Research On Structural Finite Element Model Updating Method Based On Minimum-rank

The develop ment of economy and techno logy enab le the engineer ingstructures to become longer in span, higher in he ight and more comp lex. It is aninevitable trend to monitor the health of large-scale structures. Structural dama geidentificatio n and structura l finite e leme nt model updating are important parts ofstructural hea lth monitoring. And most structural damage identificatio n methodsare based on structura l finite ele ment model. There fore the sound ness ofstructural finite ele ment model is very important, and it is necessary and pressingto develop advanced finite element model updating techniques.Structura l finite ele ment model updating method observes a rapid growth inrecent years and numerous methods have been developed. But most of theupdating methods tend to disperse the difference between the analytical and testmodels all over the updated models, and hence their results don’t presentlocalized damages. Minimum-rank model updating method is studied in thispaper to try to keep the characteristics of damage localization in the finiteelement models. The main research contents are as follows.Aiming at the localization characteristics of the damage of finite ele mentmodels, the structura l finite ele ment model updating method based on t heminimum-rank is investigated. Model condensation and modal expans ionmethods are used when partial DOFs are measured to expand its application.The concept of the first-le vel matr ix is applied to the calculation of stiffnessconnectivity matr ix. The proposed method is more concise, clear and accurate asno need to perfor m sensitivity ana lys is to solve the stiffness connectivity matr ix.The method for determining the number of structural dama ge units is improved,which in turn improves the model updating results.A plane stee l truss and a simp le beam are used for numerica l study of theproposed method. Results show that the corrected result o f the minimum-rankbased on updated model is very c lose to the pre-set model a nd it highlights thelocality of the dama ge. When partia l DOFs are measured, the method also doeswell, but the updated model has a bit lower precision.