Filtern
Dokumenttyp
- Beitrag zu einem Tagungsband (2)
- Vortrag (2)
- Dissertation (1)
Sprache
- Englisch (5)
Schlagworte
- Distributed Brillouin sensing (5) (entfernen)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (2)
This thesis reports on the development of a system for distributed strain measurement in silica optical fibers. The system was developed to provide a solution for monitoring the structural health of river embankments with a measurement length > 5 km and a spatial resolution < 5 m. It is based on stimulated Brillouin scattering (SBS), a nonlinear optical effect which converts the mechanical strain of an optical fiber into a frequency shift of the backscattered light of an optical signal. The measurement technique that is employed and significantly advanced within this work is the Brillouin optical frequency domain analysis (BOFDA). Prior to this work, this technique had been presented as a laboratory setup providing its proof of concept; however, at this development stage, the technique had been limited in performance, theoretical foundation, integrability and robustness for real-life applications when compared to the state-of-the-art Brillouin optical time domain analysis (BOTDA). The thesis comprises the theoretical background of the measurement system, advancements regarding its implementation into a practically applicable device – including the proposal of techniques for performance enhancements in signal processing – as well as the evaluation of the system performance in experimental studies. First, the physical nature of SBS in optical fibers is analyzed. Here, special focus is set on the frequency domain properties of the interaction between the optical signals, which provides a deep understanding of the system behaviour in the special case of distributed measurements in the frequency domain. The BOFDA technique is then presented with a thorough analysis from a system point of view and considerations on its implementation in a practical setup. The laboratory setup is presented with all components and different aspects of advancement in accuracy and efficiency over the state of development prior to this work, along with representative measurement results. As the major advancement regarding the system implementation, a digital approach to frequency domain measurements is presented, with a system description, a demonstrator setup and measurement results. A detailed analysis of the physical occurrences within the system that lead to a limitation of its spatial resolution is given, founding on the description of SBS earlier in the thesis. It comprises a novel point of view on the measurement artifacts that degrade frequency domain measurements of SBS; a connection to corresponding studies that apply to the BOTDA technique is made. From here, a novel approach to overcome the limitations by means of signal correction in post-processing is presented. Finally, the application of the measurement system in dike monitoring is presented. A method for integrating optical fibers into geotextiles is introduced, together with considerations on the coating material and handling on construction sites. By presenting several experimental tests in the laboratory and the field, the feasibility of the system for monitoring of the structural health of river embankments is confirmed. With the advancements achieved within this work, the BOFDA technique meets the specifications in accuracy and resolution of state-of-the-art BOTDA devices, while offering new perspectives in terms of dynamic range, robustness and cost efficiency.
In the last few years, the use of machine learning has emerged in the field of distributed fiber optic sensors as a promising approach to enhance their performance and provide new capabilities. In this study, we use machine learning for simultaneous measurements of temperature and humidity in polyimide (PI)-coated optical fibers based on Brillouin Brillouin optical frequency domain analysis (BOFDA). Different non-linear machine learning algorithms are employed, namely polynomial regression, decision trees and artificial neural networks (ANNs), and their discrimination performance is benchmarked against that of the conventional linear regression. The performance is evaluated using leave-one-out cross-validation to ensure that the models are reliable and able to generalize well on new data. We show that nonlinear machine learning algorithms outperform the conventional linear regression and thus could pave the way towards simultaneous cost-effective temperature and humidity distributed sensing, which has the potential to find attractive new applications in the field of civil and geotechnical engineering, from structural health monitoring of dikes and bridges to subsea cables and long pipelines corrosion detection.
In the last few years, the use of machine learning has emerged in the field of distributed fiber optic sensors as a promising approach to enhance their performance and provide new capabilities. In this study, we use machine learning for simultaneous measurements of temperature and humidity in polyimide (PI)-coated optical fibers based on Brillouin Brillouin optical frequency domain analysis (BOFDA). Different non-linear machine learning algorithms are employed, namely polynomial regression, decision trees and artificial neural networks (ANNs), and their discrimination performance is benchmarked against that of the conventional linear regression. The performance is evaluated using leave-one-out cross-validation to ensure that the models are reliable and able to generalize well on new data. We show that nonlinear machine learning algorithms outperform the conventional linear regression and thus could pave the way towards simultaneous cost-effective temperature and humidity distributed sensing, which has the potential to find attractive new applications in the field of civil and geotechnical engineering, from structural health monitoring of dikes and bridges to subsea cables and long pipelines corrosion detection.