological sensors, the Kalman filter algorithm is used to achieve optimal estimation of multiple data sources;

[0023] Adaptive fusion: Dynamically adjust data fusion weights according to real-time environmental changes to ensure the reliability of fused data in different environments;

[0024] Multi-source data redundancy elimination: Principal component analysis (PCA) is used to eliminate redundant data and extract the most representative meteorological features for prediction.

[0025] S4: Prediction model construction

[0026] The prediction model module is used to build a mathematical model or machine learning model suitable for new energy power prediction based on multi-source fusion meteorological data, including:

[0027] Deep learning model: Use long short-term memory network (LSTM) to predict meteorological time series data and capture the long-term dependency between renewable energy power and meteorological elements;

[0028] Support Vector Regression (SVR) Model: The nonlinear relationship between meteorological data and power output is modeled using the support vector machine regression model (SVR);

[0029] Random forest model: The random forest algorithm is used to build a prediction model based on multi-dimensional meteorological data, and multiple decision trees are used to predict power;

[0030] Combined model: Combine multiple models and use ensemble learning methods such as XGBoost to combine the advantages of deep learning and machine learning models to output high-precision power prediction results.

[0031] S5: Result output and evaluation

[0032] The result output and evaluation module is used to present the prediction results to users in intuitive forms such as charts, reports, dynamic curves, etc., including:

[0033] Power forecast curve: a curve chart showing the forecast of new energy power generation in the next 24 hours;

[0034] Model evaluation: Use indicators such as mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE) to evaluate the performance of the prediction model in real time;

[0035] Dynamic adjustment: Based on real-time evaluation results, the model parameters are automatically adjusted through the model self-learning mechanism to optimize the prediction effect;

[0036] Forecast report: Generate a detailed forecast report, including the trend forecast of renewable energy power generation in the next week, analysis of the impact of meteorological factors on power, etc.

[0037] Hardware and software support:

[0038] Hardware equipment:

[0039] Meteorological sensor nodes: sensors used to collect meteorological parameters, such as temperature, humidity, wind speed, etc.

[0040] Satellite receiving equipment: a receiver used to obtain satellite meteorological data;

[0041] Edge computing device: A computing unit used for real-time data preprocessing, data cleaning and format conversion through embedded devices.

[0042] Software Support:

[0043] Data processing and fusion module: data preprocessing and fusion processing through Python, R or MATLAB;

[0044] Prediction algorithm implementation: Model construction and training based on deep learning frameworks such as TensorFlow and PyTorch;

[0045] Result display and evaluation platform: Use the Web platform and front-end technologies (such as D3.js and Tableau) to achieve user interface visualization and report generation.

[0046] Through the above-mentioned specific data processing method, fusion algorithm and prediction model, the application of the present invention in actual renewable energy power prediction improves data consistency, power prediction accuracy and real-time performance, and significantly improves renewable energy scheduling and power grid operation efficiency.

[0047] Furthermore, the data preprocessing module uses a Kalman filter to filter the noise of the meteorological data, combines the Z-score method to eliminate abnormal data, further uses linear interpolation to complete the incomplete time series data, and uses the principal component analysis (PCA) method to extract the most representative meteorological feature data for subsequent model training and prediction.

[0048] Furthermore, the data fusion module fuses meteorological data from different data sources through a weighted average algorithm, automatically adjusts the weighting coefficient according to the historical accuracy and reliability of the data source, and uses an adaptive fusion algorithm to dynamically weight the meteorological data to achieve redundancy elimination and optimal fusion of multi-source data.

[0049] Furthermore, the prediction model module uses a long short-term memory network (LSTM) to model time series meteorological data, and achieves high-precision power prediction by capturing the long-term dependency between meteorological elements and renewable energy power output. The training process of the LSTM model uses the Adam optimizer, the learning rate is set to 0.001, and the loss function is the mean square error (MSE).

[0050] Furthermore, the meteorological data sources specifically include: ground observation stations, satellite remote sensing platforms, numerical weather forecasting systems, etc.

[0051] Further, the S2 specifically includes:

[0052] S21: Format conversion. Since different data sources have different data formats, such as file types CSV, JSON, XML, data encoding UTF-8, GBK, time format, etc., it is necessary to convert the received data to a unified data format for subsequent processing and analysis.

[0053] S22: Data cleaning and outlier removal. Identify and remove outliers in the data through statistical methods such as the 3σ criterion, physical rules such as wind speed cannot be negative, or machine learning algorithms such as isolation forests.

[0054] S23: Missing value processing: For missing values ​​in the data, filling methods such as mean filling, median filling, mode filling, interpolation such as linear interpolation, polynomial interpolation or deletion are used to process them;

[0055] S24: Data standardization or normalization is to scale the data so that it falls into a small specific interval, such as between 0 and 1; normalization is to scale the data to a fixed range according to certain rules, such as maximum and minimum values;

[0056] S25: Data integration: The cleaned and pre-processed meteorological data are integrated according to time sequence, spatial location, etc. to form a complete meteorological data set for subsequent analysis.

