TIME-DOMAIN STRUCTURAL DAMAGE IDENTIFICATION:FROM A DICTIONARY LEARNING PERSPECTIVE

摘要

Structures inevitably deteriorate during their service lives.To accurately evaluate their structural condition,the methods capable of identifying and assessing damage in a structure timely and accurately have drawn increasing attention.Compared to widely-used frequency-domain methods,the processing of time-domain data is more efficient,but remains difficult since it is usually hard to discern signals from different conditions.In fact,the signal processing fields have observed the evolution of techniques,from such traditional fixed transforms as Fourier,to dictionary learning(DL).DL leads to better representation and hence can provide improved results in many practical applications.In this paper,an innovative time-domain damage identification algorithm is proposed from a DL perspective,using D-KSVD algorithm.The numerical simulated soil-pipe system is used for verifying the performance of the proposed method.The results demonstrate that this damage identification scheme is a promising tool for structural health monitoring.