Multi-space-based wind and light output prediction method and system

Abstract

The invention relates to the technical field of electric digital data processing, in particular to a wind and light output prediction method and system based on multiple spaces, and the method comprises the following steps: obtaining topographic data and meteorological data, and carrying out the preprocessing of the topographic data and the meteorological data; constructing a mesoscale meteorological model; topographic features in the topographic data are extracted, and feature values are calculated; performing coordinate transformation on the topographic data to enable the topographic data and the meteorological data to be consistent in space; carrying out photovoltaic module modeling, calculating the output of the photovoltaic system, and evaluating the power generation efficiency of the photovoltaic system; constructing a wind power plant model, calculating fan output, and evaluating the performance of the wind power plant according to the fan output; and generating a prediction result according to the evaluation of the performance of the wind power plant and the evaluation of the power generation efficiency of the photovoltaic system. The method supports fine prediction under the conditions of complex terrains and changeable weather, and the high-precision CFD and the photovoltaic simulation model ensure that accurate output prediction is realized under the conditions of complex terrains and changeable weather.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical digital data processing, and in particular to a method and system for predicting wind and solar power output based on multiple spaces. Background Art

[0002] With the transition to clean energy, wind and solar energy have become key renewable energy sources. However, the output characteristics of wind and solar energy are easily affected by many factors such as meteorological conditions, terrain characteristics and time changes, showing extremely high randomness and uncertainty. Most of the current wind and solar output prediction methods are based on physical models, statistical models or machine learning algorithms, which usually have the following shortcomings: 1. High model complexity: Physical models usually require a large number of input parameters, and the calculation process is complex and time-consuming, making it difficult to achieve real-time predictions.

[0003] 2. Poor forecast accuracy: Many statistical models rely on historical data and cannot adapt to dynamic changes in meteorological conditions in real time, resulting in insufficient forecast accuracy.

[0004] 3. Single data source: Existing forecasting systems often rely on a single data source and fail to fully integrate multi-source data, affecting the reliability and stability of forecasting results. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a wind and solar power output prediction method and system based on multiple spaces.

[0006] The present invention provides a wind and solar power output prediction method based on multiple spaces, which adopts the following technical solutions: A method for predicting wind and solar power output based on multiple spaces comprises the following steps: Acquire terrain data and meteorological data, and pre-process the terrain data and meteorological data; Constructing mesoscale meteorological models; Extract terrain features from terrain data and calculate feature values; transform the coordinates of terrain data to make the terrain data consistent with the meteorological data space; Model photovoltaic modules, calculate the output of photovoltaic systems, and evaluate the power generation efficiency of photovoltaic systems; Build a wind farm model, calculate wind turbine output, and evaluate the performance of the wind farm based on wind turbine output; The prediction results are generated based on the evaluation of wind farm performance and the evaluation of photovoltaic system power generation efficiency.

[0007] In a specific embodiment, the sources of terrain data and meteorological data include ERA5, NCEP, WRF, ground stations and satellite remote sensing.

[0008] In a specific implementation scheme, preprocessing the terrain data and meteorological data includes the following steps: Remove missing values ​​and outliers from terrain and meteorological data; Complete the missing data in terrain data and meteorological data.

[0009] In a specific feasible implementation scheme, the constructed mesoscale meteorological model is based on the WRF model, and high-resolution meteorological data are generated by setting grid magnetic fields and boundary conditions.

[0010] In a specific implementation scheme, calculating the output of a photovoltaic system and evaluating the power generation efficiency of the photovoltaic system include the following steps: Based on the modeling of photovoltaic components and combined with the simulated light reception effect, the output of the photovoltaic system is calculated, and the power generation potential of the photovoltaic system is further calculated. Based on the output of the photovoltaic system, the power generation efficiency of the photovoltaic system is evaluated.

[0011] In a specific possible implementation scheme, calculating the wind turbine output and evaluating the performance of the wind farm based on the wind turbine output includes the following steps: The boundary conditions are configured according to the data output by the mesoscale meteorological model, the output of wind turbines under different meteorological conditions is simulated and calculated, and the performance of the wind farm is evaluated based on the wind turbine output.

[0012] In a specific feasible implementation scheme, before generating prediction results based on the evaluation of wind farm performance and the evaluation of photovoltaic system power generation efficiency, real-time data is acquired, input into a calculation model, real-time results are obtained, and prediction results are generated based on the real-time results.

