WO2022241883A1 - 基于三维液相色谱指纹的污染源识别方法及装置 - Google Patents

基于三维液相色谱指纹的污染源识别方法及装置 Download PDF

Info

Publication number
WO2022241883A1
WO2022241883A1 PCT/CN2021/100133 CN2021100133W WO2022241883A1 WO 2022241883 A1 WO2022241883 A1 WO 2022241883A1 CN 2021100133 W CN2021100133 W CN 2021100133W WO 2022241883 A1 WO2022241883 A1 WO 2022241883A1
Authority
WO
WIPO (PCT)
Prior art keywords
sample
pollution source
liquid chromatography
dimensional liquid
fingerprint
Prior art date
Application number
PCT/CN2021/100133
Other languages
English (en)
French (fr)
Inventor
吴静
刘博�
Original Assignee
清华大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 清华大学 filed Critical 清华大学
Publication of WO2022241883A1 publication Critical patent/WO2022241883A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
    • G01N30/02—Column chromatography
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
    • G01N30/02—Column chromatography
    • G01N30/86—Signal analysis
    • G01N30/8675—Evaluation, i.e. decoding of the signal into analytical information
    • G01N30/8686—Fingerprinting, e.g. without prior knowledge of the sample components

Definitions

  • This application relates to the technical field of environmental supervision, in particular to a method for identifying pollution sources based on three-dimensional liquid chromatography fingerprints
  • Pollution source identification has always been a hot and difficult point in the field of environmental supervision.
  • the identification of pollution sources mainly relies on manual investigation.
  • the manual investigation method based on the pollution source database can greatly reduce the workload of manual investigation and improve the timeliness of pollution source identification, which has become a popular method for pollution source identification in recent years.
  • the database proposed by Wan Pingyu et al. of Beijing University of Chemical Technology contained anion types, organic matter types, metal element types and fluorescence information, but the operability was not strong.
  • the databases of these methods are relatively complex and require more instruments and equipment, and these methods only provide a database establishment method, and do not provide an automatic comparison algorithm between contaminated samples and pollution source databases.
  • experienced experts are often required to perform manual comparison and judgment, which has strong subjectivity. In the absence of experts, the accuracy is low, limiting the scope and application of the method. Therefore, it is necessary to develop a simple, intelligent and accurate pollution source identification method based on the existing technology.
  • This application aims to solve one of the technical problems in the related art at least to a certain extent.
  • one purpose of this application is to propose a pollution source identification method based on three-dimensional liquid chromatography fingerprints.
  • This method uses a self-organizing neural network to realize automatic comparison and identification of three-dimensional liquid chromatography fingerprints, which is simple, intelligent and accurate. It is of great significance to trace the source of pollution.
  • Another object of the present application is to propose a pollution source identification device based on three-dimensional liquid chromatography fingerprints.
  • an embodiment of the present application proposes a method for identifying pollution sources based on three-dimensional liquid chromatography fingerprints, including the following steps:
  • a pollution source identification device based on three-dimensional liquid chromatography fingerprints, including:
  • the sample processing module is used for preprocessing the pollution source samples and the samples to be identified;
  • the sample collection module is used to collect the three-dimensional liquid chromatography fingerprints of the pretreated pollution source samples and the samples to be identified, and preprocess the collected three-dimensional liquid chromatography fingerprints;
  • the identification module is used to use the processed pollution source sample and the three-dimensional liquid chromatography fingerprint of the sample to be identified to establish a pollution source identification model and perform pollution source identification, so as to determine the pollution source to which the sample to be identified belongs.
  • the self-organizing neural network is used to realize the automatic comparison of three-dimensional liquid chromatography fingerprints, which is simple, intelligent and accurate, and has great significance for pollution traceability.
  • Fig. 1 is a flow chart of a pollution source identification method based on a three-dimensional liquid chromatography fingerprint according to an embodiment of the present application
  • FIG. 2 is a block diagram of a pollution source identification based on a three-dimensional liquid chromatography fingerprint according to an embodiment of the present application
  • Fig. 3 is a self-organizing neural network model training error diagram according to one embodiment of the present application.
  • FIG. 4 is a U-matrix matrix diagram according to an embodiment of the present application.
