AI Wind Profile Prediction for Drone Turbulence Safety
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wind condition prediction technologies, such as those using Doppler LiDAR and machine learning, lack clarity on the specific data and machine learning methods required for accurate wind condition prediction, limiting their implementability.
Innovation Solution
A wind condition learning device employing AI for supervised learning using a dataset that includes wind condition altitude distribution model values following a power law on the inflow side and turbulence energy or intensity in environmental spaces obtained by simulation, enabling the prediction of wind conditions for drone safety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine learning is used to predict wind conditions, then prediction capability is improved, but implementation clarity and reproducibility deteriorate due to lack of specific data and method specifications
Solution Approach 1:
The patent specifies exact parameters for machine learning implementation: using wind speed at 10m height (V10) and wind direction at 10m height (D10) as input features, with power law exponent 0.14 for altitude distribution. These concrete parameter specifications transform the vague machine learning approach into a reproducible implementation while maintaining prediction reliability.
2Measurement precision
If Doppler LiDAR measurement points are increased, then measurement accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The patent introduces altitude as a new dimension for wind measurement. By measuring wind conditions at multiple altitude levels (10m, 50m, 100m, etc.) and applying power law relationships, the system achieves comprehensive wind field characterization with a single vertical profiler rather than requiring multiple horizontal measurement points, thus reducing device complexity while maintaining measurement precision.
3Measurement precision
If power law model with specific exponent is used for wind altitude distribution, then prediction accuracy is improved, but adaptability to different terrain conditions deteriorates
Solution Approach 1:
The patent specifies power law exponent 0.14 for wind altitude distribution, which provides accurate predictions for standard atmospheric conditions. This fixed parameter approach ensures consistency and reliability for typical applications, while the model structure allows for parameter adjustment when dealing with non-standard terrain conditions, balancing accuracy and adaptability.
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 device provides a clear mode of machine learning for wind condition prediction, enabling real-time wind condition forecasting for drone safety, preventing inadvertent movement due to wind pressure, and contributing to safe flight operations.
Implementation Method 1
A Doppler LiDAR is known as a device that measures a wind condition
Implementation Method 2
one piece of the learning data set is a wind condition altitude distribution model value following a power law
Data Source
AI summary
A wind condition learning device according to the present disclosed technique includes: an input terminal to which a learning data set is input; and a calculator including AI to perform learning on the basis of the learning data set, in which one piece of the learning data set is a wind condition altitude distribution model value following a power law on an inflow side, and the other piece of the learning data set includes a wind speed average value, a wind speed maximum value, turbulence energy, or turbulence intensity in a wind condition distribution of an environmental space obtained by simulation.