A method of substructural identification is presented for the estimati
on of local damage in complex structural systems. For this purpose, an
auto-regressive and moving average with stochastic input (ARMAX) mode
l is derived for a substructure to process the measurement data impair
ed by noises. The sequential prediction error method is used for the e
stimation of unknown parameters related to damages. Using the substruc
tural method, the number of unknown parameters for each identification
can be significantly reduced, hence the convergence and accuracy of e
stimation can be improved. For some substructures, the effect of the i
nput excitation is expressed in terms of the responses at the interfac
es with the main structure, and substructural identification may be ca
rried out without measuring the actual input excitation to the structu
re. Direct and indirect methods for estimation of the element damage i
ndices are also developed for local damage assessments. Example analys
es are carried out for idealized structural models of a multistory bui
lding and a truss bridge. The results indicate that the present method
is effective and efficient for local damage estimation of complex str
uctures. (C) 1997 Elsevier Science Ltd.