  • Fig. 5 is a best matching neuron (BMU) map according to one embodiment of the present application.
  • Fig. 6 is a K-means clustering result diagram of the self-organizing neural network model according to one embodiment of the present application.
  • FIG. 7 is a diagram of a self-organizing neural network model recognition result according to an embodiment of the present application.
  • Fig. 8 is a schematic structural diagram of a pollution source identification device based on a three-dimensional liquid chromatography fingerprint according to an embodiment of the present application.
  • Fig. 1 is a flowchart of a pollution source identification method based on three-dimensional liquid chromatography fingerprints according to an embodiment of the present application.
  • the pollution source identification method based on the three-dimensional liquid chromatography fingerprint includes the following steps:
  • Step S1 preprocessing the pollution source samples and the samples to be identified.
  • the number of pollution sources to which the pollution source samples belong is greater than or equal to 2.
  • the pollution source samples and the samples to be identified include liquid samples, solid samples or gaseous samples.
  • the pretreatment of liquid samples refers to filtration with a 0.2-10.0 ⁇ m filter membrane.
  • the pretreatment of solid samples refers to taking appropriate quality samples and dissolving them with ultrapure water, and then filtering the leaching solution of solid samples with a 0.2-10.0 ⁇ m filter membrane.
  • the pretreatment of gaseous samples refers to extracting an appropriate volume of gas and dissolving it in ultrapure water, and then filtering the water sample with a 0.2-10.0 ⁇ m filter membrane.
  • Step S2 collecting the three-dimensional liquid chromatography fingerprints of the pretreated pollution source samples and the samples to be identified, and performing preprocessing on the collected three-dimensional liquid chromatography fingerprints.
  • the collected three-dimensional liquid chromatography fingerprints include, but are not limited to, multi-absorption chromatographic fingerprints, multi-excitation chromatographic fingerprints, and multi-emission chromatographic fingerprints.
  • the multi-absorption chromatographic fingerprint refers to the spectrum collected by the liquid chromatography of the sample under the multi-wavelength mode of the diode array detector, the wavelength range is 200-800 nm, and the wavelength interval is 1-20 nm.
  • the multi-excitation chromatographic fingerprint refers to the spectrum collected by the liquid chromatography of the sample under the multi-excitation mode of the fluorescence detector, the excitation wavelength range is 200-600nm, and the wavelength interval is 1-20nm.
  • the multi-emission chromatographic fingerprint refers to the spectrum collected by liquid chromatography in the multi-emission mode of the fluorescence detector, the emission wavelength scanning range is 220-750nm, and the wavelength interval is 1-20nm.
  • the chromatographic columns used for three-dimensional liquid chromatography fingerprint collection include but are not limited to size exclusion chromatographic columns, reversed phase chromatographic columns, forward chromatographic columns and hydrophilic chromatographic columns.
  • a size exclusion chromatographic column can be selected.
  • the collected three-dimensional liquid chromatography fingerprints are preprocessed, including but not limited to blank subtraction and normalization, principal component analysis, and multivariate curvature resolution.
  • Step S3 using the preprocessed pollution source samples and the three-dimensional liquid chromatography fingerprints of the samples to be identified to establish a pollution source identification model to identify the pollution sources to which the samples to be identified belong.
  • the algorithm used to establish the pollution source identification model includes but not limited to multi-layer perceptron, backpropagation (BP) neural network, radial basis function neural network (RBF), convolutional neural network, self-organizing neural network, as a preferred Self-organizing neural network model is adopted.
  • BP backpropagation
  • RBF radial basis function neural network
  • convolutional neural network self-organizing neural network
  • self-organizing neural network as a preferred Self-organizing neural network model is adopted.
  • the steps to establish a pollution source identification model are:
  • n is generally greater than or equal to 10;
  • the visual analysis of the SOM model includes exporting the best matching neuron (BMU) map and the unified clustering matrix (U-matrix) map, both of which can indicate the clustering boundary and initially analyze the clustering results;
  • SOM model cluster analysis refers to utilizing the k-means clustering algorithm to divide the neurons of the best SOM model into several categories;
  • the determination of the recognition results of the SOM recognition model refers to the determination of the final sample class results after comprehensive analysis of the BMU map, U-matrix graph and k-means clustering results. If the samples to be identified and samples from a certain pollution source are grouped together, it is considered that the pollution is likely to come from this pollution source. If the sample to be identified is not clustered with any sample from any pollution source in the database, it means that there are other potential pollution sources.
