Estimation of strain of distorted FBG sensor spectra using a fixed FBGfilter circuit and an artificial neural network

Fibre Bragg Grating (FBG) sensors are extremely sensitive to changes of strain, and are therefore an extremely useful candidate for Structural Health Monitoring (SHM) systems of composite structures. Sensitivity of FBGs to strain gradients originating from damage was observed as an indicator of initiation and propagation of damage in composite structures. To date there have been numerous research works done on distorted FBG spectra due to damage accumulation under controlled environments. Unfortunately, a number of related unresolved problems remain in FBG-based SHM systems development, making the present SHM systems unsuitable for real life applications. This paper reveals a novel configuration of FBG sensors to acquire strain reading and an integrated statistical approach to analyse data in real time. The proposed configuration has proven its capability to overcome practical constraints and the engineering challenges associated with FBG-based SHM systems. A fixed filter decoding system and an integrated artificial neural network algorithm for extracting strain from embedded FBG sensor were proposed and experimentally proved. Furthermore, the laboratory level experimental data was used to verify the accuracy of the system and it was found that the error levels were less than 0.3% in strain predictions.