Multi-component floating algae concentration measurement algorithm based on layering and partitioning of fluorescence spectrum

Abstract

The invention discloses a multi-component floating algae concentration measurement algorithm based on layering and partitioning of a fluorescence spectrum. The invention belongs to the technical field of marine water ecological environment monitoring. According to the algorithm, an unknown to-be-analyzed aliasing three-dimensional fluorescence spectrum is analyzed according to the following layered and partitioned analysis thought. In the first layer, the whole spectrum # imgabs0 # participates in analysis, so that the blue algae gate concentration and the cryptophyta concentration are analyzed, and a difference spectrum # imgabs1 # in a full spectrum region is obtained; and a second layer: partition analysis: taking a region # imgabs3 # containing effective spectral characteristics in # imgabs2 # to participate in analysis, so as to analyze the chlorophyll a concentration of chlorophyta, diatom phyta, dinoflagellum and xanthophyta. The algorithm is obviously superior to a common analytical algorithm in the aspect of reducing misrecognition, especially in processing high-similarity living body fluorescence spectrums such as diatom phyta, dinoflagellum and xanthophyta.

Description

Technical Field

[0001] The invention belongs to the technical field of marine water ecological environment monitoring, and specifically relates to a multi-component phytoplankton concentration measurement algorithm based on fluorescence spectrum stratification and partitioning. Background Art

[0002] Algae is a general term for a class of microscopic algae that live in the water in a planktonic manner. Algae are widely distributed and are the material and energy basis of aquatic ecosystems. Their species composition, community structure and abundance changes directly affect water quality, ecosystem energy flow, material flow and biological resource changes.

[0003] The in vivo fluorescence method has the advantages of fast measurement speed, high sensitivity, and no need for sample pretreatment. It is currently the most promising method for rapid on-site measurement of phytoplankton concentration in large areas of water. The basic idea of ​​the in vivo fluorescence method is to invert the chlorophyll a concentration of algae through the fluorescence intensity generated by stimulated radiation of pigment molecules in living algae cells. Different types of living algae have their own characteristic fingerprint fluorescence spectra. When there are multiple components in the sample, within the range of Lambert-Beer's law, the algae fluorescence spectrum satisfies linear additivity, so the quantitative problem of algae concentration can be transformed into a linear regression problem of the matrix. This is the principle basis for the rapid classification and quantitative measurement of phytoplankton through in vivo fluorescence spectroscopy.

[0004] As a byproduct of the photosynthesis process of phytoplankton, the yield of living fluorescence is closely related to the efficiency of light energy absorption, transmission and release during the photosynthesis of living cells. The instability of living fluorescence of phytoplankton has an adverse effect on the accurate quantification of the concentration of living fluorescent chlorophyll a of phytoplankton. By constructing an interpolation standard spectral library of phytoplankton, the classification accuracy and measurement precision of the measurement results of multivariate regression analysis can be significantly improved. However, it also brings new problems: the standard spectral library constructed by the interpolation method has a certain distribution range, resulting in a large overlap range of standard spectral libraries of phytoplankton of multiple categories, especially phytoplankton with highly similar fluorescence spectral characteristics, and the similarity is too high, which leads to difficulties in analysis. Summary of the invention

[0005] In order to solve the technical problem mentioned in the background technology that the standard spectral library of phytoplankton with highly similar fluorescence spectral features has a large overlap range and is too similar, which leads to difficulty in analysis, the present invention proposes a multi-component phytoplankton concentration measurement algorithm based on fluorescence spectral stratification and partitioning.

