%0 Conference Proceedings
%4 sid.inpe.br/plutao/2015/12.04.14.15.37
%2 sid.inpe.br/plutao/2015/12.04.14.15.38
%F lattes: 5142426481528206 3 TorresChiwCamp:2015:MuCoAl
%T Multi-particle collision algorithm with Hooke-Jeeves for solving a structural damage detection problem
%D 2015
%A Torres, Reynier Hernandez,
%A Chiwiacosky, Leonardo Dagnino,
%A Campos Velho, Haroldo Fraga de,
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@affiliation Universidade do Vale do Rio dos Sinos (UNISINOS)
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@electronicmailaddress reynier.torres@inpe.br
%@electronicmailaddress ldchiwiacowsky@unisinos.br
%@electronicmailaddress haroldo@lac.inpe.br
%B International Conference on Composite Science and Technology, 10 (ICCST)
%C Lisboa, Portugal
%8 2-4 Sept.
%I Instituto Superior Técnico
%V 1
%P 1-12
%S Proceedings
%K Inverse vibration problem, Damage identifitcation, Metaheuristic algorithm, MPCA: multi-particle collision algorithm, Hooke-Jeeves method.
%X A structural damage detection problem is tackled using a recent metaheuristic algorithm (Multi-Particle Collision Algorithm MPCA) associated with a deterministic approach the Hooke-Jeeves approach. The inverse problem of damage identification is formulated as an optimization problem assuming the displacement time history as experimental data. The objective function is defined by the square difference between the measured displacement and the displacement computed by using the forward model. The proposed hybrid approach consists of the MPCA metaheuristic coupled with the local search method of Hooke-Jeeves. A parallel version of MPCA technique using Message Passing Interface (MPI) is implemented since it reduces the computation time. Three different structures were considered for testing the methodology: damped springmass system, truss, and beam structure models. Finite element method was used for structure modeling with a different number of degrees of freedom (DOF). Experimental data was created in silico (synthetic experimental data). Time-invariant damages were assumed to generate the synthetic displacement data. Experiments with noiseless and noisy data were carried out. Level of noise of two and five percent were considered. Good estimations of damage location and quantification have been achieved.
%@language en
%3 1_torres2.pdf
%U http://www.dem.ist.utl.pt/iccst10/files/ICCST10_Proceedings/pdf/WEB_PAPERS/ICCST10_Upload_195.pdf