Modal identification of boiler plant structures on AR spectral analysis of seismic records

This paper deals with a modal identification method for large-scale structures such as boiler plants in thermal power station. Practical and accurate modal identification has been carried out by the proposed method, which is composed of two stages; processing frequency transfer functions by autoregressive (AR) spectral analysis, and a curve-fitting technique to extract modal parameters. Seismic records of base acceleration records at various points of the structure are used as multi-output data. This method is examined using time-series data of seismic response simulation. Introduction of the two techniques, namely, decimation of time data and FPE criterion to optimize the order of AR models have realized effective and accurate identification. This method has actually been applied to seismic observation data of boiler plants in operation. As a result of this study, the authors' modal identification has proven to be effective for seismic modeling of large-scale structures