| dc.description.abstract | This master thesis project concerns the structural monitoring applied to a windturbine. The algorithm analysed in this master thesis is an extended Kalman Filter called the joint input-state estimation algorithm.
Kalman Filter algorithm is usually fed with a measured signal that is to be filtered and some model parameter. The filtering process is based on a weighted average between the model and the measure at each time step. The weighting process involve calculation of noise covariance matrices. This master thesis is an analysis on Kalman Filtering in case of an operating windturbine, the measured signal correspond to motion data, either displacement, velocity or acceleration taken at a given height. As on a windturbine, the excitation force applied through the rotor can not be exactly known, an extended version of the Kalman Filter called the joint input-state estimation algorithm will be used. This algorithm estimate at the same time the state and the input -excitation force - of the system.
A special attention is given to detection and quantification of the modeling error which is the main personal contribution to this field. A process to include modeling error in calculation of covariance matrices is given and an analysis of the impact of each dynamical model parameter error on the overall estimation error is carried out. Besides, the stability of the algorithm is also a concern in this thesis. Finally, an optimisation of the sensor layout is performed.
As the Kalman Filter is linear, the algorithm will be analysed for periodic sinusoidal signals as every signal can be seen as the infinite sum of harmonic signals. Then, algorithm will be analysed on a virtual windturbine through FedEm. | |
