Wind Data Forecasting With Machine Learning Before Turbine Installation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The efficiency of wind turbines depends on prevailing wind conditions, and installing them in locations with excessive loads can compromise their integrity and stability, necessitating accurate wind condition prediction before installation.
Innovation Solution
A method using machine learning to predict wind conditions at a forecast location by training a prediction model with transformed training data sets from various sources, including public and internal databases, to provide precise wind data without on-site measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If on-site measurements are conducted to obtain accurate wind data, then measurement precision is improved, but loss of time and productivity deteriorate due to the lengthy measurement period required
Solution Approach 1:
The patent applies preliminary action by pre-training a machine learning model using historical wind data from multiple locations before actual wind turbine installation. The model is trained in advance on comprehensive datasets including terrain information, climate data, and wind measurements from various sources. When a new installation location is assessed, the pre-trained model can rapidly predict wind conditions without requiring lengthy on-site measurements, thus resolving the contradiction between measurement accuracy and time consumption.
2Reliability
If comprehensive training data from multiple sources is used to train the prediction model, then reliability of wind condition prediction is improved, but device complexity worsens due to the complexity of data processing and model training
Solution Approach 1:
The patent applies universality by creating a multi-functional machine learning model that processes diverse data types from multiple sources including public databases, private databases, terrain information, and climate data. The single model handles various input formats and predicts multiple wind parameters (average wind speed, turbulence intensity, extreme wind speeds) simultaneously. This universal approach improves prediction reliability while managing complexity through a unified processing framework rather than separate systems for each data type.
Solution Approach 2:
The patent uses feature transformation as an intermediary step between raw multi-source data and the prediction model. The training data is transformed into standardized features that the machine learning model can process efficiently. This intermediary transformation layer simplifies the complexity of handling diverse data sources by converting them into a common format, allowing the model to focus on learning patterns rather than managing data heterogeneity.
3Productivity
If machine learning is used to predict wind conditions without on-site measurements, then productivity is improved by accelerating the installation process, but measurement precision deteriorates compared to actual on-site measurements
Solution Approach 1:
The patent applies preliminary action by pre-training the machine learning model extensively using comprehensive historical data from multiple verified sources before deployment. The model learns from decades of wind measurements across various terrains and climates, capturing complex patterns and relationships. When deployed at new locations, this pre-trained knowledge enables rapid predictions with high accuracy, resolving the contradiction between speed and precision by doing the heavy learning work in advance rather than during site assessment.
Solution Approach 2:
The patent applies copying by using the machine learning model to replicate wind patterns and conditions from known locations to predict conditions at new locations. The model learns from copied historical data and patterns, transferring knowledge from training locations to forecast locations. This copying approach enables rapid prediction without physical measurements while maintaining accuracy through pattern recognition and statistical modeling.
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.