[0006] The technical solution of the present invention is as follows: A multi-component phytoplankton concentration measurement algorithm based on fluorescence spectrum stratification and partitioning includes the following steps: Step 1: The unknown algae sample is mixed with the in vivo three-dimensional fluorescence spectrum, and a non-negative least squares method multivariate linear regression analysis is performed on the cyanobacteria and cryptophytes in the full spectrum: , in: It represents the three-dimensional fluorescence spectrum of living phytoplankton within the excitation wavelength range of 370 nm to 650 nm and the emission wavelength range of 600 nm to 720 nm; It represents the average spectrum of the standardized spectral library of phytoplankton of other phyla except Cyanobacteria and Cryptophyta; They represent the standard spectrum combination with the smallest error obtained by traversing all the combination modes in the standard spectrum library of Cyanobacteria and Cryptophyta when the standard spectrum library of phytoplankton of other phyla except Cyanobacteria and Cryptophyta is averaged; represent the least squares analytical concentrations of phytoplankton of Cyanobacteria and Cryptophyta, respectively; represents the minimum iteration error; Step 2: Obtain the difference fluorescence spectrum after deducting Cyanobacteria and Cryptophyta from the whole spectrum : , Step 3: Select a new characteristic fluorescence spectrum range within the difference fluorescence spectrum range , The fluorescence characteristics of most of the Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta were covered in the experiment, and the excitation wavelength ranged from 580 nm to 620 nm, and the emission wavelength ranged from 660 nm to 700 nm. Step 4: Within the same range as step 3, intercept the standard spectra of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta to form a new standard spectral library ; Step 5: Within the same range as step 3, perform non-negative least squares multivariate linear regression analysis on Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta: , in: They represent the standard spectrum combination with the smallest error obtained by traversing all the combination modes in the standard spectrum library of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta within the same range of step 3; represent the least squares analytical concentrations of Chlorophyta, Bacillariophyta, Dinophyta, and Xanthophyta, respectively; represents the minimum iteration error.

[0007] Step 6: Output the chlorophyll a concentration of different types of phytoplankton, namely: .

[0008] Beneficial effects: The present invention discloses a multi-component phytoplankton concentration measurement algorithm based on fluorescence spectrum stratification and partitioning. During analysis, the unknown aliased three-dimensional fluorescence spectrum to be analyzed is analyzed according to the following stratification and partitioning analysis ideas. First layer: full spectrum Participate in the analysis, the purpose is to analyze the concentration of Cyanobacteria and Cryptophyta, and obtain the difference spectrum in the full spectrum area ; Second layer: partition analysis, take The area containing effective spectral features The purpose of the algorithm is to analyze the chlorophyll a concentration of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta. This algorithm performs significantly better than common analysis algorithms in reducing misidentification, especially when dealing with fluorescent spectra of living organisms with high similarity, such as Bacillariophyta, Dinophyta and Xanthophyta. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 Schematic diagram of the overlap range of the standardized spectral libraries for phytoplankton (especially diatoms and dinoflagellates).

[0010] Figure 2 Schematic diagram of the similarity of living three-dimensional fluorescence spectra of phytoplankton (especially green algae, diatoms, yellow algae and dinoflagellates).

[0011] Figure 3 Schematic diagram of layered and partitioned analysis of aliased spectra.

[0012] Figure 4 Schematic diagram of the comparison of average relative errors of different spectral analysis methods for measuring algae samples in the test set. DETAILED DESCRIPTION

[0013] The specific implementation mode of the present invention is described in detail below. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation mode and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiments.

[0014] Example 1 like Figure 1 As shown in the figure, the standard spectral library constructed by the interpolation method has a certain distribution range, which leads to a large overlap range of standard spectral libraries of multiple types of phytoplankton, especially phytoplankton with highly similar fluorescence spectral characteristics, and the similarity is too high, which leads to difficulties in analysis. Figure 2As shown in the figure, green algae, diatoms and dinoflagellates have similar characteristic spectral regions and similar noise regions, and the characteristic spectral regions account for a small proportion of the full spectrum, only about 30-50%. Therefore, when substituting into the full spectrum analysis, nearly half or more of the noise region will reduce the recognition and quantitative accuracy of these algae. However, directly narrowing the recognition spectral region may make it difficult to cover the characteristic fluorescence spectral regions of cyanobacteria and cryptophytes.

