実験と有限要素解析をハイブリッドして求めた状態遷移行列による振動解析

抄録

In complex structures such as machine tools and automobiles, much effort is required to accurately model the entire system. In addition, the miniaturization and performance of sensors has made it possible to collect a large amount of timeseries data simultaneously, and experimental identification of the state transition matrix of the entire system from such data is now being conducted. However, experimentally identified state transition matrices alone cannot predict vibration changes associated with structural changes. In this paper, we propose a method for predicting vibration changes by identifying the state transition matrix using a finite element model of the part undergoing structural change and experimentally measured time series data of the entire product. The method identifies the state transition matrix in the frequency domain using the z-transform by converting the displacement and velocity, which are difficult to measure, into acceleration and jerk, and then identifying the state transition matrix using only acceleration data. Numerical verification of the proposed method was also conducted.