[0057] Further, the S3 specifically includes:

[0058] (1) Using the weighted average method, different weights are assigned to different data sources according to their credibility or accuracy, and then the weighted average is calculated as the fusion result;

[0059] Suppose there are n data sources, and the data provided by the i-th data source is x i , whose corresponding weight is w i , then the fused data x can be calculated by the following formula:

[0060]

[0061] Among them, the weight w i This can be determined based on the historical performance of the data source, accuracy assessment, or other relevant factors;

[0062] (2) Use the Kalman filter algorithm to estimate the state of the system in a recursive manner, while considering the prediction error and observation error of the model to obtain the optimal fusion result;

[0063] In Kalman filtering, the state equation and observation equation of the system can be expressed as:

[0064] x k =A·X K-1 +B·u k+ w k

[0065] Z k =H·x k +v k

[0066] Among them, x k represents the system state at time k, A is the state transfer matrix, B is the control input matrix, and u k is the control input, w k is the process noise; zk represents the observation value at time k, H is the observation matrix, v k is the observation noise.

[0067] Further, the S4 specifically includes:

[0068] (1) Select features related to renewable energy power prediction from multi-source fused meteorological data, such as wind speed, wind direction, temperature, humidity, and air pressure;

[0069] (2) Based on the physical characteristics of renewable energy power generation and the impact mechanism of meteorological data, a mathematical model suitable for renewable energy power forecasting is constructed;

[0070] (3) Model training and optimization data division: divide the fused meteorological data and the corresponding renewable energy power generation data into training sets and test sets; use the training set data to train the selected mathematical model or machine learning model, and adjust the model parameters to minimize the prediction error; optimize the model parameters through cross-validation, grid search and other methods to improve the generalization ability and prediction accuracy of the model;

[0071] (4) Use the test set data to make predictions for the trained model and calculate performance indicators such as prediction error, accuracy, and recall rate; verify the prediction performance and reliability of the model by comparing it with actual renewable energy power generation data.

[0072] Further, the step (2) specifically includes:

[0073] Taking wind power as an example, the wind speed-power curve model can be expressed by the following formula:

[0074]

[0075] Where: -P is the output power of the wind turbine; -v is the wind speed; -v ci is the cut-in wind speed, i.e. the minimum wind speed at which the wind turbine starts generating electricity; -v co is the cut-out wind speed, i.e. the maximum wind speed at which the wind turbine stops generating electricity; -v r is the rated wind speed, that is, the wind speed when the wind turbine reaches the rated power; -P r is the rated power of the wind turbine; -k is the shape parameter of the power curve, usually a positive real number;

[0076] This model describes the output power of the wind turbine at different wind speeds. When the wind speed is lower than the cut-in wind speed or higher than the cut-out wind speed, the wind turbine does not generate electricity and the output power is 0. When the wind speed is between the cut-in wind speed and the rated wind speed, the output power increases with the increase of wind speed and follows a power function relationship. When the wind speed reaches or exceeds the rated wind speed, the wind turbine generates electricity at the rated power.

[0077] Another object of the present invention is to provide a new energy power prediction system based on multi-source meteorological data fusion based on the new energy power prediction method based on multi-source meteorological data fusion, the system specifically comprising:

[0078] A data acquisition module, used to acquire meteorological data from different meteorological data sources in real time or at a scheduled time;

[0079] The data processing module is connected to the data acquisition module and is used to perform preliminary screening and preprocessing on the acquired data, such as format conversion, outlier removal, etc.;

[0080] The data fusion module is connected to the data processing module and uses advanced data fusion algorithms to integrate and process meteorological data from different sources, eliminate data redundancy, and improve data consistency and reliability;

[0081] The prediction model module is connected with the data fusion module to build a mathematical model or machine learning model suitable for new energy power prediction based on multi-source fused meteorological data; these models can capture the complex relationship between new energy power generation and meteorological elements and output high-precision power prediction results;

[0082] The result output and evaluation module is connected to the prediction model module to output the prediction results to the user in an intuitive form, such as charts, reports, etc., and conduct real-time or regular evaluation of the prediction results.

[0083] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the new energy power prediction method based on multi-source meteorological data fusion.

[0084] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the new energy power prediction method based on multi-source meteorological data fusion.

[0085] Another object of the present invention is to provide an information data processing terminal, which is used to implement the new energy power prediction system based on multi-source meteorological data fusion.

[0086] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0087] First, the new energy power prediction method and system based on multi-source meteorological data fusion provided by the present invention is a system that integrates multiple advanced technologies and is mainly used to improve the prediction accuracy of new energy power generation. The following are the key points of the method and system:

[0088] Diversified data sources: The system obtains meteorological data from multiple authoritative meteorological data sources (such as ground observation stations, satellite remote sensing platforms, numerical weather forecasting systems, etc.) to ensure the comprehensiveness and accuracy of the data.

[0089] Data fusion technology: Use advanced data fusion algorithms to integrate meteorological data from different sources, eliminate redundant information, and improve data consistency and reliability.

[0090] Prediction model construction: Based on multi-source fused meteorological data, mathematical models or machine learning models suitable for renewable energy power forecasting are constructed. These models can capture the complex relationship between renewable energy power generation and meteorological elements, and achieve high-precision power forecasting.

[0091] Real-time and accuracy: The system can obtain meteorological data in real time or near real time and quickly output forecast results. By continuously optimizing algorithms and models, the forecast results are ensured to be highly accurate and stable.

