Particle Filter for temperature compensation of FBG strain sensors for event segmentation

Structural Health Monitoring (SHM) plays a crucial role in assessing the integrity and performance of structures. Fibre-Bragg-Grating (FBG) technology has emerged as a valuable tool for strain measurements in SHM applications. However, FBG strain sensors are susceptible to temperature variations that can confound accurate strain measurements, potentially leading to errors in fatigue damage calculations and event detection. This research addresses the challenge of temperature compensation in FBG-based SHM setups. Instead of the traditional approach of placing temperature sensors near every strain sensor, which can escalate costs and complexity with a growing number of sensors, we propose a filter-based alternative. The proposed approach leverages the slower strain changes induced by temperature effects compared to those caused by external loads, e.g. a passing train. The key innovation lies in a filtering method that isolates temperature-induced strain variations as filtered data while treating load-induced strain as process noise. We explore the application of state-based filters such as the Kalman Filter (KF) and Particle Filter (PF). While the KF is limited to normally distributed noise, which does not represent load-induced strain effectively, the PF can handle various, adjustable noise distributions, that can be tailored to specific requirements. The research showcases the PF's ability for temperature compensation on multiple structures, such as a pedestrian bridge equipped with only 8 temperature sensors for 82 strain sensors loaded either in tension or compression. This novel approach enables more reliable strain measurements of events for detection and analysis in SHM systems.