Improving System Identification using Clustering

System identification involves identification of a behavioral model that best explains the measured behavior of a structure. This research uses a strategy of generation and iterative filtering of multiple candidate models for system identification. The task of model filtering is supported by measurement cycles. During each measurement cycle, the location for subsequent measurement can be chosen using the predictions of current candidate models. In this paper, data mining techniques are proposed to support such measurement-interpretation cycles. Candidate models, representing possible states of a structure, are clustered using a technique that combines principal component analysis and K-means clustering. Representative models of each cluster are used to place sensors for subsequent measurement on the basis of the entropy of their predictions. Models are filtered from candidate model sets using new measurements. Results show that clustering is necessary to identify the different groups of candidate models. The entropy of predictions is found to be a valid stopping criterion for iterative sensor addition. While measurement-interpretation cycles can lead to a unique model for structures with low levels of complexity, engineers may be left with large numbers of models for structures with higher levels of uncertainty. In those situations, clustering is a powerful tool to classify models and thus provide much fewer representative models to engineers for further decision making. Grad. Res. Assist. in Comp. Sc., IMAC, Struct. Eng. Inst., Station 18, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. E-Mail: [email protected]. Post Doc. Res. in Civil Eng., IMAC, Struct. Eng. Inst., Station 18, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. E-Mail: [email protected]. Assist. Prof. of Civil Eng., Department of Building, National University of Singapore, 117566, Singapore. E-Mail: [email protected]. Prof. of Civil Eng., F. ASCE, IMAC, Struct. Eng. Inst., Station 18, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. E-Mail: [email protected]. 1 Saitta S., Kripakaran, P., Raphael, B. and Smith, I.F.C. "Improving System Identification Using Clustering" Journal of Computing in Civil Engineering, Vo1 22, No 5, 2008, pp 292-302. BEST PAPER AWARD for 2008. -------------------------------------------------------------------------------------------------------------------------------------------------