In this paper, an in-line optical fiber spectral analysis system is presented and analyzed, which will be used in the field for measuring the mount of components of oil, water and gas in down-well. Instead of full spectrum analysis of crude oil with a spectroscopy on the ground, this system uses a disperse-spectrum-selecting algorithm, which makes the system in small and compact size and easy to use in down-well. The correlation of the model of disperse wavelengths is equivalent to the model of full range wavelengths and the system measurement accuracy reaches 2 percent by experiments. In addition, the math's model of the system is established, and several regression methods such as Multivariate linear regression (MLR) and Principle components regression (PCR) are discussed and compared in detail. The performance of the system is analyzed with the math's model also. The result indicates that the PCR has a higher accuracy then MLR and 6 channels are enough for practical system used in the field. Besides, an accuracy of less than 1 percent may be got by the results of performance analysis.

Date Published: 5 March 2007
PDF: 9 pages
Proc. SPIE 6595, Fundamental Problems of Optoelectronics and Microelectronics III, 65953S (5 March 2007); doi: 10.1117/12.726604

Published in SPIE Proceedings Vol. 6595:
Fundamental Problems of Optoelectronics and Microelectronics III
Yuri N. Kulchin; Jinping Ou; Oleg B. Vitrik; Zhi Zhou, Editor(s)