[0013] In a specific feasible implementation scheme, after obtaining the prediction result, the deviation between the prediction result and the measured value is calculated, and the parameters of the calculation model are dynamically adjusted through an optimization algorithm.

[0014] The present invention also provides a wind and solar power output prediction system based on multiple spaces, which adopts the following technical solution: A wind and solar power output prediction system based on multiple spaces, used to execute the wind and solar power output prediction method based on multiple spaces, further comprising a data interface, through which the wind and solar power output prediction system based on multiple spaces obtains terrain data and meteorological data; The application layer is used to record interface requests and corresponding logs and error handling mechanisms, and is also used to obtain the latest forecast data; At the model layer, multiple nodes are deployed in a distributed manner, and a model is deployed on each node; Airflow platform, used to define tasks and set dependencies between various tasks; PostgreSQL library, used for relational storage of meteorological data, model configuration information, operation results, etc.; Mongodb cluster for non-relational storage of large-scale historical data and model output; Data analysis tools are used to regularly analyze the differences between historical forecast results and actual output and generate analysis result reports; Output module for report generation and downloading.

[0015] In a specific feasible implementation plan, the Airflow platform adopts a DAG (directed acyclic graph) structure; a custom operator is written in the Airflow platform to encapsulate the operation logic of each model, and the model operation results are stored through XCom.

[0016] In summary, the present invention has the following beneficial effects: 1. Integrate multi-source meteorological data (ERA5, WRF, etc.), generate high-resolution meteorological models by integrating and dynamically adjusting meteorological data at different spatial scales. This modular design enables the system to adapt to different meteorological conditions, adjust forecast parameters in real time, and ensure the accuracy and real-time performance of meteorological input.

[0017] 2. Using a variety of CFD and photovoltaic simulation tools such as OpenFOAM, SOWFA, PVLib and Bifacial__Radiance, we achieved refined evaluation of wind farms and photovoltaic systems through high-precision modeling and distributed simulation. Through multi-dimensional flow field and radiation simulation, we ensured high-precision predictions under complex terrain and changeable weather.

[0018] 3. An adaptive adjustment mechanism based on real-time data feedback has been designed to monitor meteorological changes in real time and automatically optimize the model based on the latest data, improving the accuracy and response speed of the system. This function makes up for the lack of real-time performance of existing technologies and provides support for the dynamic scheduling of wind and solar resources.

[0019] 4. Support users to generate visualization results and analysis reports based on real-time feedback, enhance decision-making support capabilities, and provide strong support for the dynamic scheduling of wind and solar resources. Through efficient integration and optimization, it has important innovation and practical value in the field of wind and solar output forecasting, and can meet the growing demand for renewable energy management.

[0020] 5. The design of the underlying storage layer uses a combination of relational databases and NoSQL databases to achieve efficient data storage and management. Regular data cleaning and analysis processes ensure the accuracy and reliability of the data, providing a solid foundation for subsequent predictions. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flow chart of the wind and solar power output prediction method based on multiple spaces.

[0022] Figure 2 It is a flow chart of wind and solar power output prediction.

[0023] Figure 3 It is a flow chart of the mesoscale meteorological model. DETAILED DESCRIPTION

[0024] The following is combined with Figure 1-3 The present invention is described in further detail.

[0025] The wind and solar power output prediction method based on multiple spaces includes the following steps: S100, acquiring terrain data and meteorological data, and preprocessing the terrain data and meteorological data.

[0026] The sources of terrain data and meteorological data can be ERA5, NCEP, WRF, ground stations and satellite remote sensing, etc. Through data cleaning, the missing values ​​and outliers in terrain data and meteorological data are removed, and the missing data are supplemented by interpolation methods.

[0027] Specifically, the Pandas and NumPy libraries in Python were used to preliminarily clean the collected terrain data and meteorological data to remove missing values ​​and outliers. The interpolation methods such as linear interpolation and spline interpolation of the SciPy library were used to complete the missing data to ensure data continuity. The integrity and continuity of the terrain data and meteorological data were ensured by preprocessing them.

[0028] It is easy to understand that the acquired terrain data and meteorological data are backed up and archived regularly. The backed up and archived data are pre-processed terrain data and meteorological data.