[0092] Wide range of application scenarios: The system is widely used in power system dispatching, new energy station operation management and other fields, providing important support for optimizing power grid operation, reducing wind and solar power abandonment, and improving the utilization rate of new energy power generation.

[0093] In summary, the new energy power prediction method and system based on multi-source meteorological data fusion is a comprehensive system that integrates a variety of advanced technologies. It has the characteristics of diversified data sources, advanced data fusion technology, flexible prediction model construction, high real-time and accuracy, and a wide range of application scenarios.

[0094] Second, the expected benefits and commercial value of the technical solution of the present invention after transformation are:

[0095] This invention solves the accuracy, real-time and stability problems in the prediction of renewable energy power generation by integrating multi-source meteorological data fusion technology and artificial intelligence algorithms. The transformation of the technical solution is expected to bring the following significant benefits and commercial value:

[0096] 1. Improve power generation management efficiency and reduce costs:

[0097] By accurately predicting future power generation, power plants can more effectively plan power generation and dispatch resources, reduce power generation losses and resource waste. Power plants can optimize equipment operating conditions based on the prediction results, reduce unnecessary equipment downtime and maintenance frequency, and reduce operation and maintenance costs (O&M costs), with an estimated overall operating cost reduction of 10%-20%.

[0098] 2. Optimize grid dispatching and improve power supply reliability:

[0099] The present invention can predict the power output of renewable energy in advance, help grid managers better balance load distribution, reduce the impact of renewable energy fluctuations on the grid, and improve the stability and reliability of the grid. Through this technology, the grid dispatching efficiency will be increased by about 15%-20%, reducing the risk of grid interruption and energy transmission loss.

[0100] 3. Expand market applications and increase commercial revenue:

[0101] This technical solution can be widely used in various new energy power generation projects such as photovoltaic power generation and wind power generation, and has good scalability and market application potential. The improvement of prediction accuracy makes more new energy power stations willing to adopt this technology, and the market penetration rate is expected to increase by 30%, bringing considerable commercial income to power station operators and technical service providers.

[0102] 4. Promote carbon emission reduction in line with the goal of green economic development:

[0103] The technology of the present invention helps to better dispatch clean energy, reduce dependence on fossil energy, reduce carbon emissions, and promote sustainable development goals. The application of new energy power prediction technology after transformation is expected to reduce carbon dioxide emissions by more than 100,000 tons per year after large-scale promotion, which will help enterprises obtain additional benefits in the carbon trading market.

[0104] Third, the technical solution of the present invention fills the technical gap in the industry at home and abroad:

[0105] 1. Innovation of multi-source meteorological data fusion technology:

[0106] At present, the prediction technology of renewable energy power generation at home and abroad mainly relies on a single meteorological data source, while the present invention proposes for the first time a prediction method based on the fusion of multi-source meteorological data (including meteorological stations, satellite remote sensing, equipment status, etc.). This multi-source fusion technology significantly improves the accuracy and timeliness of data, especially under complex meteorological conditions, and can better eliminate data errors and ensure prediction accuracy.

[0107] 2. Power prediction model integrating deep learning and traditional machine learning:

[0108] Internationally, the field of renewable energy power forecasting mainly adopts experience-based statistical models or simple machine learning algorithms. This invention combines deep learning models (such as LSTM) with traditional machine learning models (such as SVM and random forest) to provide a new hybrid model architecture, which effectively improves the accuracy and stability of power forecasting. This model can capture complex nonlinear relationships under different meteorological conditions, breaking through the limitations of existing technologies.

[0109] 3. Adaptive data fusion and model dynamic optimization:

[0110] The present invention uses adaptive data fusion algorithms and dynamic model optimization technology to adjust model parameters in real time according to changes in weather and equipment status. This technology breaks through the limitations of existing fixed models at home and abroad, can self-optimize to cope with real-time changing complex environments, and ensure that the prediction results of new energy power generation are more accurate and reliable.

[0111] 4. Universal application in various new energy scenarios:

[0112] Most of the new energy power prediction technologies on the market are aimed at a single type of power generation scenario (such as photovoltaic or wind power), while the present invention has strong versatility and can be applied to a variety of new energy power generation scenarios. By expanding data processing and model architecture, the present invention can be compatible with multiple types of power generation facilities such as photovoltaic and wind power, filling the technical gap in multiple new energy power generation scenarios.

[0113] In summary, the technical solution of the present invention fills the domestic and foreign technical gaps in data fusion, algorithm innovation, real-time optimization, etc., and is expected to have a far-reaching impact and broad application prospects in the field of renewable energy power generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0114] Figure 1 This is a flow chart of a new energy power prediction method based on multi-source meteorological data fusion provided by an embodiment of the present invention;

[0115] Figure 2 It is a flow chart of a method for performing preliminary screening and preprocessing of acquired data using a data processing module, such as format conversion and outlier removal, provided by an embodiment of the present invention;

[0116] Figure 3 It is a structural diagram of a new energy power prediction system based on multi-source meteorological data fusion provided by an embodiment of the present invention;

[0117] In the figure: 1. Data acquisition module; 2. Data processing module; 3. Data fusion module; 4. Prediction model module; 5. Result output and evaluation module. DETAILED DESCRIPTION

[0118] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0119] like Figure 1 As shown, an embodiment of the present invention provides a new energy power prediction method based on multi-source meteorological data fusion, the method comprising:

