|
Abstract
|
Early identification of damages in tubular structures is crucial for their
long-term safety and functionality, as they are essential in various
modern life applications. Experimental and numerical modal data
may slightly differ due to unknown structural characteristics and
uncertainties, which are typically addressed using finite element
(FE) model updating procedures. Instead of using the Euler-
Bernoulli beam element, this paper utilises the semi-rigidly connected
frame element (S-RCFE). By incorporating extra design parameters,
such as the end fixity factor of all connections, the S-RCFE
offers a unique opportunity to establish a strong agreement between
experimental and numerical models through an optimisation-based
FE model updating procedure. A well-calibrated FE model represents
the actual behaviour of the structure and leads to achieving accurate
results in the damage detection step. This paper employs the
improved grey wolf optimiser (IGWO) and weIghted meaN oF
vectOrs (INFO) to minimise 11 objective functions with adjustable
coefficients. The statistical investigations reveal that the IGWO effectively
minimised five out of six objective functions, which were
defined based on the modified total modal assurance criterion
(MTMAC). The rest of the objective functions based on the modal
assurance criterion (MAC), natural frequency vector assurance criterion
(NFVAC), differences in natural frequencies, and a combination of
the MAC and NFVAC could not obtain accurate outcomes for the
model updating problem. The statistical comparison indicates that
the INFO algorithm is unreliable for the FE model updating despite
achieving at least one successful result in ten independent runs. The
INFO algorithm and the IGWO algorithm demonstrate comparable
performance in damage detection. The analysis also shows that the
coefficients of MTMAC, alpha and beta, should be adjusted to 0.65
and 1, respectively, to achieve the most accurate damage detection
result.
|