[0029] S200, builds a mesoscale meteorological model to provide input for wind farm simulation and photovoltaic system simulation.

[0030] The mesoscale meteorological model is based on the WRF model. By setting the grid size and boundary conditions, it generates high-resolution meteorological data of the target area. The generated high-resolution meteorological data can be used as input for wind farm simulation and photovoltaic system simulation.

[0031] S300, extracting terrain features from terrain data and calculating feature values; performing coordinate transformation on the terrain data so that the terrain data and the meteorological data are spatially consistent.

[0032] Specifically, the GDAL library and Rasterio library of the Python module are used to read high-resolution digital elevation model (DEM) data, extract terrain features such as slope, aspect, terrain undulation, etc., and further calculate the characteristic values ​​of the terrain features, which can be used for subsequent evaluation of the impact of wind speed and light on the terrain.

[0033] Perform coordinate conversion on the terrain data that has undergone format conversion to ensure the spatial consistency between the terrain data and the meteorological data.

[0034] S400: Model the photovoltaic components, calculate the output of the photovoltaic system, and evaluate the power generation efficiency of the photovoltaic system. Build a wind farm model, calculate the output of the wind turbine, and evaluate the performance of the wind farm based on the wind turbine output.

[0035] Build a wind farm CFD (computational fluid dynamics) model in OpenFOAM and configure boundary conditions based on the WRF model output data. Run SOWFA (Scale Resolving Wind Farm Simulation) to simulate the output of wind turbines under different meteorological conditions and calculate the output of wind turbines. Optimize the layout of wind turbines based on the output of wind turbines and evaluate the performance of the wind farm.

[0036] The PVLib library of the Python module is used to model photovoltaic components, and the BifacialRadiance is used to simulate the light reception effect, and then the output of the photovoltaic system is calculated. This can be used to analyze the power generation potential of the photovoltaic system under different meteorological conditions, and can also evaluate the power generation efficiency of the photovoltaic system.

[0037] S500, acquiring real-time data, and dynamically adjusting parameters of the calculation model according to the real-time data.

[0038] The computational models are mesoscale meteorological models, CFD models, SOWFA, PVLib library, BifacialRadiance and other models used to calculate and analyze the performance of wind farms and the power generation efficiency of photovoltaic systems.

[0039] After inputting real-time data, the real-time results are obtained according to the calculation results of the real-time data in the calculation model, and the wind and solar power output is predicted based on the real-time results, and the prediction results are generated. The deviation between the prediction results and the measured values ​​is calculated, and the parameters of the calculation model are dynamically adjusted to adapt to the real-time changes in meteorological conditions. The dynamic adjustment of the calculation model parameters adopts optimization algorithms such as genetic algorithms or particle swarm optimization algorithms (PSO).

[0040] Taking the particle swarm optimization algorithm as an example, a group of parameters of the computational model that need to be optimized are taken as particles, such as wind speed, wind shear, layout parameters of wind turbines in a wind farm, or solar radiation and photovoltaic component parameters in a photovoltaic simulation system.

[0041] Particle Vector:

[0042] In the above formula, For the The vector of particles, Indicates The particle parameters, .

[0043] Minimizing the prediction error of the wind and solar output model is taken as the optimization goal to improve the accuracy of the model. By adjusting the position of the particles, that is, the parameters of each model during the execution process, the error between the predicted value and the actual value is minimized. The fitness function is:

[0044] In the above formula, Indicates The fitness of each particle; is the total number of samples; Indicates The predicted value of samples; Indicates The actual value of the samples.

[0045] By updating the position and velocity of the particle, it moves towards a more direct spatial direction. The particle follows the velocity update formula:

[0046] Indicates The particle in The speed of the The updated value of the iteration; represents the inertia weight; Indicates The particle in The speed of the The updated value of the iteration; and is the learning factor; and is a random number uniformly distributed in the interval [0,1]; Indicates The particle in The historical best position on the dimension; Indicates that the entire particle group is The global optimal position in dimension.

[0047] The particle position update formula is:

[0048] In the above formula, Indicates The particle in Position on the dimension The updated value of the iteration; Indicates The particle in Position on the dimension Update value for iteration.

