An optical fibre distributed sensor based on pattern recognition

Machine Learning Methods for Pipeline Surveillance Systems Based on Distributed Acoustic Sensing: A Review

“…These methods are typically based on artificial neural networks [12][13][14][15][16]19,27] or support vector machine [20,45,46] models. In [12], a 3-layer backpropagation and momentum backpropagation ANNs is employed, while in [13] the authors use a momentum backpropagation ANN.…”

“…In [12], a 3-layer backpropagation and momentum backpropagation ANNs is employed, while in [13] the authors use a momentum backpropagation ANN. The authors of [14][15][16] employed a 3-layer feed forward ANN for pattern classification.…”

“…Next, we describe the main proposals in the literature. In the published works that use time domain features, some of them did not carry out any feature extraction process [12,13], and the raw recorded signal was given to the pattern classification algorithm. The authors of [14] also employed the raw recorded signal, but normalized between −1 and 1 as an input to the pattern classification algorithm.…”

“…The works presented in [12,13] aimed to detect water along the fiber. In [12], the length of the sensed fiber segment is 200 m. The sensor is 250 m far from the beginning of the optical fiber cable and signal recordings were carried out near the sensor.…”

“…Although some works have been presented in this direction [3,[12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27], there is still, many times, a lack of understanding of how a reliable and efficient system can be built, and what is the adequate and rigorous methodology to apply in the system evaluation processes. In addition, to the best of our knowledge, no thorough written report that summarizes the main characteristics of these systems exists in the literature applied to DAS+PRS technology.…”