Triaxial accelerometer based azimuth estimator for horizontal axis wind turbines

Abstract

One of the elements that receives the greatest stresses is the main shaft. Its damage is directly related to the cyclical nature of its rotational motion. However, the vast majority of horizontal axis wind turbines (HAWT) do not have sensors to measure the main-shaft angular position (azimuth), or they are not always easily accessible. Using a main-shaft placed single triaxial accelerometer for the estimation of the azimuth is proposed as a low intrusion approach that can be easily deployed in machines already in use. An approach using a tandem of two extended Kalman filters (calibration/prediction), aiming for a precise and robust estimation, is presented. The estimator is able to calibrate for accelerometer positional and orientation errors, as well as for bias drift. To simplify the burden of deployment, a simple procedure is proposed to determine the covariance matrices for a particular HAWT from those determined in a synthetic case. The proposed approach is analyzed using synthetic data, OpenFAST simulation of NREL-5MW HAWT. It outperforms the ATAN naive approach by an order of magnitude, showing errors smaller than 0.4o. The filter shows a good behavior, coherent with that of the synthetic setup, when tested on experimental data obtained from a 3MW HAWT.