2020 年 86 巻 883 号 p. 19-00369
In this paper, we propose a new method to identify the error factors between an actual structure and the finite element (FE) model. The FE model updating using sensitivity and optimization algorithm conventionally is used to ensure high accuracy. Parametric design variables such as thickness and elastic modulus are adopted as updating variables in many cases. However, their application is limited because the actual error factors are often non-parametric variables. The existing concept of utilizing the driving point frequency response function (FRF) has often been applied in engineering. With this technique, the driving point FRFs of the actual structure and FE model are compared to determine the error factors. However, it is difficult to accurately recognize such factors. This is because FRF is calculated from time history, which includes the reflection wave of the entire system. To overcome this limitation, a new method is proposed that involves the use of driving point transient response. When a particular component is excited, its vibration propagates as a wave that moves toward the entire system. The driving point transient response is measured immediately after excitation and is restricted to the excited component only. The comparison of the driving point transient response of the actual structure and the FE model response enables us to determine if the excited component is the cause of the error. To determine the contribution of the excited component to the transient response, a time domain mutual mean compliance is used. As an application, the proposed method using the driving point transient response and the mutual mean compliance is applied to a simple beam structure.
