Tunable Power Forecasting Model for Wind Farm Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wind farms face performance constraints due to variability in power production caused by changing wind conditions, leading to inaccurate short and medium-term forecasts, resulting in penalties and revenue loss, as existing persistence models fail to account for sudden changes and are complex and time-intensive.
Innovation Solution
A method and system using a tunable power forecasting model with an asymmetric loss function to generate accurate power production forecasts, allowing selective favoring of over or under forecasting based on differential penalties, and incorporating a wind farm management system for real-time data processing and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a persistence forecasting model is used for short and medium term power forecast, then the forecasting method is simple to implement, but the forecast accuracy deteriorates during ramp-up events leading to consistent under-forecasting
Solution Approach 1:
The forecasting model transitions from a static persistence approach to a dynamic hybrid model that adapts to changing wind conditions. The system dynamically selects between persistence forecasting and ramp-up detection mechanisms based on real-time wind data patterns, allowing the model to respond flexibly to ramp-up events while maintaining simplicity during stable conditions.
Solution Approach 2:
The forecasting system combines two different forecasting approaches (persistence model and ramp-up detection model) into a hybrid composite model. This composite approach integrates the simplicity of persistence forecasting with the accuracy of ramp-up event detection, creating a unified forecasting system that leverages the strengths of both methods.
2Measurement precision
If complex wind turbine parametric data techniques are used to address persistence model limitations, then forecast accuracy may improve, but the system complexity and time intensity increase making them unsuitable for short term forecasting
Solution Approach 1:
The system extracts only the essential features needed for accurate short-term forecasting from the complex wind turbine parametric data. Instead of using full parametric models, the invention extracts key wind condition indicators and ramp-up detection parameters, discarding unnecessary complexity while retaining the critical information needed for accurate forecasting.
Solution Approach 2:
The forecasting system is segmented into distinct functional modules: a persistence forecasting component, a ramp-up detection component, and a hybrid integration component. This segmentation allows each module to operate independently with appropriate complexity levels, combining simplicity where possible with targeted complexity only where needed for ramp-up event detection.
3Use of energy by moving object
If persistence model forecasting is used, then computational resources are minimized, but revenue loss increases due to inability to sell total power produced and curtailment penalties
Solution Approach 1:
The system performs preliminary detection of ramp-up events using simplified wind condition monitoring before the actual forecasting period. By identifying impending ramp-up events in advance, the system can prepare more accurate forecasts proactively, preventing revenue loss from curtailment penalties while maintaining low computational resource usage during normal operating conditions.
Data Source
AI summary
The system and method described herein relate to production of power from the wind farm that incorporate tunable power production forecasts for optimal wind farm performance, where the wind farm power production is controlled at least in part by the power production forecasts. The system and method use a tunable power forecasting model to generate tunable coefficients based on asymmetric loss function applied on actual power production data, along with tuning factor(s) that tune forecast towards under forecasting or over forecasting. The power production forecasts are generated using the tunable coefficients 34 and power characteristic features that are derived from actual power production data. The power production forecasts are monitored for any degradation, and a control action to regenerate the coefficients or retune the model is undertaken if degradation is observed.