Novel technique to improve optical fibre sensor's response for industrial applications

A general major disadvantage of optical fibre sensors (OFS) is a non-linear (sigmoidal) response which produces a limited useable linear region. This work presents a novel method of overcoming this problem by the use digital instrumentation and signal processing. Previously we have shown that by appropriate selection of eight wave ordinates from the spectral response of an optical fibre pH sensor as inputs to a trained multilayer artificial neural network, the response range of the sensor can be extended from the normal pH 5.0-7.25 range to the full sensitivity of the sensor, i.e. pH 2.51-9.76 range with a worst error of only 0.08 pH. Application of this method is illustrated. The method is currently adopted to a new, simple and portable instrument based on the use of three LEDs as light sources. In this preliminary study, the sensor just simply measures colour changes of thymol blue indicator in an acid-base pH titration. The application of the neural network to process the colour changes of thymol blue indicator in the acid-base titration is illustrated.