  • One sample X1 is collected from the polluted water body, and six samples are collected from pollution sources A and B, namely A1 ⁇ A6 and B1 ⁇ B6.
  • the three-dimensional liquid chromatography fingerprint refers to the multi-absorption chromatography fingerprint.
  • the wavelength range of the multi-absorption chromatographic fingerprint is 220-360nm, the wavelength interval is 2nm, and the chromatographic column is size exclusion chromatography.
  • the multi-absorbance chromatographic fingerprint data of all samples were subtracted from the multi-absorbance chromatographic fingerprint data of the blank sample, and then the multi-absorbance chromatographic fingerprint data were normalized.
  • MQE mean quantization error
  • TGE topological graph error
  • the unified clustering matrix (U-matrix) diagram is shown in Figure 4, and the results show that the distance between neurons on the left and right sides is relatively large.
  • the best matching neuron (BMU) map shows that the left sample comes from pollution source A, and the right sample comes from pollution source B, and there are more empty neurons between the neurons of A and B samples, indicating clustering boundary.
  • the unknown sample X1 is closer to the sample of pollution source A, indicating that the pollution source of X1 is probably A.
  • the pollution source sample and the sample to be identified are pretreated; the three-dimensional liquid chromatography fingerprints of the pretreated pollution source sample and the sample to be identified are collected.
  • the three-dimensional liquid chromatography fingerprint of the pretreated pollution source sample and the three-dimensional liquid chromatography fingerprint of the sample to be identified are used to establish a pollution source identification model and identify the pollution source to determine the pollution source to which the sample to be identified belongs.
  • This application requires less equipment, less sample size, rich fingerprint information, and low cost, which is conducive to large-scale promotion.
  • the self-organizing neural network is used to realize the automatic comparison and identification of three-dimensional liquid chromatography fingerprints. It is simple, intelligent and accurate, and is of great significance to trace the source of pollution.
  • Fig. 8 is a schematic structural diagram of a pollution source identification device based on a three-dimensional liquid chromatography fingerprint according to an embodiment of the present application.
  • the device for identifying pollution sources based on three-dimensional liquid chromatography fingerprints includes: a sample processing module 801 , a sample collection module 802 and an identification module 803 .
  • the sample processing module 801 is used for preprocessing the pollution source samples and the samples to be identified.
  • the sample collection module 802 is configured to collect the three-dimensional liquid chromatography fingerprints of the pretreated pollution source samples and the samples to be identified, and perform preprocessing on the collected three-dimensional liquid chromatography fingerprints.
  • the identification module 803 is configured to use the preprocessed pollution source samples and the three-dimensional liquid chromatography fingerprints of the samples to be identified to establish a pollution source identification model and perform pollution source identification, so as to determine the pollution sources to which the samples to be identified belong.
  • the pollution source identification device based on the three-dimensional liquid chromatography fingerprint proposed in the embodiment of the present application, by preprocessing the pollution source sample and the sample to be identified; collecting the three-dimensional liquid chromatography fingerprint of the pretreated pollution source sample and the sample to be identified, the collected The three-dimensional liquid chromatography fingerprint of the pretreated pollution source sample and the three-dimensional liquid chromatography fingerprint of the sample to be identified are used to establish a pollution source identification model and identify the pollution source to determine the pollution source to which the sample to be identified belongs.
  • This application requires less equipment, less sample size, rich fingerprint information, and low cost, which is conducive to large-scale promotion.
  • the self-organizing neural network is used to realize the automatic comparison and identification of three-dimensional liquid chromatography fingerprints. It is simple, intelligent and accurate, and is of great significance to trace the source of pollution.
  • first and second are used for descriptive purposes only, and cannot be interpreted as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features.
  • the features defined as “first” and “second” may explicitly or implicitly include at least one of these features.
  • “plurality” means at least two, such as two, three, etc., unless otherwise specifically defined.

