An online two-stage adaptive algorithm for strain profile estimation from noisy and abruptly changing BOTDR data and application to underground mines

Strain measurement using BOTDR (Brillouin Optical Time-Domain Reflectometry) is nowadays a standard tool for structural health monitoring. In this context, weak data quality and noise, usually owed to defective fiber installation, hinders discriminating actual level shifts from outliers and might entail a biased risk assessment. We propose a novel online adaptive algorithm for strain profile estimation in strain time series with abrupt and gradual changes and missing data. It relies on a convolution filter in Brillouin spectrum domain and a smoothing technique in time domain. In simulated data, the convolution filter is shown to reduce strain measurement uncertainty by up to 8 times the strain resolution. The two-stage method is illustrated with systematic outliers removal from real data of a Chilean copper mine and the improvement of the associated gain spectrum quality by up to 18 dB in SNR terms.