Online Sifting Technique for Structural Health Monitoring Data Based on Recursive EMD Processing Framework

Abstract

Real-time and online screening techniques for single load effect signal monitoring are one of the key issues in smart structure monitoring. In this paper, an online signal sifting framework called online recursive empirical mode decomposition (EMD) is proposed. The framework is based on an improved EMD that optimizes the boundary effect by using extreme value recursion and eigensystem realization algorithm (ERA) extension, and combines the intrinsic mode functions (IMFs) correlation coefficient and adaptive filtering to select IMFs for signal reconstruction to achieve the sifting purpose. When applied to simulated signals, the method satisfies the requirements of signal sifting in an online environment with high adaptivity, low parameter sensitivity and good robustness. The method was applied to the dynamic strain data collected by the health monitoring system of Daishan Second Bridge to achieve real-time online sifting of strain signals caused by traffic loads, which provided the basis for subsequent data analysis applications and confirmed the value of the application in a real bridge health monitoring system.