[0120] S1: using the data acquisition module to obtain meteorological data from different meteorological data sources in real time or at a fixed time;

[0121] S2: Use the data processing module to perform preliminary screening and preprocessing on the acquired data, such as format conversion, outlier removal, etc.;

[0122] S3: The data fusion module uses advanced data fusion algorithms to integrate meteorological data from different sources, eliminate data redundancy, and improve data consistency and reliability;

[0123] S4: Use the prediction model module to build mathematical models or machine learning models suitable for renewable energy power prediction based on multi-source fused meteorological data; these models can capture the complex relationship between renewable energy power generation and meteorological elements and output high-precision power prediction results;

[0124] S5: Use the result output and evaluation module to output the prediction results to the user in an intuitive form, such as charts, reports, etc., and conduct real-time or regular evaluation of the prediction results.

[0125] The meteorological data sources specifically include: ground observation stations, satellite remote sensing platforms, numerical weather forecasting systems, etc.

[0126] like Figure 2 As shown, the S2 specifically includes:

[0127] S21: Format conversion. Since different data sources have different data formats, such as file types CSV, JSON, XML, data encoding UTF-8, GBK, time format, etc., it is necessary to convert the received data to a unified data format for subsequent processing and analysis.

[0128] S22: Data cleaning and outlier removal. Identify and remove outliers in the data through statistical methods such as the 3σ criterion, physical rules such as wind speed cannot be negative, or machine learning algorithms such as isolation forests.

[0129] S23: Missing value processing: For missing values ​​in the data, filling methods such as mean filling, median filling, mode filling, interpolation such as linear interpolation, polynomial interpolation or deletion are used to process them;

[0130] S24: Data standardization or normalization is to scale the data so that it falls into a small specific interval, such as between 0 and 1; normalization is to scale the data to a fixed range according to certain rules, such as maximum and minimum values;

[0131] S25: Data integration: The cleaned and pre-processed meteorological data are integrated according to time sequence, spatial location, etc. to form a complete meteorological data set for subsequent analysis.

[0132] The S3 specifically includes:

[0133] (1) Using the weighted average method, different weights are assigned to different data sources according to their credibility or accuracy, and then the weighted average is calculated as the fusion result;

[0134] Suppose there are n data sources, and the data provided by the i-th data source is x i , whose corresponding weight is w i , then the fused data x can be calculated by the following formula:

[0135]

[0136] Among them, the weight w i This can be determined based on the historical performance of the data source, accuracy assessment, or other relevant factors;

[0137] (2) Use the Kalman filter algorithm to estimate the state of the system in a recursive manner, while considering the prediction error and observation error of the model to obtain the optimal fusion result;

[0138] In Kalman filtering, the state equation and observation equation of the system can be expressed as:

[0139] x k =A·X K-1 +B·u k+ w k

[0140] Z k =H·x k +v k

[0141] Among them, x k represents the system state at time k, A is the state transfer matrix, B is the control input matrix, and u k is the control input, w k is the process noise; z k represents the observation value at time k, H is the observation matrix, v k is the observation noise.

[0142] The S4 specifically includes:

[0143] (1) Select features related to renewable energy power prediction from multi-source fused meteorological data, such as wind speed, wind direction, temperature, humidity, and air pressure;

[0144] (2) Based on the physical characteristics of renewable energy power generation and the impact mechanism of meteorological data, a mathematical model suitable for renewable energy power forecasting is constructed;

[0145] (3) Model training and optimization data division: divide the fused meteorological data and the corresponding renewable energy power generation data into training sets and test sets; use the training set data to train the selected mathematical model or machine learning model, and adjust the model parameters to minimize the prediction error; optimize the model parameters through cross-validation, grid search and other methods to improve the generalization ability and prediction accuracy of the model;

[0146] (4) Use the test set data to make predictions for the trained model and calculate performance indicators such as prediction error, accuracy, and recall rate; verify the prediction performance and reliability of the model by comparing it with actual renewable energy power generation data.

[0147] The step (2) specifically comprises:

[0148] Taking wind power as an example, the wind speed-power curve model can be expressed by the following formula:

[0149]

[0150] Where: -P is the output power of the wind turbine; -v is the wind speed; -v ci is the cut-in wind speed, i.e. the minimum wind speed at which the wind turbine starts generating electricity; -v co is the cut-out wind speed, i.e. the maximum wind speed at which the wind turbine stops generating electricity; -v r is the rated wind speed, that is, the wind speed when the wind turbine reaches the rated power; -P r is the rated power of the wind turbine; -k is the shape parameter of the power curve, usually a positive real number;

[0151] This model describes the output power of the wind turbine at different wind speeds. When the wind speed is lower than the cut-in wind speed or higher than the cut-out wind speed, the wind turbine does not generate electricity and the output power is 0. When the wind speed is between the cut-in wind speed and the rated wind speed, the output power increases with the increase of wind speed and follows a power function relationship. When the wind speed reaches or exceeds the rated wind speed, the wind turbine generates electricity at the rated power.

