Mechanical and Civil Engineering Seminar
This seminar addresses important problems of sparse output-only identification, sparse input-output identification of nonlinear structures, and structural damage detection. Sparse component analysis based linear modal identification method for underdetermined problems using -minimization is presented. Structural damage localization and damage severity identification by means of sparse representation and -minimization recast as classification problem is presented. A sparse system identification algorithm with -regularization suitable for single degree of freedom (SDOF) and multiple degree of freedom (MDOF) structural systems with either nonlinear elastic or inelastic/hysteretic behavior is presented. This new parametric technique is capable of discovering the underlying governing equations of the system of interest from input-output data. The technique includes basis functions that allow the discovery of significant nonlinearities, and hystertic or inelastic behavior with permanent deformation. Model selection is performed using Akaike information criterion. Experimental validation of the selected model is presented. A new two stage semi-supervised technique for nonlinear structural system identification is presented, which allows for discovery of type of nonlinearity involved.