[0015] To solve the above problems, the steps of the hierarchical partition analysis algorithm in this embodiment are as follows: 1. Establish a standard spectral library of phytoplankton interpolation of different classes based on non-uniform interpolation method, and follow the steps below: Step 1: Obtain the fluorescence spectrum of living phytoplankton, with an excitation wavelength range of 370 nm to 650 nm and an emission wavelength range of 600 nm to 720 nm. Measure the standard chlorophyll a concentration of the living phytoplankton; Step 2: Dividing the living phytoplankton fluorescence spectrum obtained in step 1 by the standard chlorophyll a concentration of the living phytoplankton to obtain a concentration normalized spectrum of the phytoplankton, i.e., a standard spectrum; Step 3: Calculate the mean fluorescence spectrum M and the corresponding relative standard deviation S of the phytoplankton according to the standard spectrum in step 2; Step 4: Calculate M±2 S, to obtain the upper and lower limits of the interpolation range of the phytoplankton standard spectrum library; Step 5: Perform uniform interpolation within the interpolation range of the phytoplankton standard spectrum library described in step 5, and the number of interpolations is 10; Step 6: Repeat steps 1 to 5 to obtain the standardized spectral library of different types of phytoplankton; 2. Obtain the living three-dimensional fluorescence spectrum of the unknown phytoplankton sample, and the spectral range is the same as step 1 in 1; 3. The present invention analyzes the spectrum based on the least squares multivariate linear regression analysis concept.

[0016] The idea of ​​multivariate linear regression analysis is briefly described as follows: For a certain type of algae, its three-dimensional fluorescence spectrum measurement value is a fluorescence intensity data matrix, theoretically: , Among them, M i is the in vivo three-dimensional fluorescence measurement spectrum of a certain algae sample, a i It indicates that the concentration of algae samples of this phylum is related to the concentration of algae samples of this phylum. i Represents the standard fluorescence spectrum of algae samples of this phylum.

[0017] When there are multiple components in the sample, within the range of Lambert-Beer's law, the algae fluorescence spectrum satisfies the linear additivity, that is: , , Where: M represents the generation of unknown algae sample aliased in vivo three-dimensional fluorescence measurement spectrum; f ij is the jth standard spectrum in the standard spectrum library of phytoplankton of the i-th class; a i is the chlorophyll concentration of the i-th phylum, χ is the error; λ mn is the fluorescence intensity of the ith algae at excitation wavelength n and emission wavelength m.

[0018] In order to minimize the error, the non-negative least squares method is used to analyze the spectrum to be tested. At the same time, since the concentration of algae in the mixed algae sample to be tested will not be negative, it is necessary to add non-negative restrictions in the iterative operation and output the least squares result.

[0019] , 4. During analysis, the unknown aliased three-dimensional fluorescence spectrum to be analyzed is analyzed according to the following hierarchical partition analysis ideas, such as Figure 3 First level: Full spectrum Participate in the analysis, the purpose is to analyze the concentration of Cyanobacteria and Cryptophyta, and obtain the difference spectrum in the full spectrum area ; Second layer: partition analysis, take The area containing effective spectral features Participate in the analysis, the purpose is to analyze the chlorophyll a concentration of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta.

[0020] 4.1 First level: Optimal analysis in the interpolation spectral library of cyanobacteria and cryptophytes in the full spectral region Step 1: The unknown algae sample is mixed with the in vivo three-dimensional fluorescence spectrum, and a non-negative least squares method multivariate linear regression analysis is performed on the cyanobacteria and cryptophytes in the full spectrum range, and the full spectrum range is the spectrum range described in step 1 in 1: , in: It represents the three-dimensional fluorescence spectrum of living phytoplankton within the excitation wavelength range of 370 nm to 650 nm and the emission wavelength range of 600 nm to 720 nm; It represents the average spectrum of the standardized spectral library of phytoplankton of other phyla except Cyanobacteria and Cryptophyta; They represent the standard spectrum combination with the smallest error obtained by traversing all the combination modes in the standard spectrum library of Cyanobacteria and Cryptophyta when the standard spectrum library of phytoplankton of other phyla except Cyanobacteria and Cryptophyta is averaged; represent the least squares analytical concentrations of phytoplankton of Cyanobacteria and Cryptophyta, respectively; represents the minimum iteration error.