Landscapes

  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Library & Information Science (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
  • Other Investigation Or Analysis Of Materials By Electrical Means (AREA)

Abstract

本申请公开了一种基于三维液相色谱指纹的污染源识别方法及装置，该方法包括：对污染源样品和待识别样品进行预处理；采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，对采集的三维液相色谱指纹进行预处理；利用预处理后的三维液相色谱指纹建立污染源识别模型进行污染源识别，以确定待识别样品所属的污染源。该装置包括样品处理模块、指纹采集模块和识别模块。该方法所需设备少、样本量少、指纹信息丰富、成本低，有利于大范围推广。利用自组织神经网络实现了三维液相色谱指纹的自动比对和识别，具有简单、智能、准确的特点，对污染源溯源具有重要意义。

Description

基于三维液相色谱指纹的污染源识别方法及装置
相关申请的交叉引用
本申请要求清华大学于2021年05月17日提交的、发明名称为“基于三维液相色谱指纹的污染源识别方法及装置”的、中国专利申请号“202110534813.9”的优先权。
技术领域
本申请涉及环境监管技术领域，特别涉及一种基于三维液相色谱指纹的污染源识别方法
背景技术
污染源识别一直是环境监管领域的热点和难点。目前对污染源的识别主要依靠人工排查。基于污染源数据库辅助的人工排查方法可以大大减少人工排查的工作量，提高污染源识别的时效性，成为近年来污染源识别的热门方法。此前，北京化工大学万平玉等人提出的数据库包含阴离子种类、有机物种类、金属元素种类和荧光信息，但可操作性不强。清华大学吴静等人提出了一个可操作性较强的水污染源数据库建立方法，包括pH值、电导率等常规水质指标和荧光水纹、三维分子量水纹等水质指纹。这些方法的数据库都较复杂，需要较多的仪器设备，而且这些方法仅仅提供了一个数据库建立方法，没有提供污染样本与污染源数据库的自动比对算法。在实际应用时往往需要有经验的专家进行人工比对和判断，具有较强的主观性。在没有专家的情况下，准确性较低，限制了该方法的范围推广和应用。因此，很有必要在现有技术基础上，发展一种简单、智能、准确的污染源识别方法。
发明内容
本申请旨在至少在一定程度上解决相关技术中的技术问题之一。
为此，本申请的一个目的在于提出一种基于三维液相色谱指纹的污染源识别方法，该方法利用自组织神经网络实现了三维液相色谱指纹的自动比对和识别，具有简单、智能、准确的特点，对污染源溯源具有重要意义。
本申请的另一个目的在于提出一种基于三维液相色谱指纹的污染源识别装置。
为达到上述目的，本申请一方面实施例提出了一种基于三维液相色谱指纹的污染源识别方法，包括以下步骤：
对污染源样品和待识别样品进行预处理；
采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，对采集的三维液相色谱指纹进行预处理；
利用预处理后的所述污染源样品和所述待识别样品的三维液相色谱指纹建立污染源识别模型进行污染源识别，以确定所述待识别样品所属的污染源。
为达到上述目的，本申请另一方面实施例提出了一种基于三维液相色谱指纹的污染源识别装置，包括：
样品处理模块，用于对污染源样品和待识别样品进行预处理；
样品采集模块，用于采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，并对采集的三维液相色谱指纹并进行预处理；
识别模块，用于利用处理后的所述污染源样品和所述待识别样品的三维液相色谱指纹建立污染源识别模型并进行污染源识别，以确定所述待识别样品所属的污染源。
本申请实施例的基于三维液相色谱指纹的污染源识别方法及装置，具有以下有益效果：
1)利用一台设备即可实现多种三维液相色谱指纹的测试，所需设备少、样本量少、指纹信息丰富、成本低，有利于大范围推广；
2)采用自组织神经网络实现了三维液相色谱指纹进行自动比对，具体简单、智能、准确的特点，对污染溯源具有重要意义。
本申请附加的方面和优点将在下面的描述中部分给出，部分将从下面的描述中变得明显，或通过本申请的实践了解到。
附图说明
本申请上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解，其中：
图1为根据本申请一个实施例的基于三维液相色谱指纹的污染源识别方法流程图；
图2为根据本申请一个实施例的基于三维液相色谱指纹的污染源识别方流程框图；
