DocumentCode :
3289361
Title :
Fluorescence optical fiber measurement system based on an advanced Neural Network (NN) ensemble algorithm
Author :
Wang Yutian ; Yingchao, Wang ; Shiying, Yang
Author_Institution :
Hebei Key Lab. of Meas. Technol. & Instrum., Yanshan Univ., Qinhuangdao, China
fYear :
2011
fDate :
15-17 April 2011
Firstpage :
415
Lastpage :
418
Abstract :
Noisy Spread based Ensemble Neural Networks is a new neural network ensemble algorithm with high integration, anti-interference and high precision. In this paper, it´s applied to fluorescence spectra of the oil in water (classification and identification). The method uses optical spectroscopy to measure excitation fluorescence spectroscopy in the pool of crude oil and mineral oil in the range 230 ~ 400 nm, belong to UV spectral region: spectral interval of 5 nm, recording fluorescence spectroscopy data ,then the integration neural networks were established by the enhanced T individual neural networks. Vote method for the classification of fluorescence spectra characteristics and the average method for nonlinear regression. Experiments show that the neural network integrated for fluorescence spectral measurement system of oil in water identification and processing higher accuracy.
Keywords :
computerised instrumentation; fluorescence spectroscopy; fuel processing; neural nets; regression analysis; UV spectral region; advanced neural network ensemble algorithm; crude oil; fluorescence optical fiber measurement system; fluorescence spectra characteristics; fluorescence spectral measurement system; fluorescence spectroscopy; mineral oil; neural networks integration; noisy spread based ensemble neural network; nonlinear regression; oil water identification; optical spectroscopy; Accuracy; Algorithm design and analysis; Artificial neural networks; Fluorescence; Noise; Pollution measurement; Signal processing algorithms; Ensemble Neural Networks; Fluorescence Spectrum; Nonlinear Regression; Oil; Spectral Anylisis;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Electric Information and Control Engineering (ICEICE), 2011 International Conference on
Conference_Location :
Wuhan
Print_ISBN :
978-1-4244-8036-4
Type :
conf
DOI :
10.1109/ICEICE.2011.5778096
Filename :
5778096
Link To Document :
بازگشت