DIAGNOSTICS OF WIND TURBINES BASED ON INCOMPLETE SENSOR DATA

Zs. J. Viharos, K. B. Kis

Abstract

A typical wind turbine monitors tens of parameters such as temperatures at different locations, rotation speed of the components, power produced, availability, etc. In many cases sensor data are not collected and stored continuously, because of different reasons like sensor or communication failure, storage size restrictions, condition and situation based information collection. The amount of the resulted incomplete information is typically a significant part of the whole collected dataset; consequently, there is a requirement for such diagnosis solutions that are able to handle incomplete data.<BR />The paper introduces an artificial intelligence based solution for exploring dependencies among monitoring parameters using up the whole incomplete dataset in order to serve with reliable models for supervision of wind turbines.

Keywords

artificial neural network, incomplete data, supervision and diagnostics, wind energy

Download

IMEKO-WC-2012-TC10-O10.pdf

DOI

-

IMEKO TC

TC10 - Measurement for Diagnostics, Optimization and Control

Back to the proceedings