Development of Automatic Temperature Compensation Software for Optical Fibre Sensing Data

Development of Automatic Temperature Compensation Software for Optical Fibre Sensing DataM.Res in Photonic Systems DevelopmentMini ProjectMalcolm Scott Supervisor: Prof. Kenichi Soga1Strain Sensing in Civil EngineeringDetermine how a structure moves over timeNew structures movement as soil settles, and as load increasesOld structures measure structural health as surroundings are redevelopedPiles, buildings, tunnels, bridges,

Reproduced from Mohamad (2008)Traditional Strain SensorsVibrating Wire Strain GaugeResonant frequency of taut wire changes with tensionPoint sensor, individually installedOnly a few points per structureMay miss important features in the spaces between sensors

Distributed Optical Fibre SensingStrain profile rather than point dataMeasure along the length of the fibreFibre itself is sensorBrillouin Optical Time-Domain Reflectometry (BOTDR)

StrainDistanceReproduced from Bennett (2006)Optical Scattering

Reproduced from Mohamad (2008)Brillouin Optical Time-Domain Reflectometry (BOTDR)Send a short, high-power laser pulse into sensing fibreWatch for backscattered light within the Brillouin frequency range(Could use forward scattering, but that requires access to both ends of fibre)Very low power!

Time of arrival distancePeak frequency strain(contaminated by temperature)

Temperature Compensation (Our Way)Two fibresOne measures strain and temperatureOne measures temperature only: Unitube gel-filled cablePerform BOTDR on both;can (nearly) just subtract(Take into account differing cableproperties: constant factor)

Case Study: Addenbrookes Access Road Bridge

Case Study: Addenbrookes Access Road Bridge

The DataOne strain profile: 10,000-100,000 points (calculated by equipment)One pile/beam: ~10 strain profiles (strain + temperature, repeated; also both ends where fibre is broken)One days readings: 2 beams + 7 pilesOne project: so far, 7 individual days of readings; more to comePotentially on the order of hundreds of millions of pointsProcessing and analysing this data is time-consumingAnalysis the Hard WayParse dataRemove bogus data from beyond the end of the fibreAlign & average repeat readings (fibre length may change: resplicing)Mark region of interestCorrect for analyser miscalibrationAlign temperature data and perform compensationFilter noise (Savitzky-Golay)Graph individual data setsResample and align successive days data setsCompute and graph changes in strain profile over timeHuge Excel spreadsheet!Analysis the Easy WaySoftware for use by geotechnical researchers and civil engineersOpen source softwareWritten in PythonObject-oriented, modular, model-view-controller basedUsing scientific computing libraries: NumPy, SciPy, MatplotlibEasy to adapt and extendWill be available online: http://strainanalyser.malc.org.uk/

Thank youhttp://strainanalyser.malc.org.uk/