Objetivo
In order to be able to predict the energy yield available at a potential wind farm site, accurate predictions of the wind regime at that site are required. Improvements in the wind speed and direction predictions will reduce the uncertainty in the available energy yield, which in turn will reduce the financial risks of the wind farm development. Current wind resource prediction methods have been found to contain errors of up to 10% in wind speed and 60 in direction. This translates to 15 - 20% errors in predicted energy yield.
This project will design and implement an improved Measure-Correlate-Predict (MCP) algorithm by using neural network techniques. Neural networks are particularly good at extracting patterns from noisy time series data which is exactly the problem facing MCP techniques.
The objectives of the project are:
1. design and develop a model a neural network which will result in a 50% improvement in the accuracy of the predicted long term wind speed compared with conventional measure correlate predict techniques;
2. quantify the uncertainties in wind speed and direction predictions;
3. translate the uncertainties in wind climate in to energy yield.
Achievement in these objectives should result in the following benefits:
- a substantial reduction in the financial risk of investment in wind power projects
- improved understanding of the physical parameters connected to wind speed and direction analysis
- transfer of neural network knowledge into the wind energy industry
The project will construct a comprehensive database of wind measurements. This will be carefully analysed to ensure the optimal neural network approach is used. Upon completion of the neural network algorithm, a user friendly software tool will be developed that provides easy access to highly accurate wind resource predictions. These predictions will be compared against the current state-of-the-art. The effect of the improved accuracy on energy yield will be calculated. Finally the reduction in financial uncertainty due to the new MCP method will be quantified.
Ámbito científico
- natural sciencescomputer and information sciencessoftware
- natural sciencescomputer and information sciencesdatabases
- engineering and technologyenvironmental engineeringenergy and fuelsrenewable energywind power
- natural sciencescomputer and information sciencesartificial intelligencecomputational intelligence
Convocatoria de propuestas
Data not availableRégimen de financiación
CSC - Cost-sharing contractsCoordinador
AL1 3AW ST ALBANS
Reino Unido