Method for improving identification accuracy of mixture components by using known mixture Raman spectrum

Patent metadata

Field	Value
Publication number	US-11614408-B2
Application number	US-202117344111-A
Country	US
Kind code	B2
Filing date	Jun 10, 2021
Priority date	May 19, 2020
Publication date	Mar 28, 2023
Grant date	Mar 28, 2023

Abstract

A method for improving an identification accuracy of mixture components by using a known mixture Raman spectrum is disclosed. After calculating a first similarity between a to-be-tested Raman spectrum characteristic vector group and a pure substance Raman spectrum characteristic vector group of an nth kind of pure substance in a Raman spectrum standard library, the method uses a known mixture library to calculate to obtain a second similarity between a to-be-identified substance Raman spectrum characteristic vector group and a spectral peak characteristic vector group with offset information corresponding to a pure substance in a known mixture, and determines a similarity between a to-be-tested mixture and the nth kind of pure substance according to the first similarity and all second similarities to thus obtain a component identification result. The present application uses the known mixture library to assist the Raman spectrum standard library in searching.

First claim

What is claimed is:

1. A method for improving an identification accuracy of mixture components by using a known mixture Raman spectrum, the method comprising:
establishing a Raman spectrum standard library and a known mixture library, wherein the Raman spectrum standard library comprises pure substance Raman spectrum characteristic vector groups of N kinds of pure substances; the known mixture library comprises known mixture Raman spectrum characteristic vector groups of M kinds of known mixtures; and each kind of the known mixtures is formed by mixing K kinds of pure substances;
acquiring a to-be-tested Raman spectrum characteristic vector group of a to-be-tested mixture;
calculating a first similarity between the to-be-tested Raman spectrum characteristic vector group and a pure substance Raman spectrum characteristic vector group of an nth kind of pure substance in the Raman spectrum standard library, and detecting whether a reference known mixture exists in the known mixture library, wherein the reference known mixture is a known mixture containing the nth kind of pure substance from among the M kinds of known mixtures; n is a parameter, and a starting value of n is 1;
if no reference known mixture exists in the known mixture library, determining that a similarity between the to-be-tested mixture and the nth kind of pure substance is the first similarity;
if Q kinds of reference known mixtures exist in the known mixture library, extracting a spectral peak characteristic vector group with offset information corresponding to the nth kind of pure substance in the known mixture Raman spectrum characteristic vector group of each kind of reference known mixture, and calculating a second similarity between the spectral peak characteristic vector group with the offset information and the to-be-tested Raman spectrum characteristic vector group; and determining that the similarity between the to-be-tested mixture and the nth kind of pure substance is a maximum value from the first similarity and Q second similarities;
after determining that the similarity between the to-be-tested mixture and the nth kind of pure substance is obtained, setting n=n+1, and re-executing steps of calculating the first similarity between the to-be-tested Raman spectrum characteristic vector group and the pure substance Raman spectrum characteristic vector group of the nth kind of pure substance in the Raman spectrum standard library and detecting whether a reference known mixture exists in the known mixture library, till N similarities between the to-be-tested mixture and the N kinds of pure substances are obtained in case of n=N; and
selecting P kinds of pure substances with the highest similarity as a component identification result of the to-be-tested mixture.

2. The method according to claim 1 , wherein the extracting a spectral peak characteristic vector group with offset information corresponding to the nth kind of pure substance in the known mixture Raman spectrum characteristic vector group of each kind of reference known mixture comprises, for any qth kind of reference known mixture:
determining an sth spectral peak that is closest to an ith spectral peak from among the known mixture Raman spectrum characteristic vector group of the qth kind of reference known mixture, for any ith spectral peak in the pure substance Raman spectrum characteristic vector groups of the nth kind of pure substance, wherein the characteristic vector of the ith spectral peak is expressed as [λ i n , I i n , ω i n ], and the characteristic vector of the sth spectral peak is expressed as [λ s q , I s q , ω s q ], wherein λ denotes a Raman shift of the spectral peak; I denotes a Raman intensity of the spectral peak; and ω denotes a full width at half maximum of the spectral peak;
if the sth spectral peak satisfies d≤h 1 or the sth spectral peak satisfies h 1 <d<h 2 ,S i (ω)≥S ω , determining that the sth spectral peak is an offset spectral peak corresponding to the ith spectral peak, wherein d=|λ i n −λ s q | denotes an absolute value of a difference value of Raman shifts between two spectral peaks; ω=|ω i n −ω s q | denotes an absolute value of a difference value of full widths at half maximum between two spectral peaks; S i (ω) is a full width at half maximum similarity calculated using a fuzzy membership function based on the absolute value of the difference value of the full widths at half maximum; and h 1 , h 2 , and S ω are all preset thresholds; and
calculating to obtain the offset spectral peaks corresponding to all spectral peaks in the pure substance Raman spectrum characteristic vector group of the nth kind of pure substance, and to obtain the spectral peak characteristic vector group with offset information of the nth kind of pure substance.

3. The method according to claim 1 , wherein the calculating the first similarity between the to-be-tested Raman spectrum characteristic vector group and the pure substance Raman spectrum characteristic vector group of the nth kind of pure substance in the Raman spectrum standard library comprises:
determining a kth spectral peak that is closest to a jth spectral peak in the to-be-tested Raman spectrum characteristic vector group for any jth spectral peak in the pure substance Raman spectrum characteristic vector group of the nth kind of pure substance, wherein the characteristic vector of the jth spectral peak is expressed as [λ j n , I j n , ω j n ], and the characteristic vector of the kth spectral peak is expressed as [λ k T , I k T , ω k T ], wherein λ denotes a Raman shift of the spectral peak; I denotes a Raman intensity of the spectral peak; and ω denotes a full width at half maximum of the spectral peak;
calculating an absolute value of a difference value of Raman shifts and an absolute value of a difference value of full widths at half maximum between the jth spectral peak and the kth spectral peak, and using a fuzzy membership function to perform a calculation based on the absolute value of the difference value between the Raman shifts to obtain a Raman shift similarity and to perform a calculation based on the absolute value of the difference value between the full widths at half maximum to obtain a full width at half maximum similarity;
calculating, based on the Raman shift similarity and the full width at half maximum similarity, to obtain the similarity

S
j

=

S
j

⁡

(
λ
)

+

S
j

⁡

(
ω
)

2

between the jth spectral peak and the kth spectral peak, wherein S j (λ) is the Raman shift similarity, and S j (ω) is the full width at half maximum similarity;
calculating to obtain similarities between all spectral peaks in the pure substance Raman spectrum characteristic vector group of the nth kind of pure substance and the closest spectral peaks in the to-be-tested Raman spectrum characteristic vector group, and obtaining the first similarity by calculation according to the following formula:

S
n

Assignees

• Univ Jiangnan

Inventors

• Zhu Qibing
• Ji Mingqiang
• Huang Min
• Guo Ya

Classifications

• G16C20/20

  Identification of molecular entities, parts thereof or of chemical compositions · CPC title

• G06F2218/10

  by analysing the shape of a waveform, e.g. extracting parameters relating to peaks · CPC title

• G01N2201/1293

  resolving multicomponent spectra · CPC title

• G01N21/65 Primary

  Raman scattering · CPC title

• G01J3/44

  Raman spectrometry; Scattering spectrometry {; Fluorescence spectrometry} · CPC title