Wind Farm Layout Optimization Using Physics-Informed Wind Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wind farm modeling methods face challenges due to multiscale nature spanning from atmospheric scales to wind turbine boundary layer, leading to excessive modeling complexity and computational cost, and insufficient accuracy in wind speed prediction, which affects the stability and reliability of wind power generation.
Innovation Solution
A physics-informed neural network model is constructed by incorporating partial differential equations into a neural network architecture, using automatic differentiation to embed physical constraints into the loss function, and training with data from lidar sensors to predict wind speed and optimize turbine layout.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computational fluid dynamics (CFD) methods are used for wind field modeling, then simulation accuracy is improved, but computational cost and modeling complexity increase significantly
Solution Approach 1:
The patent replaces traditional computational fluid dynamics (CFD) mechanical modeling approaches with a neural network-based computational model. The neural network learns wind field patterns from historical data and predicts wind speed distributions without requiring complex physical equations, thereby maintaining prediction accuracy while significantly reducing modeling complexity and computational burden.
Solution Approach 2:
The patent uses neural networks to create a virtual copy of the wind field behavior based on historical measurement data. Instead of solving complex physical equations, the model learns and replicates wind field patterns from training data, enabling accurate predictions with simpler computational structures.
2Productivity
If larger row and column spacing between turbines is used, then wake effects are reduced and power output increases, but the overall layout scale of wind farms increases
Solution Approach 1:
The patent employs dynamic layout optimization that adapts turbine positioning based on predicted wind field conditions. Rather than using fixed spacing, the system dynamically adjusts turbine placement to minimize wake effects while maximizing power output, allowing for more compact wind farm layouts that maintain high productivity through condition-dependent optimization.
3Speed
If traditional neural networks are used for wind speed prediction, then computational speed is improved, but prediction accuracy is insufficient
Solution Approach 1:
The patent segments the wind field prediction problem into multiple neural network models that operate at different spatial and temporal scales. This segmentation allows each sub-model to focus on specific aspects of wind behavior, improving overall prediction accuracy while maintaining computational efficiency through specialized, modular architectures.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The model provides high-precision, ultra-short-term velocity field predictions, enhancing wind farm efficiency and stability by integrating physical laws into the neural network, reducing computational costs and improving accuracy.
Implementation Method 1
calculating partial derivatives of partial differential equations using automatic differentiation
Implementation Method 2
The physical equations include two-dimensional Navier-Stokes equations, continuity equation, and turbulence model
Implementation Method 3
obtaining wind speed data at a specific location in a velocity field based on the physics-informed neural network model
Data Source
AI summary
A method, system, and device for wind speed prediction and layout optimization in wind power generation are provided. The method includes: obtaining a basic wind resource dataset of a target region; constructing a physics-informed neural network model based on the basic wind resource dataset; obtaining wind speeds data at a specific location in a velocity field based on the physics-informed neural networks and constructing a training dataset; training the physics-informed neural network model based on the training dataset; reconstructing a wind speed distribution within the velocity field and predicting wind speeds for a next time period with a wind farm using the trained physics-informed neural network model; and optimizing a layout of a wind turbine cluster based on a reconstructed wind speed distribution within the velocity field. The present application reconstructs a two-dimensional velocity field of the wind farm by training the PINN and enables accurate ultra-short-term wind speed prediction.