Identification and evolutionary response analysis of continuous and discrete structural systems using wavelet transform

Abstract

Performance based design of structures require stringent serviceability criteria to be satisfied based on functional requirements and economic implications. Thus, performance evaluation of structural dynamic systems not only helps to detect any deterioration, alerting the user of subsequent structural failure in advance, but also helps to estimate the remaining service life of the structure. This helps in deciding on the preventive measures those are required to maintain design serviceability criteria during operation in the remaining period of service life. The objective of this research work is, therefore, to investigate the mathematical techniques to evaluate the performance of structural systems via system identification and their behaviour, when subjected to evolutionary stochastic processes. First of all, a wavelet-based algorithm is developed to solve the inverse problem so that the parameters of a dynamical system can be estimated from the measured time history of the response. Although conventionally this problem of system idenfication is performed either in time domain (e.g. using Kalman filter technique) or in frenquency domain (using Fourier transform), both the techniques are known to have their own limitations in the estimation of system parameters. In this context, time-frequency based approach has gained momentum in the recent past. A wavelet-based time-frequency analysis is adopted in the present thesis to estimate the natural frequencies and mode shapes of linear structural systems and their changes. The proposed wavelet-based formulation, which is applicable for both off-line and on-line identification and monitoring of dynamical systems is then validated using numerical simulations and experimental results. The proposed identification technique is further used to identify localized modes of a class of non-linear systems