[0152] like Figure 3 As shown, an embodiment of the present invention provides a new energy power prediction system based on multi-source meteorological data fusion based on the new energy power prediction method based on multi-source meteorological data fusion, and the system specifically includes:

[0153] Data acquisition module 1, used to acquire meteorological data from different meteorological data sources in real time or at regular intervals;

[0154] The data processing module 2 is connected to the data acquisition module 1 and is used to perform preliminary screening and preprocessing on the acquired data, such as format conversion, outlier removal, etc.;

[0155] The data fusion module 3 is connected to the data processing module 2 and uses advanced data fusion algorithms to integrate and process meteorological data from different sources, eliminate data redundancy, and improve data consistency and reliability;

[0156] The prediction model module 4 is connected to the data fusion module 3, and constructs a mathematical model or a machine learning model suitable for new energy power prediction based on multi-source fused meteorological data; these models can capture the complex relationship between new energy power generation and meteorological elements, and output high-precision power prediction results;

[0157] The result output and evaluation module 5 is connected to the prediction model module 4, outputs the prediction results to the user in an intuitive form, such as charts, reports, etc., and performs real-time or regular evaluation on the prediction results.

[0158] The embodiment of the present invention provides a new energy power prediction system based on multi-source meteorological data fusion. The system realizes the whole process from data acquisition, processing, fusion to power prediction and evaluation through multi-module collaboration. The specific working principle is as follows:

[0159] Step 1: Data acquisition and collection

[0160] One of the core modules of the system is the data acquisition module, which is responsible for acquiring data in real time or at regular intervals from different meteorological data sources, including weather station data, satellite remote sensing data, equipment status data, and power grid load data. The diversity of data acquisition frequency and sources ensures the comprehensiveness of the data, enabling the system to grasp the meteorological conditions and equipment status information in the renewable energy power generation environment in real time. These data provide basic support for subsequent power forecasting, ensuring the real-time and accuracy of model input.

[0161] Step 2: Data Preprocessing

[0162] The acquired data first enters the data processing module, where it is initially screened and cleaned. The core tasks of data preprocessing include outlier removal, data format conversion, and data interpolation and completion. By removing abnormal data and processing missing values, the system can ensure the consistency and standardization of the data and effectively reduce the impact of data noise. The processing of this module makes the data more suitable for subsequent fusion and prediction models, improving the reliability of the data and the robustness of the system.

[0163] Step 3: Multi-source data fusion

[0164] The preprocessed data enters the data fusion module, which uses a multi-source data fusion algorithm to merge data from different data sources. The focus of data fusion is to eliminate data redundancy and optimize data consistency through weighted averaging, Kalman filtering and other technologies. For example, the weight of satellite remote sensing data is increased on sunny days, and more reliance is placed on weather station data on cloudy days, thereby ensuring the applicability of data under different weather conditions. The fused data has higher accuracy and provides comprehensive and accurate input data for the prediction model.

[0165] Step 4: Prediction model construction and power prediction

[0166] After data fusion is completed, the data will be transmitted to the prediction model module. In this module, the system will build an adaptive mathematical model or machine learning model (such as LSTM, support vector regression, etc.) according to the meteorological characteristics of renewable energy power generation. These models can capture the complex relationship between meteorological factors and renewable energy power output, and output high-precision power prediction results by analyzing key factors such as temperature, humidity, wind speed, and solar radiation. The power prediction results not only cover real-time output in the next few hours, but also provide long-term trend predictions to help optimize grid dispatching.

[0167] Step 5: Result output and user feedback

[0168] The power forecast results will be presented to the user through the result output and evaluation module. This module supports a variety of result display methods, including generating charts and reports, so that users can intuitively see the power change trend. This module also supports the generation of detailed power trend reports and provides comparative analysis with the target power. In addition, by real-time monitoring of the model prediction error, users can adjust the power station operation strategy according to the grid load demand to ensure the smooth operation of the grid.

[0169] Step 6: Forecast result evaluation and optimization

[0170] Finally, the system conducts real-time or regular evaluation of the forecast results, mainly evaluating the forecast accuracy through indicators such as mean square error (MSE) and mean absolute error (MAE), and uses this to guide model optimization. By dynamically optimizing the parameters in the forecast model and the fusion algorithm, the system can adapt to changes in different seasons and meteorological conditions, and continuously improve the forecast accuracy and system stability.

[0171] Application Example 1: Application of New Energy Power Forecasting System Based on Multi-source Meteorological Data Fusion in High Altitude Areas

[0172] A new energy power generation company manages multiple photovoltaic power stations in a plateau area at an altitude of 2,500 meters. The climate in the area is changeable, and cloud coverage and sunshine intensity vary greatly. In order to ensure the stability of power grid dispatch and optimize power generation management, the power station introduced a new energy power prediction system based on multi-source meteorological data fusion. The system realizes power output prediction within the next 24 hours through real-time data fusion and prediction analysis. The process and technical details of the implementation of the system are as follows:

[0173] 1. Data collection and preprocessing

[0174] The system data acquisition module mainly collects data from the following sources and pre-processes it based on the climate characteristics of high-altitude areas:

[0175] Weather station data: collects temperature, humidity, wind speed, wind direction and solar radiation intensity, and is updated every hour.

[0176] Sample data: The temperature in the past 7 days was -5℃ to 20℃, the humidity was 20%-70%, the wind speed was 1.5m / s to 12m / s, and the daily average solar radiation intensity was 650W / m 2 .

