Premium
Synergistic and combinatorial optimization of finite element models for monitored buildings
Author(s) -
Miraglia Gaetano,
Lenticchia Erica,
Ceravolo Rosario,
Betti Raimondo
Publication year - 2019
Publication title -
structural control and health monitoring
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.587
H-Index - 62
eISSN - 1545-2263
pISSN - 1545-2255
DOI - 10.1002/stc.2403
Subject(s) - finite element method , particle swarm optimization , simulated annealing , calibration , ranking (information retrieval) , mathematical optimization , computer science , structural health monitoring , genetic algorithm , rank (graph theory) , global optimization , engineering , structural engineering , algorithm , mathematics , machine learning , statistics , combinatorics
Summary This paper investigates the use of rank aggregation strategies for the finite element model calibration of monitored masonry structures subjected to earthquakes. Ranking is used to obtain optimal results from several competing optimization strategies, with the final aim of establishing a numerical model of reference to support the existing monitoring systems installed on the structures. For the tuning of the model, different optimization methods are currently employed (i.e., genetic optimization, particle swarm optimization, and simulated annealing optimization), which can provide an initial nonunique definition of the structural parameters. Starting from the results obtained from selected updating methods, a combinatorial parameter selection is proposed to define the best finite element model among several optimal results. In this case, the ranking problem is solved by using a Plackett‐Luce model‐based strategy. The calibrated model is a useful tool to investigate the dynamic response of the structure, allowing a preliminary structural assessment at the same time. The data recorded by the permanent dynamic structural health monitoring system installed on a masonry building, the Town Hall of Pizzoli in central Italy, are used to demonstrate the proposed model calibration strategy.