Averaged cov-driven subspace identification for modal analysis of a modified troposkien blade with displacement measurement

An operational modal analysis study has been carried out on a 1- kW, vertical-axis wind turbine rotor blade, using a stereo vision technique. Numerical simulation has also been carried out and results were compared to classical modal analysis and OMA results. The displacement time series used in the analysis are very short because of limitations in the image acquisition system. Short time series are not fully qualified for OMA and analyzing the data needs a proper method. Covariance driven Stochastic Subspace Identification method (SSI-cov) has been used for short time series like earthquakes. In the SSI-cov method, a block Toeplitz matrix is formed which contains output correlation functions. 10 displacement time series have been recorded with 187 Hz sampling rate, and about 3 time series were chosen to be analyzed. The block Toeplitz matrix of 3 time series are averaged out and the procedure how to analyse short time series is described in this paper. For OMA validation, a classical modal analysis is done with signals from accelerometers mounted on the blade. The blade is exited with a hammer. The natural frequencies are estimated by picking the peaks from the FRFs (Frequency Response Function) obtained at different points. Finally the natural frequencies and mode shapes obtained from classical modal analysis and OMA results are compared to numerical simulations of the blade with COMSOL.