Renewable Energy Site Layout Using Terrain-Aware Weather Downscaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for planning renewable energy sites, such as wind and solar energy plants, are computationally expensive and unreliable due to the use of simulations that require large data sets and complex terrain analysis, leading to inefficiencies in layout planning.
Innovation Solution
A method using a deep learning algorithm to derive a data model from correlated historical meteorological and terrain data, allowing for fast and reliable downscaling of meteorological data to increase resolution and optimize renewable energy site layout.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If simulations are performed using large data sets and complex terrain models to obtain operational conditions, then measurement precision and reliability are improved, but computing time and computational cost increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-training a deep learning model on historical meteorological data and terrain information before actual operational conditions are needed. The model learns patterns and relationships in advance, so when operational conditions are required, the pre-trained model can quickly generate accurate predictions without performing time-consuming simulations from scratch.
Solution Approach 2:
The patent uses copying by creating a virtual representation of the complex simulation system through a deep learning model. Instead of repeatedly executing full physical simulations, the trained model serves as a computational copy that replicates simulation outcomes much faster, maintaining accuracy while dramatically reducing computing time for operational condition assessments.
2Measurement precision
If high-resolution climate models are used for dynamical downscaling to improve data resolution, then measurement precision is improved, but computational cost and complexity increase
Solution Approach 1:
The patent applies mechanics substitution by replacing the complex mechanical/dynamical downscaling models with a data-driven deep learning approach. Instead of using physical climate models that require solving complex differential equations, the system uses a neural network trained on historical data to predict high-resolution meteorological conditions, substituting computational physics with statistical learning.
Solution Approach 2:
The patent applies parameter changes by transforming the input meteorological data into enhanced resolution output through the deep learning model. The model learns to map lower-resolution input parameters to higher-resolution output parameters by identifying patterns in historical data, effectively changing the resolution parameter without requiring complex downscaling physics.
3Productivity
If statistical downscaling methods are used to reduce computational cost, then computing time is reduced, but reliability decreases due to inconsistent results
Solution Approach 1:
The patent uses copying by training a deep learning model on extensive historical meteorological data to create a reliable computational representation. The model learns consistent patterns and relationships from the training data, providing reliable and reproducible results for operational conditions while maintaining computational efficiency, unlike ad-hoc statistical methods.
Solution Approach 2:
The patent applies feedback by using historical operational conditions and meteorological data to train and validate the deep learning model. The model's predictions are continuously refined through training on past data, ensuring that it learns accurate relationships between meteorological conditions and operational performance, thereby improving reliability through iterative learning from historical feedback.
Data Source
AI summary
Correlated sets of historical meteorological data and terrain data are obtained for at least one geographical area. A data model is derived based on the basis of the correlated sets, by training the data model. The trained data model is adapted to identify coherence between meteorological data and terrain data relating to the same geographical area. Meteorological data and terrain data related to the renewable energy site are fed to the trained data model, the terrain data having a higher resolution than the meteorological data. Using the trained model, meteorological data related to the renewable energy site with increased resolution is estimated by downscaling the meteorological data. The estimated meteorological data with increased resolution for the renewable energy site is then used for planning a layout of the renewable energy site.