Wind Farm Control via Statistical Prediction Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wind power prediction methods are limited to macroscopic levels and longer time horizons, failing to provide accurate short-term predictions necessary for flexible energy market requirements, and do not consider high-frequency signals or mechanical loads effectively.
Innovation Solution
A statistical prediction model is used to control wind farms by reading data from one wind power plant and applying it to others, incorporating high-frequency data from sensors and historical data to predict electrical power and mechanical loads, enabling optimized operation based on energy demand and real-time measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If physical weather models are used for wind power prediction, then prediction coverage at macroscopic level is improved, but prediction accuracy for short-term and individual turbines deteriorates
Solution Approach 1:
The patent divides the wind farm into multiple zones based on spatial distribution and wind flow characteristics. Each zone is equipped with its own anemometer and prediction model, allowing for localized high-precision predictions while maintaining overall farm-wide coverage. This segmentation resolves the contradiction by enabling both broad coverage and detailed accuracy simultaneously.
Solution Approach 2:
The patent transitions from traditional temporal prediction (single time dimension) to spatio-temporal prediction by incorporating spatial coordinates and multiple measurement points. The system uses three-dimensional wind flow fields and multi-point anemometer data to predict wind power at different locations and times, adding spatial dimensionality to resolve the accuracy-coverage trade-off.
2Loss of time
If statistical prediction models are used for short-term predictions, then prediction window is reduced, but data requirements and model complexity increase
Solution Approach 1:
The patent pre-processes and stores historical wind data, turbine performance data, and meteorological data in structured databases before prediction is needed. Prediction models are pre-trained with extensive historical data to capture statistical patterns. This preliminary preparation reduces real-time computational complexity while enabling accurate short-term predictions.
Solution Approach 2:
The patent introduces intermediate variables such as wind speed at hub height, wind direction angles, and turbulence intensity as mediators between raw sensor data and final power predictions. These intermediate representations simplify the statistical modeling process by capturing essential physical relationships, reducing model complexity while maintaining prediction accuracy.
3Measurement precision
If high-frequency sensor data is incorporated for mechanical load prediction, then prediction accuracy is improved, but data processing requirements increase
Solution Approach 1:
The patent extracts and separates different frequency components of sensor data using signal processing techniques. High-frequency components related to mechanical loads are extracted and processed separately from lower-frequency wind speed variations. This extraction allows focused processing of only the relevant high-frequency data needed for load predictions, reducing overall processing requirements while maintaining accuracy.
Solution Approach 2:
The patent transforms raw high-frequency sensor data into aggregated statistical parameters such as root-mean-square values, peak factors, and spectral density estimates. By changing the representation from raw time-series data to condensed statistical parameters, the system reduces data volume and processing complexity while preserving the essential information needed for accurate mechanical load predictions.
Data Source
AI summary
A method (300) for the control of a wind farm (10) is disclosed. The method (300) comprises: read-in of data from at least one first wind power plant (200) of the wind farm; supply of the read-in data from the at least one first wind power plant to a statistical prediction model for the control of at least one second wind power plant (200) of the wind farm based on the read-in data from the at least one first wind power plant; and use of the statistical prediction model to control the at least one second wind power plant (200).