Improving Temperature Resolution of Distributed Temperature Sensor Using Artificial Neural Network

Temperature resolution is a key factor for the performance of a Distributed Temperature Sensor (DTS). One can define the resolution as the degree of uncertainty in the temperature information. Thus, the temperature measured in a steady-state condition at a given point in the fiber will vary between successive measurements and between adjacent points that are at the same temperature. Temperature resolution of the system becomes worse as return signal level decreases, as in the case of measurements in longer fibers or as a result of increased loss due to bends or connectors. Besides, recent studies show temperature resolution becomes worse for high measurement temperatures. In this context, this paper discusses the use of an Artificial Neural Network (ANN) algorithm to improve the temperature resolution in a DTS by correctly reconstructing hot regions in the fiber without new extra information of the system, such as: impulsive response, attenuation of the signal of interest, local losses due to fiber curvatures and connectors. Therefore, the use of ANN has a strong application in the calibration of DTS systems.