Simultaneous identification of structural parameters and dynamic input with incomplete output-only measurements Hao Sun * ,† and Raimondo Betti Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, NY 10027, USA SUMMARY A hybrid heuristic optimization strategy is presented to simultaneously identify structural parameters and, when possible, dynamic input time histories from incomplete sets of output measurements. The proposed strategy combines a novel swarm intelligence algorithm, the artificial bee colony algorithm, with a local search operator, Nelder–Mead simplex method, integrated in a search space reduction approach, so as to improve the convergence efficiency of the overall identification process. Because of the independent nature of the algorithm, a parallel scheme is implemented so as to improve the computational efficiency. If the time histories of the structural response and information about the mass of the structural system are available, then the algorithm can also be used for the identification of the time histories of the dynamic input force through a modified Newmark integration scheme, using the current estimates of the structural parameters. To investigate the applicability of the proposed technique, three numerical examples, two shear-type building models and a coupled building system model under different conditions of data availabilities and noise corruption levels are presented. The results show that the proposed technique is powerful, robust and efficient in the simultaneous identification of the structural parameters and input force even from an incomplete set of noise-contaminated structural response measurements. Copyright © 2013 John Wiley & Sons, Ltd. Received 16 March 2013; Revised 16 July 2013; Accepted 1 September 2013 KEY WORDS: output-only system identification; artificial bee colony algorithm; Nelder–Mead simplex method; search space reduction; parallel computation 1. INTRODUCTION Identification of the dynamic characteristics and structural parameters of models representing complex structural systems plays a key role in SHM for model updating, damage detection, active control, non- destructive evaluation and others. The system identification (SI) process, formulated as an inverse problem, aims to determine a set of parameters, either physical or non-physical, of a model that is representative of the structure in question. Physical parameters might be considered as the mass, damping and stiffness of the structural elements while the coefficients of an autoregressive model can be labeled as non-physical parameters. These estimated parameters can then be used, among other quantities, to predict the structural response to a future excitation or to assess the structural conditions. In essence, SI can be considered as an optimization process in which the objective is to identify a model of a system so that its predicted response to a given input is close enough to the measured response from the real system. In recent years, considerable efforts have been carried out in developing reliable models of structural systems using time histories of the input and/or of the corresponding structural response as shown in [1–8]. *Correspondence to: Hao Sun, Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, NY 10027, USA. † E-mail: hs2595@columbia.edu STRUCTURAL CONTROL AND HEALTH MONITORING Struct. Control Health Monit. 2014; 21:868–889 Published online 3 October 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/stc.1619 Copyright © 2013 John Wiley & Sons, Ltd.