Machine-Learning Wind Data Forecasting Without On-Site Measurements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wind turbine installations face challenges in accurately predicting wind conditions at specific locations, which affects efficiency and stability, and require time-consuming and error-prone on-site measurements to assess potential yield.
Innovation Solution
A method using machine learning to predict wind conditions by training a prediction model with transformed data sets from various locations, including geographical and meteorological features, to provide accurate wind data without on-site measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If on-site measurements are conducted to assess wind conditions, then measurement precision is improved, but loss of time and productivity deteriorate due to time-consuming field work
Solution Approach 1:
The system performs preliminary wind condition assessments using historical data and machine learning models before actual turbine installation or operation. By pre-calculating wind resource potential and turbulence characteristics at candidate locations, the need for extensive on-site measurements is reduced, saving time while maintaining adequate precision for decision-making
Solution Approach 2:
The system creates virtual replicas of wind conditions by training machine learning models on historical measurement data from reference stations. These learned models can predict wind parameters at target locations without requiring physical measurements, effectively copying the behavior of actual wind fields through data-driven simulations
2Measurement precision
If on-site measurements are conducted to assess wind conditions, then measurement precision is improved, but device complexity and ease of operation worsen due to manual data collection processes
Solution Approach 1:
The system enables automated wind resource assessment that performs itself without manual intervention. Machine learning models automatically process historical data from public databases and reference stations, generate predictions for target locations, and provide results without requiring field technicians to conduct measurements or manually process data
Solution Approach 2:
The system replaces physical measurement devices and manual data collection processes with computational models. Instead of using anemometers and other hardware to measure wind conditions at candidate sites, the system uses machine learning algorithms that process existing data to predict wind parameters, substituting mechanical measurement systems with information processing systems
3Productivity
If machine learning prediction models are used to forecast wind conditions, then productivity is improved by eliminating on-site measurements, but measurement precision may deteriorate without direct field data
Solution Approach 1:
The system incorporates feedback mechanisms where prediction results are continuously refined using actual measurement data when available. The machine learning models can be retrained and validated against real-world measurements to improve accuracy, creating a closed-loop system that learns from discrepancies between predictions and actual conditions
Solution Approach 2:
The system is designed to handle multiple data sources and prediction scenarios universally. It can process data from public databases, reference stations, and direct measurements, and can adapt its prediction approach based on data availability. This multi-functional capability allows the system to maintain reasonable precision across different contexts while always providing productivity benefits
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Method for providing wind data at a forecast location, comprising: providing (S101) training data sets for a plurality of installation locations for wind turbines, in particular comprising training data sets from public databases, preparing (S103) the training data sets for machine learning by transforming the training data sets into features, training (S105) a prediction model for predicting at least one statistical wind condition at a forecast location based on the features, obtaining (S107) a target forecast location, in particular from a CRM system, predicting (S109) the at least one statistical wind condition at the target forecast location using the trained prediction model, and providing (S111) wind data comprising the predicted at least one statistical wind condition, in particular providing the wind data for the CRM system.