图3为根据本申请一个实施例的自组织神经网络模型训练误差图；
图4为根据本申请一个实施例的U-matrix矩阵图；
图5为根据本申请一个实施例的最佳匹配神经元(BMU)映射图；
图6为根据本申请一个实施例的自组织神经网络模型K-均值聚类结果图；
图7为根据本申请一个实施例的自组织神经网络模型识别结果图；
图8为根据本申请一个实施例的基于三维液相色谱指纹的污染源识别装置结构示意图。
具体实施方式
下面详细描述本申请的实施例，所述实施例的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的，旨在用于解释本申请，而不能理解为对本申请的限制。
下面参照附图描述根据本申请实施例提出的基于三维液相色谱指纹的污染源识别方法及装置。
首先将参照附图描述根据本申请实施例提出的基于三维液相色谱指纹的污染源识别方法。
图1为根据本申请一个实施例的基于三维液相色谱指纹的污染源识别方法流程图。
如图1所示，该基于三维液相色谱指纹的污染源识别方法包括以下步骤：
步骤S1，对污染源样品和待识别样品进行预处理。
进一步地，在本申请的实施例中，污染源样品所属的污染源个数大于等于2。
具体地，污染源样品和待识别样品包括液态样品、固态样品或气态样品。
进一步地，针对不同的样品使用不同的预处理方法，对液态样品预处理是指用0.2～10.0μm滤膜过滤。对固态样品预处理是指取适当质量样品用超纯水溶解，然后固态样品浸出液用0.2～10.0μm滤膜过滤。对气态样品预处理是指抽取适当体积的气体溶于超纯水，然后水样用0.2～10.0μm滤膜过滤。
步骤S2，采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，对采集的三维液相色谱指纹进行预处理。
具体地，采集的三维液相色谱指纹包括但不限于多吸收色谱指纹、多激发色谱指纹和多发射色谱指纹。
其中，多吸收色谱指纹是指样品用液相色谱在二极管阵列检测器多波长模式下采集的图谱，波长范围200～800nm，波长间隔1～20nm。多激发色谱指纹是指样品用液相色谱在荧光检测器多激发模式下采集的图谱，激发波长范围200～600nm，波长间隔1～20nm。多发射色谱指纹是指用液相色谱在荧光检测器多发射模式下采集的图谱，发射波长扫描范围220～750nm，波长间隔1～20nm。
进一步地，三维液相色谱指纹采集所用色谱柱包括但不限于体积排阻色谱柱、反相色谱柱、正向色谱柱和亲水色谱柱。作为优选的方案，可以选择体积排阻色谱柱。
进一步地，在本申请的实施例中，还包括：
采集空白样本，对空白样本进行预处理；
采集预处理后的空白样本的三维液相色谱指纹。
进一步地，对采集的三维液相色谱指纹并进行预处理，包括：但不限于扣除空白和归一化、主成分分析、多元曲率分辨。
步骤S3，利用预处理后的污染源样品和待识别样品的三维液相色谱指纹建立污染源识别模型进行污染源识别，以确定待识别样品所属的污染源。
进一步地，建立污染源识别模型所用算法包括但不限于多层感知器、反向传播(BP)神经网络、径向基函数神经网络(RBF)、卷积神经网络、自组织神经网络，作为优选可以采用自组织神经网络模型。通过污染源识别模型对污染源样品和待识别样品的三维液相色谱指纹进行可视化分析和聚类分析，根据聚类分析结果判断待识别样品所属的污染源。
具体地，建立污染源识别模型的步骤为：
a)导入预处理后的污染样品和污染源样品的三维液相色谱指纹数据；
b)线性初始化建立并训练至少1个SOM模型；
c)随机初始化建立并训练n个SOM模型，n一般大于等于10；
d)选择最佳SOM模型：从b)和c)建立的SOM模型中选择平均量化误差(MQE)和拓扑图形误差(TGE)最小化的模型作为最佳SOM模型；
e)SOM模型可视化分析：SOM模型可视化分析包括导出最佳匹配神经元(BMU)映射图和统一聚类矩阵(U-matrix)图，二者可以指示聚类边界，初步分析聚类结果；
f)SOM模型聚类分析：SOM模型聚类分析是指利用k-均值聚类算法将最佳SOM模型的神经元分为若干类别；
g)SOM模型识别结果的确定
SOM识别模型识别结果的确定是指综合分析BMU映射图、U-matrix图和k-均值聚类结果后确定最终的样品类结果。若待识别样品和某个污染源的样品聚为一类，则认为污染很可能来自该污染源。若待识别样品未与数据库中任何一个污染源的样品聚为一类，则说明还存在其他潜在污染源。
下面通过一个具体实施例对本申请的基于三维液相色谱指纹的污染源识别方法进行说明。
如图2所示，以液态样品为例，基于三维液相色谱指纹的污染源识别方法步骤为：
1)污染水体样品和污染源样品采集
从污染水体中采集一个样品X1，从污染源A和B分别采集6个样品即A1～A6和B1～B6。
2)样品预处理
