Design of hetero-core smart fiber optic macrobend sensors

This paper presents a design technique for smart hetero-core fiber optic macrobend displacement sensors using Artificial Neural Networks. Experimental results for several macrobend displacement sensors are used for measuring the ability of Artificial Neural Networks (ANNs) to predict sensor measurand. Multilayer Perceptron algorithm was used as an ANN model. It is shown that all the algorithms are able to predict the displacement values with acceptable errors. Furthermore, a smart sensor architecture composed of multiple macrobend sensors using ANNs is proposed.