[0177] Satellite remote sensing data: Obtain regional light intensity, cloud thickness and other data through the satellite API, updated every 4 hours.

[0178] Equipment status data: Collect PV panel cleanliness (70%-85%), inverter operating efficiency (90%-95%), and battery energy storage status (60%-90% charge) from the SCADA system.

[0179] Grid load data: Load information sampled every 15 minutes and used to balance grid load and power generation output.

[0180] Geographic location information: The latitude (36.5°N) and altitude (2500 m) of the power station are used to calculate the impact of sunlight angle and meteorological conditions.

[0181] Data preprocessing:

[0182] Kalman filtering removes high-frequency noise from wind speed and solar radiation intensity to ensure data quality.

[0183] Linear interpolation is used to fill in the missing parts of satellite remote sensing data and improve data integrity.

[0184] Principal component analysis (PCA) extracts key features such as temperature, humidity, solar radiation intensity, and wind speed.

[0185] 2. Data Fusion

[0186] The system is based on a multi-source data fusion module to perform personalized data fusion processing of the plateau climate environment:

[0187] Weighted average method: Considering the particularity of satellite remote sensing data in high altitude areas, the weights of weather station data and satellite data are 50% and 50% respectively.

[0188] Kalman filter fusion: In clear weather, satellite data is given more weight, while weather station data is given priority in cloudy weather.

[0189] PCA analysis: Remove redundant information and retain only variables that have a greater impact on power prediction (such as solar radiation, temperature, and wind speed).

[0190] 3. Prediction model construction

[0191] Based on the fused data, LSTM, SVR and random forest models are constructed for comparative analysis:

[0192] LSTM model: A 64-unit two-layer LSTM network is used, and the input data includes weather, equipment status, etc.

[0193] Example output: The forecast for the next 24 hours is between 150kW and 300kW.

[0194] SVR model: RBF kernel function, tuned hyperparameters C = 1.5, γ = 0.15.

[0195] Output forecast value: The power in the next hour is 220kW.

[0196] Random Forest Model: Use 50 decision trees and input fusion data.

[0197] Output forecast: 225kW in the next hour.

[0198] 4. Model comparison and selection

[0199] The LSTM model outperforms other models, and the prediction accuracy fluctuates by less than 5% under different weather conditions.

[0200] LSTM model results: MSE is 12.8, RMSE is 3.6.

[0201] 5. Result output and evaluation

[0202] The power forecast curve and analysis report for the next 24 hours are output through the system interface, including power influencing factors, equipment status and cleanliness reminders.

[0203] Accuracy evaluation: The system prediction accuracy reaches 93%.

[0204] 6. System hardware and software support

[0205] The hardware includes meteorological sensor nodes and satellite receiving equipment. The software supports the use of Python's Pandas, NumPy, TensorFlow and other tools for data processing and model training.

[0206] Implementation effect: Achieve high-precision prediction in high-altitude areas, optimize power grid dispatch, increase computing efficiency by 30%, and significantly improve power generation management efficiency.

[0207] Application Example 2: Application of New Energy Power Forecasting System Based on Multi-source Meteorological Data Fusion in Coastal Areas

[0208] A coastal new energy company manages multiple photovoltaic power stations and wind power generation equipment. The company uses a new energy power prediction system based on multi-source meteorological data fusion to solve the impact of marine meteorological changes on power generation. The system implementation process is as follows:

[0209] 1. Data collection and preprocessing

[0210] The coastal climate has high humidity and large wind speed changes. Therefore, the system collects and preprocesses data based on coastal characteristics:

[0211] Weather station data: collects temperature, humidity, wind speed, wind direction and solar radiation, and is updated every 30 minutes.

[0212] Sample data: The temperature in the past 7 days was 20℃ to 32℃, the humidity was 60%-95%, the wind speed was 2.5-15m / s, and the average daily solar radiation intensity was 900W / m 2 .

[0213] Satellite remote sensing data: The satellite API obtains regional meteorological data such as light and wind speed, which is updated every 3 hours.

[0214] Equipment status data: The SCADA system collects PV panel cleanliness (80%), inverter efficiency (98%), and battery charge (75%).

[0215] Grid load data: sampled every 15 minutes, used to optimize power generation and grid load balance.

[0216] Geographical location information: The power station is located at latitude (24.1°N) and 5 meters above sea level, which is used for dynamic regulation of meteorological variables.

[0217] Data preprocessing:

[0218] Kalman filtering removes noise from solar radiation, wind speed and temperature.

[0219] Interpolation is used to fill in missing data and ensure data consistency.

[0220] PCA is used to extract key features such as temperature, humidity, wind speed, and solar radiation intensity.

[0221] 2. Data Fusion

[0222] The system uses a multi-source fusion algorithm to process meteorological data under changing weather conditions along the coast:

[0223] Weighted average method: The weights of weather station and satellite data are 60% and 40% to balance the accuracy of data sources.

[0224] Kalman filter fusion: In clear weather, satellite data is more relied upon, while in rainy weather, weather station data is prioritized.

[0225] PCA: Remove redundant variables and retain only the main factors such as temperature, humidity, and solar radiation intensity.

[0226] 3. Prediction model construction

[0227] Based on the fused data, three models are built and optimized:

[0228] LSTM model: A two-layer LSTM network with 32 and 64 units respectively, with meteorological and equipment status data as input.

