Feature Extraction for Robust Crack Monitoring Using Passive Wireless RFID Antenna Sensors

Abstract

Radio-frequency identification (RFID) tag antenna-based passive wireless sensors are receiving increasing attentions for structural health monitoring in large-scale infrastructure. For permanently installed monitoring, robustness to measurement variation including environmental conditions is a practical issue. This paper retrospects the communication principle of magnetic resonant coupling for low-frequency passive wireless RFID antenna sensors. The influence of communication is uncovered and essentially separated from sensing through two steps: sweep (source) frequency to capture the system behavior, i.e., a resonance frequency range, and multiple feature extraction, fusion, and selection in conjunction with normalization for the selected time-domain feature of peak-to-peak. By this way, the robustness of the low-cost RFID sensing system is enhanced. The proposed method is validated by the case study in open crack detection and characterization under varied measurement conditions.