A novel time domain structural damage diagnosis method using jensen-shannon divergence

Structural vibration response under ambient excitation is a stochastic process. The change in the characteristics of stochastic process can be used to evaluate the state of structure. Jensen-Shannon Divergence (JSD) is a mathematical expression widely used in time series analysis to account for the similarity of measurements. The aim of this article is to extend the JSD index, in conjunction with the Natural Excitation Technique, phase space reconstruction and singular value decomposition, for detecting damage in structures at an early stage. A simulation on the ASCE Benchmark structure with various damage scenarios and a laboratory experiment for a truss-type structure are carried out to validate the proposed JSD damage index. Both numerical and experimental results show that the proposed approach has better performance in contrast to other existing similar damage index such as Cross-entropy, and no mathematical model of the analyzed structure is required a priori.