Damping Identification of an Offshore Wind Turbine
A Predictor-Based Subspace Identification Approach
More Info
expand_more
Abstract
Increasing the capacity of offshore wind energy is a necessary step in the transition from fossil fuels to renewable energy sources. In order to accelerate this process, the costs of wind energy should be reduced to make it more competitive with traditional energy sources. Cost reduction could be achieved by increasing the lifetime of an Offshore Wind Turbine (OWT). The lifetime is directly affected by the amount of damping present in an OWT. Hence, it is vital to have accurate estimates of the damping in order to predict the lifetime. All Offshore Wind Turbines are equipped with a bidirectional accelerometer, resulting in large amounts of data that can be used for identification purposes. Using Operational Modal Analysis (OMA) it is possible to estimate the structural parameters of an OWT from output measurement data. Due to technical limitations, however, the available measurement data does not have a constant sampling frequency. Consequently, conventional OMA methods based on discrete-time models cannot be applied directly. The goal of this thesis is to investigate whether the damping of an OWT can be accurately estimated using non-uniformly sampled measurement data. Two different identification approaches are taken to answer this problem. The first approach consists of resampling the measurement data with a constant sampling frequency and subsequently applying OMA methods. The second approach uses a continuous-time identification method, which can be applied directly to non-uniformly sampled measurement data. In this thesis, the Predictor-Based Subspace IDentification (PBSID) method is used for the identification of measurement data. This method is based on a discrete-time state-space representation of a system. Using Laguerre projections the PBSID algorithm can also be extended to the continuous-time domain. The performance of both the discrete and continuous-time PBSID algorithms is examined by applying them to a simulation example. It is observed that both methods are able to provide damping estimates of similar accuracy. Following the simulation example, both identification approaches are applied to wind turbine simulation and measurement data, respectively. Furthermore, the results from these two approaches are compared with the results obtained from uniformly sampled measurement data taken from an experimental OWT. It is concluded that accurate damping estimates of an OWT can be obtained from non-uniformly sampled measurement data.