[0021] Step 2: Obtain the difference fluorescence spectrum after deducting Cyanobacteria and Cryptophyta from the whole spectrum : .

[0022] Step 3: Select a new characteristic fluorescence spectrum range within the difference fluorescence spectrum range , It covers most of the fluorescence characteristics of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta, and is set as follows: excitation wavelength range 580 nm~620 nm, emission wavelength range 660 nm~700 nm.

[0023] 4.2 Second layer: optimization analysis of the interpolation spectral library of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta within local features.

[0024] Step 4: Within the same range as step 3, intercept the standard spectra of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta to form a new standard spectral library .

[0025] Step 5: Within the same range as step 3, perform non-negative least squares multivariate linear regression analysis on Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta: , in: They respectively represent the standard spectrum combination with the smallest error obtained after traversing all the combination methods in the standard spectrum library of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta within the same range of step 3; represent the least squares analytical concentrations of Chlorophyta, Bacillariophyta, Dinophyta, and Xanthophyta, respectively; represents the minimum iteration error.

[0026] Step 6: Output the chlorophyll a concentration of different types of phytoplankton, namely: , Example 2 Three-dimensional fluorescence spectra of living phytoplankton under different habitat conditions Common phytoplankton species in my country's coastal and freshwater waters were selected, including Cyanobacteria, Chlorophyta, Bacillariophyta, Dinophyta, Xanthophyta and Cryptophyta, see Table 1. They were cultured according to GB / T21805-2008 "Chemicals - Algal Growth Inhibition Test".

[0027] Table 1

[0028]

[0029] Take an appropriate amount of algae sample mother solution at different growth stages, dilute it according to a certain volume ratio, and measure the three-dimensional fluorescence spectrum of the sample in vivo with a fluorescence spectrophotometer. The excitation wavelength range is 370 nm to 650 nm, and the emission wavelength range is 600 nm to 720 nm. At the same time, refer to the national standard method to measure the standard concentration of chlorophyll a in the algae sample.

[0030] Example 3 Obtaining interpolation standard spectral libraries of different phytoplankton The algae samples with known chlorophyll a standard concentration obtained in Example 2 were divided into an experimental set and a test set in a ratio of 7:3.

[0031] The three-dimensional fluorescence spectrum of the living algae in the experimental set was divided by the standard concentration of chlorophyll a, thereby obtaining a series of three-dimensional fluorescence standard spectra of living algae. According to the standard spectra of different categories, the mean fluorescence spectrum M and the corresponding relative standard deviation S of the phytoplankton were calculated; M ± 2 S, obtaining the upper and lower limits of the interpolation range of the standard spectrum library of phytoplankton; performing uniform interpolation within the interpolation range of the standard spectrum library of phytoplankton described in step 4, the number of interpolations is 10, thereby establishing the standard spectrum library of phytoplankton.

[0032] The above method obtains the standardized spectral library of phytoplankton of other specific phyla including Cyanobacteria, Chlorophyta, Bacillariophyta, Dinophyta, Xanthophyta and Cryptophyta.

[0033] Example 4 Comparison of hierarchical and partitioned analysis effects The analytical method of the present invention is used to analyze the living fluorescence spectra of the phytoplankton in the test set obtained in Example 2, and the analytical results are compared with the standard chlorophyll a concentration of the sample and the direct analytical results of the least squares multivariate linear regression algorithm (i.e., the common analytical algorithm) of the non-stratified partition. The test set samples include 209 pure algae samples, including 44 cyanobacteria, 60 green algae, 85 diatoms, 60 dinoflagellates, 46 yellow algae, and 14 cryptophytes, and 20 mixed samples. The false recognition rate is 0. and relative measurement error Indicates the parsing accuracy of the algorithm.

