2024 年 70A 巻 p. 303-312
In recent years, studies has been conducted to detect bridge damage from changes of vibration characteristics. We studies estimation method of bridge vibration characteristics from dynamic responses with high accuracy using stochastic subspace identification (SSI) with frequency bands separation method. However, trial and error is necessary to determine the estimation parameters. In this study, we examine the optimum value of model order and singuler vector length in SSI and freqency threshold in frequency bands separation method. As a result of examination, optimal value for the three parameters are found, and frequencies are estimated by SSI using these parameters. In addition, dominant frequencies are detected from kernel density estimation.