Evaluation of System Identification Methodology and Application.

摘要

As part of an assessment of the role of dynamic testing in the design of nuclear power plant (NPP) structures, the state of knowledge concerning system identification methodologies and applications is evaluated. In general, it is found that system identification is limited to parameter estimation for a priori chosen models. The well developed parameter identification techniques assume linear system behavior. For structural applications the most commonly used method is the modal analysis technique. Nonlinear parameter estimation and generalized system identification are still in the developmental stage. In applications to NPP buildings system identification can play a useful role in the verification and improvement of design methods. Low-level testing of NPP structures is feasible and would suffice to determine many of the parameters of linear design models. Since testing of NPP structures at design level excitations is not practical, the verification by system identification of design methods in this regime requires an appropriate databank. To apply system identification methods to NPP structures the accuracy and reliability of these techniques must be assessed. No systematic attempt of verifying even the current, well developed methods with real data can be documented.