Measurement accuracy enhancement with multi-event detection using the deep learning approach in Raman distributed temperature sensors

Abstract

In this work, we present a novel deep learning framework for multi-event detection with enhanced measurement accuracy from the measured data of a Raman Optical Time Domain Reflectometer (Raman-OTDR). We demonstrate the utility of a deep learning-based approach by comparing the results from three popular neural networks, i.e. vanilla recurrent neural network (RNN), long short-term memory (LSTM), and gated recurrent unit (GRU). Before feeding the experimentally obtained data to the neural network, we sanitize our data through a correlation filtering operation to suppress outlier noise spikes. Based on experiments with Raman-OTDR traces consisting of single temperature event, we show that the GRU is able to provide better performance compared to RNN and LSTM models. Specifically, a bidirectional-GRU (bi-GRU) architecture is found to outperform other architectures owing to its use of data from both previous as well as later time steps. Although this feature is similar to that used recently in one dimension convolutional neural network (1D-CNN), the bi-GRU is found to be more effective in providing enhanced measurement accuracy while maintaining good spatial resolution. We also propose and demonstrate a threshold-based algorithm for accurate and fast estimation of multiple events. We demonstrate a 4x improvement in the spatial resolution compared to post-processing using conventional total variational denoising (TVD) filters, while the temperature accuracy is maintained within ± 0.5 o C of the set temperature.

Multichannel Classifier for Recognizing Acoustic Impacts Recorded with a phi-OTDR

“…An autoencoder learns to compress the data, minimizing reconstruction errors. Similar studies were carried out by Abdelli K. et al [ 24 , 25 ]. In their work, an autoencoder with the most suitable architecture ( Figure 6 ) was trained to solve the problem of noise reduction and artifact removal.…”

Physics and applications of Raman distributed optical fiber sensing

“…It sanitizes the data using correlation filtering to remove any undesirable noise spikes before feeding it to a deep learning network to achieve multi-event detection. The experiment obtained a temperature accuracy of 0.5 °C at a sensing distance of 11.5 km 100 . Since, the algorithm denoising technology does not increase the hardware cost of the sensor, the temperature demodulation scheme based on the algorithm has attracted extensive attention from researchers in recent years.…”

Automatic detection of the height of poured concrete via fiber-optic temperature sensing

“…The distributed temperature sensing technology mainly includes temperature sensing based on Brillouin scattering and temperature sensing based on Raman scattering. The former has a sensing distance of 100 kilometers and the temperature measurement accuracy can reach 0.1℃ [9], but it is affected by the cross sensitivity of temperature and strain, therefore, decoupling and compensation are required in the actual use process, and the use is relatively cumbersome; The latter is based on the principle of Raman scattering, without external interference such as strain and vibration, and is only related to the change of external temperature, so it is used in engineering more mature [10][11].To sum up, the current applications based on distributed optical fiber sensing are mainly the strain stress and state monitoring of structures, as well as the temperature monitoring of environmental perception. There is no report on RDTS used to measure the liquid level elevation of diaphragm wall concrete pouring.…”