所有样品均用0.45μm的滤膜过滤。
3)三维液相色谱指纹采集
采集步骤1)中样品的三维液相色谱指纹，同时测试一个空白样品的三维液相色谱指纹。本案例中，三维液相色谱指纹是指多吸收色谱指纹。多吸收色谱指纹的波长范围为220～360nm，波长间隔2nm，色谱柱为体积排阻色谱。
4)多吸收色谱指纹预处理
将所有样品的多吸收色谱指纹数据减去空白样品的多吸收色谱指纹数据，然后将多吸收色谱指纹数据归一化。
5)建立基于多吸收色谱指纹的污染源识别模型
(1)将预处理后的多吸收色谱指纹数据导入MATLAB。
(2)线性初始化建立并训练1个SOM模型。
(3)随机初始化建立并训练10个SOM模型。
(4)选择最佳SOM模型。
步骤(2)和(3)建立的11个SOM模型的平均量化误差(MQE)和拓扑图形误差(TGE)如图3所示。模型4(SOM4)的MQE最小，而所有模型的TGE都为0，故SOM4为最佳模型。
(5)SOM模型可视化分析
统一聚类矩阵(U-matrix)图如图4所示，结果显示左右两侧的神经元距离较大。如图5所示，最佳匹配神经元(BMU)映射图显示左侧样品来自污染源A，右侧样品来自污染源B，A和B样品所在神经元间有较多空神经元，指示出聚类边界。未知样品X1与污染源A的样品更为接近，表明X1的污染源很可能是A。
(6)SOM模型聚类分析
只有2个污染源，因此利用k-均值聚类将SOM模型的神经元聚为2类，结果如图6所示。
(7)SOM模型识别结果的确定
如图7所示，综合分析BMU映射图、U-matrix图可以发现k-均值聚类结果合理可靠。未知样品X1与和污染源A的样品聚为一类，因此则认为X1的污染源是A。
根据本申请实施例提出的基于三维液相色谱指纹的污染源识别方法，通过对污染源样品和待识别样品进行预处理；采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，对采集的三维液相色谱指纹进行预处理；利用预处理后的污染源样品和待识别样品的三维液相色谱指纹建立污染源识别模型并进行污染源识别，以确定待识别样品所属的污染源。本申请所需设备少、样本量少、指纹信息丰富、成本低，有利于大范围推广。首次利用自组织神经网络实现了三维液相色谱指纹的自动比对和识别，具有简单、智能、准确的特点，对污染源溯源具有重要意义。
其次参照附图描述根据本申请实施例提出的基于三维液相色谱指纹的污染源识别装置。
图8为根据本申请一个实施例的基于三维液相色谱指纹的污染源识别装置结构示意图。
如图8所示，该基于三维液相色谱指纹的污染源识别装置包括：样品处理模块801、样 品采集模块802和识别模块803。
样品处理模块801，用于对污染源样品和待识别样品进行预处理。
样品采集模块802，用于采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，对采集的三维液相色谱指纹并进行预处理。
识别模块803，用于利用预处理后的污染源样品和待识别样品的三维液相色谱指纹建立污染源识别模型并进行污染源识别，以确定待识别样品所属的污染源。
需要说明的是，前述对方法实施例的解释说明也适用于该实施例的装置，此处不再赘述。
根据本申请实施例提出的基于三维液相色谱指纹的污染源识别装置，通过对污染源样品和待识别样品进行预处理；采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，对采集的三维液相色谱指纹进行预处理；利用预处理后的污染源样品和待识别样品的三维液相色谱指纹建立污染源识别模型并进行污染源识别，以确定待识别样品所属的污染源。本申请所需设备少、样本量少、指纹信息丰富、成本低，有利于大范围推广。首次利用自组织神经网络实现了三维液相色谱指纹的自动比对和识别，具有简单、智能、准确的特点，对污染源溯源具有重要意义。
此外，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。在本申请的描述中，“多个”的含义是至少两个，例如两个，三个等，除非另有明确具体的限定。
在本说明书的描述中，参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本申请的至少一个实施例或示例中。在本说明书中，对上述术语的示意性表述不必须针对的是相同的实施例或示例。而且，描述的具体特征、结构、材料或者特点可以在任一个或多个实施例或示例中以合适的方式结合。此外，在不相互矛盾的情况下，本领域的技术人员可以将本说明书中描述的不同实施例或示例以及不同实施例或示例的特征进行结合和组合。
尽管上面已经示出和描述了本申请的实施例，可以理解的是，上述实施例是示例性的，不能理解为对本申请的限制，本领域的普通技术人员在本申请的范围内可以对上述实施例进行变化、修改、替换和变型。