[0229] Output example: The power forecast value for the next 24 hours is between 200kW and 360kW.

[0230] SVR model: RBF kernel function, hyperparameters C = 1.2, γ = 0.08.

[0231] Output forecast value: The power forecast for the next hour is 260kW.

[0232] Random Forest Model: 150 decision trees are used as input to the ensemble data.

[0233] Output forecast value: The power forecast for the next hour is 255kW.

[0234] 4. Model comparison and selection

[0235] The optimal LSTM model is selected, with an MSE of 10.5 and an RMSE of 2.9 under different weather conditions.

[0236] Stability: Under humid weather conditions, the error fluctuation is small, meeting the long-term power prediction needs.

[0237] 5. Result output and evaluation

[0238] The system outputs the power prediction results through the interface and generates an analysis report:

[0239] The chart shows: The power curve within 24 hours shows that the maximum output power is 310kW at 10 am.

[0240] Accuracy evaluation: The system showed that the prediction accuracy reached 97%.

[0241] 6. System hardware and software support

[0242] The hardware includes meteorological sensors and satellite receiving equipment. The software uses Python, TensorFlow and Scikit-learn to implement data processing and model calculation.

[0243] Implementation effect: The system application in coastal areas has significantly optimized the prediction accuracy, increased the grid dispatch efficiency by 35%, and significantly reduced the operating costs.

[0244] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0245] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A new energy power prediction method based on multi-source meteorological data fusion, characterized in that: The method includes: S1. Acquisition of meteorological data: Acquiring meteorological data from multiple meteorological data sources in real time or at regular intervals through a data acquisition module. The meteorological data includes: Weather station data: obtain temperature, humidity, wind speed, wind direction, and solar radiation intensity; Satellite remote sensing data: obtain large-scale meteorological change data, including cloud coverage and radiation intensity distribution; Online API interface: Get real-time weather forecast data from weather service providers; Local environmental sensors: monitor wind speed, wind direction, humidity, and light intensity; S2. Data preprocessing: preprocessing the acquired data using a data processing module, the preprocessing includes: Format conversion: converting data formats from different meteorological sources into a unified standardized format; Outlier removal: Identify and remove outliers in meteorological data through anomaly detection algorithms; Time series synchronization: Time align data from different sources and use linear interpolation to handle incomplete time series; Noise filtering: Use filtering algorithms to remove high-frequency noise to ensure data accuracy and reliability; S3, Data Fusion: Use the data fusion module to integrate data from different meteorological sources, including: Weighted average method: according to the precision and historical accuracy of the data source, different weights are assigned to perform weighted fusion of data at the same time point; Kalman filter algorithm: Based on the dynamic characteristics of meteorological sensors, the Kalman filter algorithm is used to optimally estimate the data; Adaptive fusion: Dynamically adjust the weight of data fusion based on real-time environmental changes; Principal component analysis (PCA): PCA is used to eliminate redundant data and extract the most representative meteorological features for prediction; S4. Prediction model construction: Using the prediction model module based on multi-source fused meteorological data, a model suitable for new energy power prediction is constructed, the model includes: Long short-term memory network (LSTM): used to capture the long-term dependency between meteorological time series data and power generation; Support Vector Regression (SVR): used to construct the nonlinear relationship between meteorological data and power output; Random forest model: constructs a power prediction model based on multi-dimensional meteorological data through multiple decision tree models; Combined model: Combines the advantages of multiple models through ensemble learning methods to output high-precision power prediction results; S5. Result output and evaluation: Use the result output and evaluation module to display and evaluate the prediction results, including: Power forecast curve: a curve chart showing the forecast of new energy power generation in the next 24 hours; Model evaluation: Use indicators such as mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE) to evaluate the performance of the prediction model; Dynamic adjustment: Based on real-time evaluation results, the model parameters are automatically adjusted through the model self-learning mechanism to optimize the prediction effect; Generate forecast report: Provide analysis and optimization suggestions on power generation trends and the impact of meteorological factors on power in the next week.

2. The new energy power prediction method based on multi-source meteorological data fusion according to claim 1 is characterized in that: The data preprocessing module uses a Kalman filter to filter the noise of meteorological data, combines the Z-score method to eliminate abnormal data, further uses linear interpolation to complete the incomplete time series data, and extracts the most representative meteorological feature data through the principal component analysis (PCA) method for subsequent model training and prediction; The data fusion module fuses meteorological data from different data sources through a weighted average algorithm, automatically adjusts weighting coefficients according to the historical accuracy and reliability of the data source, and uses an adaptive fusion algorithm to dynamically weight meteorological data to achieve redundancy elimination and optimal fusion of multi-source data.

3. The new energy power prediction method based on multi-source meteorological data fusion according to claim 1 is characterized in that: The prediction model module uses a long short-term memory network (LSTM) to model time series meteorological data, and achieves high-precision power prediction by capturing the long-term dependency between meteorological elements and new energy power output. The training process of the LSTM model uses the Adam optimizer, the learning rate is set to 0.001, and the loss function is the mean square error (MSE); the meteorological data sources specifically include: ground observation stations, satellite remote sensing platforms, numerical weather forecasting systems, etc.

