Machine-Learning Wind Data Prediction Without On-Site Measurements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting wind conditions at wind turbine locations are inaccurate and require time-consuming on-site measurements, which can lead to errors and increased setup time.
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, allowing for accurate estimation of average wind speed, turbulence intensity, and other parameters without on-site measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If on-site wind speed measurements are performed continuously and in real time, then measurement precision is improved, but loss of time increases due to time-consuming setup and measurement periods
Solution Approach 1:
The patent applies preliminary action by training machine learning models in advance using historical wind data from multiple sources (reanalysis data, numerical weather prediction models, and measured data from reference stations). This pre-training enables the model to make accurate predictions without requiring time-consuming on-site measurements, thus resolving the contradiction between measurement precision and time loss.
Solution Approach 2:
The patent uses copying by creating virtual representations of wind conditions through machine learning models that replicate the behavior of physical measurement systems. The model learns from copied historical data patterns and generates predictions that match the precision of actual measurements without requiring physical presence at the measurement location, thereby eliminating setup time while maintaining accuracy.
2Measurement precision
If machine learning models are trained with comprehensive data from multiple sources, then prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The patent introduces an intermediary layer (the machine learning model) that processes and integrates data from multiple complex sources including reanalysis data, numerical weather prediction models, and reference station measurements. This intermediary model simplifies the complexity by learning patterns from diverse data sources and providing unified, accurate predictions, thus resolving the contradiction between prediction accuracy and system complexity.
Solution Approach 2:
The patent applies universality by designing a multi-functional machine learning model that can process and integrate various types of data (reanalysis data, NWP model data, measured data from multiple reference stations) through a single unified framework. This universal approach improves prediction accuracy while managing complexity by using one model to handle multiple data sources rather than separate systems for each.
Data Source
AI summary
The present disclosure is directed to method for providing wind data at a prediction location includes providing training data sets for a plurality of installation locations for wind turbines, wherein the training data sets are obtained from public databases, preparing the training data sets for machine learning by transforming the training data sets into features, training a prediction model for predicting at least one statistical wind condition at the location based on the features, obtaining a target location, in particular from a Customer Relationship Management (CRM) system, predicting the at least one statistical wind condition at the target location using the trained prediction model, and providing wind data including the predicted at least one statistical wind condition to the CRM system.