[0034] , , in: and They represent the total number of samples and the number of samples misidentified by the analysis for a specific phylum of phytoplankton, respectively; and They respectively represent the standard concentration and analytical concentration of chlorophyll a in a certain phylum of phytoplankton samples.

[0035] Table 2

[0036]

[0037] Table 2 and Figure 4 It shows that the total number of misidentifications of the test set by the common parsing algorithms is 34, accounting for 11.0% of the total number of samples. The number of misidentifications of Bacillariophyta, Dinophyta and Xanthophyta are 14, 10 and 9 respectively, and the misidentification rates are 16.5%, 16.7% and 64.3% respectively, and the misidentification rate of Chlorophyta and Cryptophyta samples is 0%. In comparison, the hierarchical partitioning parsing algorithm significantly reduces the number of misidentified samples of algae in the test set, with a total of only 4 misidentified samples, accounting for 1.3% of the total number of samples. The number of misidentifications of Cyanobacteria, Bacillariophyta, Dinophyta and Xanthophyta is reduced to 0, 0, 3 and 1 respectively, and the misidentification rates are reduced by 2.3%, 16.5%, 11.7% and 57.1% respectively. Table 2 and Figure 4 The results show that the hierarchical partitioning analysis algorithm performs significantly better than the common analysis algorithms in reducing misidentification, especially when dealing with fluorescence spectra of living organisms with high similarity such as Bacillariophyta, Dinophyta and Xanthophyta.

[0038] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-component phytoplankton concentration measurement algorithm based on fluorescence spectrum stratification and partitioning, characterized in that: The steps include: Step 1: The unknown algae sample is mixed with the in vivo three-dimensional fluorescence spectrum, and a non-negative least squares multivariate linear regression analysis is performed on the cyanobacteria and cryptophytes in the full spectrum: ， in: It represents the three-dimensional fluorescence spectrum of living phytoplankton within the excitation wavelength range of 370 nm to 650 nm and the emission wavelength range of 600 nm to 720 nm; It represents the average spectrum of the standardized spectral library of phytoplankton of other phyla except Cyanobacteria and Cryptophyta; They represent the standard spectrum combination with the smallest error obtained by traversing all the combination modes in the standard spectrum library of Cyanobacteria and Cryptophyta when the standard spectrum library of phytoplankton of other phyla except Cyanobacteria and Cryptophyta is averaged; represent the least squares analytical concentrations of phytoplankton of Cyanobacteria and Cryptophyta, respectively; represents the minimum iteration error; Step 2: Obtain the difference fluorescence spectrum after deducting Cyanobacteria and Cryptophyta from the whole spectrum : ， Step 3: Select a new characteristic fluorescence spectrum range within the difference fluorescence spectrum range , The fluorescence characteristics of most of the Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta were covered in the experiment, and the excitation wavelength ranged from 580 nm to 620 nm, and the emission wavelength ranged from 660 nm to 700 nm. Step 4: Within the same range as step 3, intercept the standard spectra of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta to form a new standard spectral library ; Step 5: Within the same range as step 3, perform a multivariate linear regression analysis of Chlorophyta, Bacillariophyta, Dinophyta, and Xanthophyta using the non-negative least squares method: ， in: They represent the standard spectrum combination with the smallest error obtained by traversing all the combination modes in the standard spectrum library of Chlorophyta, Bacillariophyta, Dinophyta and Xanthophyta within the same range of step 3; represent the least squares analytical concentrations of Chlorophyta, Bacillariophyta, Dinophyta, and Xanthophyta, respectively; represents the minimum iteration error; Step 6: Output the chlorophyll a concentration of different types of phytoplankton, namely: 。

Images