4. The new energy power prediction method based on multi-source meteorological data fusion as claimed in claim 1, characterized in that: The S2 specifically includes: S21: Format conversion. Since different data sources have different data formats, such as file types CSV, JSON, XML, data encoding UTF-8, GBK, time format, etc., it is necessary to convert the received data to a unified data format for subsequent processing and analysis. S22: Data cleaning and outlier removal. Identify and remove outliers in the data through statistical methods such as the 3σ criterion, physical rules such as wind speed cannot be negative, or machine learning algorithms such as isolation forests. S23: Missing value processing: For missing values ​​in the data, filling methods such as mean filling, median filling, mode filling, interpolation such as linear interpolation, polynomial interpolation or deletion are used to process them; S24: Data standardization or normalization is to scale the data so that it falls into a small specific interval, such as between 0 and 1; normalization is to scale the data to a fixed range according to certain rules, such as maximum and minimum values; S25: Data integration: The cleaned and pre-processed meteorological data are integrated according to time sequence, spatial location, etc. to form a complete meteorological data set for subsequent analysis.

5. The new energy power prediction method based on multi-source meteorological data fusion as claimed in claim 1, characterized in that: The S3 specifically includes: (1) Using the weighted average method, different weights are assigned to different data sources according to their credibility or accuracy, and then the weighted average is calculated as the fusion result; Suppose there are n data sources, and the data provided by the i-th data source is x i , whose corresponding weight is w i , then the fused data x can be calculated by the following formula: Among them, the weight w i This can be determined based on the historical performance of the data source, accuracy assessment, or other relevant factors; (2) Use the Kalman filter algorithm to estimate the state of the system in a recursive manner, while considering the prediction error and observation error of the model to obtain the optimal fusion result; In Kalman filtering, the state equation and observation equation of the system can be expressed as: x k =A·X K-1 +B·u k+ w k Z k =H·x k +v k Among them, x k represents the system state at time k, A is the state transfer matrix, B is the control input matrix, and u k is the control input, w k is the process noise; z k represents the observation value at time k, H is the observation matrix, v k is the observation noise.

6. The new energy power prediction method based on multi-source meteorological data fusion as claimed in claim 1, characterized in that: The S4 specifically includes: (1) Select features related to renewable energy power prediction from multi-source fused meteorological data, such as wind speed, wind direction, temperature, humidity, and air pressure; (2) Based on the physical characteristics of renewable energy power generation and the impact mechanism of meteorological data, a mathematical model suitable for renewable energy power forecasting is constructed; (3) Model training and optimization data division: divide the fused meteorological data and the corresponding renewable energy power generation data into training sets and test sets; use the training set data to train the selected mathematical model or machine learning model, and adjust the model parameters to minimize the prediction error; optimize the model parameters through cross-validation, grid search and other methods to improve the generalization ability and prediction accuracy of the model; (4) Use the test set data to make predictions for the trained model and calculate performance indicators such as prediction error, accuracy, and recall rate; verify the prediction performance and reliability of the model by comparing it with actual renewable energy power generation data.

7. The new energy power prediction method based on multi-source meteorological data fusion as claimed in claim 6, characterized in that: The step (2) specifically comprises: Taking wind power as an example, the wind speed-power curve model can be expressed by the following formula: Where: -P is the output power of the wind turbine; -v is the wind speed; -v ci is the cut-in wind speed, i.e. the minimum wind speed at which the wind turbine starts generating electricity; -v co is the cut-out wind speed, i.e. the maximum wind speed at which the wind turbine stops generating electricity; -v r is the rated wind speed, that is, the wind speed when the wind turbine reaches the rated power; -P r is the rated power of the wind turbine; -k is the shape parameter of the power curve, usually a positive real number; This model describes the output power of the wind turbine at different wind speeds. When the wind speed is lower than the cut-in wind speed or higher than the cut-out wind speed, the wind turbine does not generate electricity and the output power is 0. When the wind speed is between the cut-in wind speed and the rated wind speed, the output power increases with the increase of wind speed and follows a power function relationship. When the wind speed reaches or exceeds the rated wind speed, the wind turbine generates electricity at the rated power.

8. A new energy power prediction system based on multi-source meteorological data fusion based on the new energy power prediction method based on multi-source meteorological data fusion as described in claims 1-7, characterized in that: The system specifically includes: A data acquisition module, used to acquire meteorological data from different meteorological data sources in real time or at a scheduled time; The data processing module is connected to the data acquisition module and is used to perform preliminary screening and preprocessing on the acquired data, such as format conversion, outlier removal, etc.; The data fusion module is connected to the data processing module and uses advanced data fusion algorithms to integrate and process meteorological data from different sources, eliminate data redundancy, and improve data consistency and reliability; The prediction model module is connected with the data fusion module to build a mathematical model or machine learning model suitable for new energy power prediction based on multi-source fused meteorological data; these models can capture the complex relationship between new energy power generation and meteorological elements and output high-precision power prediction results; The result output and evaluation module is connected to the prediction model module to output the prediction results to the user in an intuitive form, such as charts, reports, etc., and conduct real-time or regular evaluation of the prediction results.

9. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the new energy power prediction method based on multi-source meteorological data fusion as described in any one of claims 1 to 6.

10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the new energy power prediction system based on multi-source meteorological data fusion as described in claim 7.

Images