Claims (10)

  1. 一种基于三维液相色谱指纹的污染源识别方法，其特征在于，包括以下步骤：
    对污染源样品和待识别样品进行预处理；
    采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，并对采集的三维液相色谱指纹进行预处理；
    利用预处理后的所述污染源样品和所述待识别样品的三维液相色谱指纹建立污染源识别模型进行污染源识别，以确定所述待识别样品所属的污染源。
  2. 根据权利要求1所述的方法，其特征在于，所述污染源样品和所述待识别样品包括液态样品、固态样品或气态样品。
  3. 根据权利要求1所述的方法，其特征在于，所述污染源样品所属的污染源个数大于等于2。
  4. 根据权利要求2所述的方法，其特征在于，对所述污染源样品和所述待识别样品进行预处理，包括：
    对所述液态样品预处理包括：用滤膜对所述液态样品进行过滤；
    对所述固态样品预处理包括：将所述固态样品用超纯水溶解，将所述固态样品浸出液用滤膜进行过滤；
    对所述气态样品预处理包括：将所述气态样品溶于超纯水，将水样用滤膜进行过滤。
  5. 根据权利要求1所述的方法，其特征在于，所述三维液相色谱指纹包括多吸收色谱指纹、多激发色谱指纹和多发射色谱指纹；
    所述多吸收色谱指纹为样品用液相色谱在二极管阵列检测器多波长模式下采集的图谱，波长范围200～800nm，波长间隔1～20nm；
    所述多激发色谱指纹为样品用液相色谱在荧光检测器多激发模式下采集的图谱，激发波长范围200～600nm，波长间隔1～20nm；
    所述多发射色谱指纹为样品用液相色谱在荧光检测器多发射模式下采集的图谱，发射波长扫描范围220～750nm，波长间隔1～20nm。
  6. 根据权利要求1或5所述的方法，其特征在于，所述三维液相色谱指纹采集所用色谱柱包括但不限于体积排阻色谱柱、反相色谱柱、正向色谱柱和亲水色谱柱。
  7. 根据权利要求1所述的方法，其特征在于，还包括：
    采集空白样本，对所述空白样本进行预处理；
    采集预处理后的所述空白样本的三维液相色谱指纹。
  8. 根据权利要求1或7所述的方法，其特征在于，对采集的三维液相色谱指纹并进行 预处理，包括但不限于扣除空白和归一化、主成分分析、多元曲率分辨。
  9. 根据权利要求1所述的方法，其特征在于，所述污染源识别模型所用算法包括但不限于多层感知器、反向传播神经网络、径向基函数神经网络、卷积神经网络、自组织神经网络，通过所述污染源识别模型对所述污染源样品和所述待识别样品的三维液相色谱指纹进行可视化分析和聚类分析，根据聚类分析结果判断所述待识别样品所属的污染源。
  10. 一种基于三维液相色谱指纹的污染源识别装置，其特征在于，包括：
    样品处理模块，用于对污染源样品和待识别样品进行预处理；
    指纹采集模块，用于采集预处理后的污染源样品和待识别样品的三维液相色谱指纹，并对采集的三维液相色谱指纹并进行预处理；
    识别模块，用于利用预处理后的所述污染源样品和所述待识别样品的三维液相色谱指纹建立污染源识别模型并进行污染源识别，以确定所述待识别样品所属的污染源。
PCT/CN2021/100133 2021-05-17 2021-06-15 基于三维液相色谱指纹的污染源识别方法及装置 WO2022241883A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110534813.9A CN113311081B (zh) 2021-05-17 2021-05-17 基于三维液相色谱指纹的污染源识别方法及装置
CN202110534813.9 2021-05-17