[0049] The parameters in the wind and solar output prediction model are adjusted through the particle swarm optimization algorithm. In each iteration, the algorithm updates the parameters based on real-time data and model calculation results to ensure that the model always adapts to changes in data parameters such as current meteorological conditions. Among them, in the wind farm simulation, the particle parameters update wind speed, wind shear, layout parameters, etc. By optimizing the particles, the wind farm wind turbine output simulation prediction is maximized. In the photovoltaic system simulation, the particle parameters can be the simulation parameters of the photovoltaic components, which can be the power, radiation intensity, temperature, etc. of the components. By optimizing the particle position, the power generation efficiency of the photovoltaic system is improved.

[0050] Finally, the stopping condition is set. When the algorithm reaches the preset stopping condition, the optimization process is terminated, and the final particle position is the optimized model parameter.

[0051] It is easy to understand that in some scenarios where real-time requirements are not high, dynamic adjustment of parameters may not be performed to reduce the computational burden. When dynamic adjustment is not performed, the prediction result is generated based on the evaluation of wind farm performance and the evaluation of photovoltaic system power generation efficiency in step S400.

[0052] S600, the calculation results are visualized and output.

[0053] The real-time results and prediction results are visualized and output in the form of generated charts, and can be converted into PDF and other formats for export as needed.

[0054] The present invention also discloses a wind and solar power output prediction system based on multiple spaces. The wind and solar power output prediction system based on multiple spaces is based on a B / S architecture. By formulating function parameter inputs, form information is provided to support the adjustment of meteorological parameters, model settings, terrain-related data, basic information of wind turbine models, photovoltaic component information, etc. This information is integrated into forms such as drop-down menus and text boxes, and the data is visualized and output. Data visualization uses d3.js to present visual views such as meteorological data, simulation data, and forecast trends. The json format is used as the data transmission format of the wind and solar power output prediction system based on multiple spaces.

[0055] The multi-space-based wind and solar power output prediction system includes a data interface, which is specifically a RESTful interface, used to integrate the data interface and input data into the system through the RESTful interface.

[0056] Interface endpoints, including interface endpoints with functions such as parameter input, data collection, model startup, result query, parameter adjustment and other related functions.

[0057] The application layer is used to record interface requests and corresponding logs as well as error handling mechanisms. It is also used to obtain the latest forecast data to facilitate the generation of visualizations including but not limited to forecast curves and data tables.

[0058] The model service layer deploys the Python environment based on the daemon process.

[0059] In the model layer, multiple nodes are deployed in a distributed manner, and each node deploys a corresponding model, including WRF model, OpenFOAM model, OpenFAST model, SWOFA model, terrain data processing model, PVLIB photovoltaic output prediction model, Bifacial Radiance photovoltaic system double-sided design and complex scene simulation model, etc. The names of the nodes are op_wrf, op_openfoam, op_openfast, op_sowfa, op_hdem, op_pvlib, op_br, etc.

[0060] Parameter scripts are used to drive the model to modify operating parameters and provide basic support for automation.

[0061] The Airflow platform is used to define tasks, set dependencies between various tasks, and ensure the logic of data flow. The Airflow platform uses a DAG (directed acyclic graph) structure to set the execution order of each task, including data collection, preprocessing, mesoscale meteorological model, wind farm and photovoltaic system simulation. Custom operators are written in the Airflow platform to encapsulate the operation logic of each model. By storing the model operation results through XCom, data transmission between different tasks is realized, and real-time data sharing between tasks is realized. The operation of each model is encapsulated as an independent task, and the parameters are passed to the tasks of the Airflow platform using the Python API. Data collection and model simulation tasks are regularly triggered through the scheduler of the Airflow platform to ensure the efficient operation of the system. In the tasks of the Airflow platform, the XCom function is used to realize data transmission and parameter sharing between different models to ensure the uniformity of model input. The parameters of subsequent models are dynamically adjusted according to meteorological data and simulation results.

[0062] The storage module is used to store the results obtained after the model runs. The corresponding storage service interface is encapsulated through the model service layer, and the results are fed back to the application layer through the service interface.

[0063] PostgreSQL library, used for relational storage of meteorological data, model configuration information, operation results, etc.

[0064] Mongodb cluster is used for non-relational storage of large-scale historical data and model outputs, storing real-time data and update records to ensure that the parameter adjustment process is traceable.

