Interpretation of fluorescence fingerprints of crude oils and other hydrocarbon mixtures using neural networks

• US 5,424,959 A
• Filed: 07/19/1993
• Issued: 06/13/1995
• Est. Priority Date: 07/19/1993
• Status: Expired due to Fees

• Current Assignee: Texaco Incorporated (Chevron Corp.)
• Original Assignee: Texaco Incorporated (Chevron Corp.)
• Inventors: Pyeatt, Larry D., Reyes, Marilyn V.
• Primary Examiner(s): Dzierzynski, Paul M.
• Assistant Examiner(s): Bruce, David Vernon
• Application Number:
  US08/093,567
• Time in Patent Office: 694 Days
• Field of Search: 364/497, 364/498, 399/911, 399/928, 399/929, 378/53, 378/45, 250/372
• US Class Current: 702/28
• CPC Class Codes: G01N 2021/6417 Spectrofluorimetric devices
• CPC Class Codes: G01N 21/64 Fluorescence; Phosphorescence
• CPC Class Codes: G01N 2201/1293 resolving multicomponent spectra
• CPC Class Codes: G01N 33/2829 mixtures of fuels, e.g. determining the RON-number
• CPC Class Codes: Y10S 706/929 Geological, e.g. seismology

Abstract

An artificial intelligence system is used with a conglomeration of fluorescence data to provide a method of improving recognition of an unknown from its spectral pattern. Customized neural network systems allow the ultimate organization and resourceful use of assumption-free variables already existing in a total scanning fluorescence database for a much more comprehensive, discrete and accurate differentiation and matching of spectra than is possible with human memory. The invention provides increased speed of fingerprinting analysis, accuracy and reliability together with a decreased learning curve and heightened objectivity for the analysis.

Claims

1. A system for analyzing and interpreting total scanning fluorescence fingerprints, which are characteristic of complex hydrocarbon fluid mixtures, in order to make a better estimation of composition of the mixture, said system comprising:

at least one fluorescence detector;

spectrum preprocessor means connected to receive data from each said at least one fluorescence detector;

data base manager means connected to said preprocessor means; and

a plurality of artificial intelligence modules each capable of performing a specific task, all of said modules being interconnected with said data base manager whereby pattern recognition ability of said artificial intelligence modules is utilized by training to identify relationships between fluorescence fingerprints of an unknown fluid sample and known fluorescence fingerprints to estimate the content of fluid being monitored.

2. A system according to claim 1 wherein one of said plurality of artificial intelligence modules subtracts fluorescence data of one sample from the fluorescence data of another sample to eliminate background noise and create a sample data file.

3. A system according to claim 1 wherein one of said plurality of artificial intelligence modules compares the fluorescence fingerprint of the unknown fluid mixture with the fingerprint of known fluid mixtures to predict the content of the unknown mixture.

4. A system according to claim 3 wherein one of said plurality of artificial intelligence modules factors in possible weathering of said unknown mixture.

5. A method using neural networks in analyzing and interpreting fluorescence fingerprints characterizing complex hydrocarbon fluid mixtures comprising the steps of:

providing at least one fluorescence detector;

providing a neural network connected to each said at least one fluorescence detector, said neural network having at least spectrum preprocessor means, data base manager means connected to said preprocessor means, and a plurality of artificial intelligence modules connected to said data base manager, each said module being capable of performing a specific task;

collecting fluorescence data from a fluid sample by said at least one fluorescence detector and feeding said sample data to said neural network;

comparing said fluorescence sample data to corresponding known standard fluorescence data;

subtracting fluorescence data of one sample from the fluorescence data of another sample to eliminate background noise and create a sample data file;

standardizing each sample data file to a common format;

using this standardized data to "train" the neural network to recognize corresponding characterizing fingerprint data which are likely to be encountered in the study area;

testing the trained network with known fluorescence fingerprint data; and

applying to the network fluorescence data from unknown fluid mixtures, whereby comparison with known data is used to make an accurate determination of the content of the unknown fluid mixture.