[0065] The ORM framework SQLALchemy is used to implement database access and management of the basic program framework.

[0066] Data analysis tool, used to regularly analyze the difference between historical forecast results and actual output, generate analysis result reports, and output to other interfaces through REST API.

[0067] Output module, used for report generation and download. Use Python program to generate PDF reports, including all key parameters, prediction results, analysis charts, etc. of each distributed model node, and support downloading reports to local The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A wind and solar power output prediction method based on multiple spaces, characterized by: The steps include: Acquire terrain data and meteorological data, and pre-process the terrain data and meteorological data; Constructing mesoscale meteorological models; Extract terrain features from terrain data and calculate feature values; transform the coordinates of terrain data to make the terrain data consistent with the meteorological data space; Model photovoltaic modules, calculate the output of photovoltaic systems, and evaluate the power generation efficiency of photovoltaic systems; Build a wind farm model, calculate wind turbine output, and evaluate the performance of the wind farm based on wind turbine output; The prediction results are generated based on the evaluation of wind farm performance and the evaluation of photovoltaic system power generation efficiency.

2. The method for predicting wind and solar power output based on multiple spaces according to claim 1 is characterized in that: The sources of terrain data and meteorological data include ERA5, NCEP, WRF, ground stations and satellite remote sensing.

3. The method for predicting wind and solar power output based on multiple spaces according to claim 1 is characterized in that: The preprocessing of terrain data and meteorological data includes the following steps: Remove missing values ​​and outliers from terrain and meteorological data; Complete the missing data in terrain data and meteorological data.

4. The method for predicting wind and solar power output based on multiple spaces according to claim 1, characterized in that: The constructed mesoscale meteorological model is based on the WRF model and generates high-resolution meteorological data by setting grid magnetic villages and boundary conditions.

5. The method for predicting wind and solar power output based on multiple spaces according to claim 1 is characterized in that: Calculating the output of a photovoltaic system and evaluating its power generation efficiency include the following steps: Based on the modeling of photovoltaic components and combined with the simulated light reception effect, the output of the photovoltaic system is calculated, and the power generation potential of the photovoltaic system is further calculated. Based on the output of the photovoltaic system, the power generation efficiency of the photovoltaic system is evaluated.

6. The method for predicting wind and solar power output based on multiple spaces according to claim 1 is characterized in that: Calculating wind turbine output and evaluating the performance of a wind farm based on wind turbine output includes the following steps: The boundary conditions are configured according to the data output by the mesoscale meteorological model, the output of wind turbines under different meteorological conditions is simulated and calculated, and the performance of the wind farm is evaluated based on the wind turbine output.

7. The method for predicting wind and solar power output based on multiple spaces according to claim 1, characterized in that: Before generating prediction results based on the evaluation of wind farm performance and the evaluation of photovoltaic system power generation efficiency, real-time data is obtained, input into the calculation model, real-time results are obtained, and prediction results are generated based on the real-time results.

8. The method for predicting wind and solar power output based on multiple spaces according to claim 7 is characterized in that: After obtaining the prediction results, the deviation between the prediction results and the measured values ​​is calculated, and the parameters of the calculation model are dynamically adjusted through the optimization algorithm.

9. A wind and solar power output prediction system based on multiple spaces, characterized by: Used to execute the wind and solar power output prediction method based on multiple spaces as described in any one of claims 1 to 8, further comprising a data interface, through which the wind and solar power output prediction system based on multiple spaces obtains terrain data and meteorological data; The application layer is used to record interface requests and corresponding logs and error handling mechanisms, and is also used to obtain the latest forecast data; At the model layer, multiple nodes are deployed in a distributed manner, and a model is deployed on each node; Airflow platform, used to define tasks and set dependencies between various tasks; PostgreSQL library, used for relational storage of meteorological data, model configuration information, operation results, etc.; Mongodb cluster for non-relational storage of large-scale historical data and model output; Data analysis tools are used to regularly analyze the differences between historical forecast results and actual output and generate analysis result reports; Output module for report generation and downloading.

10. The multi-space-based wind and solar output prediction system according to claim 9, characterized in that: The Airflow platform uses a DAG (directed acyclic graph) structure. Custom operators are written in the Airflow platform to encapsulate the operation logic of each model and store the model operation results through XCom.

Images