Publications (1)

Publication Number Publication Date
WO2022241883A1 true WO2022241883A1 (zh) 2022-11-24

Family

ID=77373403

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/100133 WO2022241883A1 (zh) 2021-05-17 2021-06-15 基于三维液相色谱指纹的污染源识别方法及装置

Country Status (2)

Country Link
CN (1) CN113311081B (zh)
WO (1) WO2022241883A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115219472B (zh) 2022-08-12 2023-05-12 生态环境部华南环境科学研究所(生态环境部生态环境应急研究所) 一种定量识别混合水体多污染源的方法及系统

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150198573A1 (en) * 2014-01-14 2015-07-16 Bio-Rad Laboratories, Inc. Method and system for liquid chromatography data analysis
CN109470667A (zh) * 2018-11-14 2019-03-15 华东理工大学 一种结合水质参数和三维荧光光谱进行污染物溯源的方法
CN111426668A (zh) * 2020-04-28 2020-07-17 华夏安健物联科技(青岛)有限公司 一种利用三维荧光光谱特征信息对污染水体溯源分类识别的方法

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
UA78194C2 (en) * 2000-12-08 2007-03-15 Council Scient Ind Res Method for revealing and identification of active components in extracts by chromatographic fingerprinting of "fingerprints" and processor of program data processing chromatograms
CN104950037A (zh) * 2015-06-15 2015-09-30 广州禾信分析仪器有限公司 一种针对挥发性有机物的在线污染源识别监测方法及系统
CN105352928A (zh) * 2015-11-14 2016-02-24 常州大学 一种基于三维荧光光谱的水污染快速识别和定量分析的方法
CN105468926A (zh) * 2015-12-29 2016-04-06 北京师范大学 地下水型饮用水水源地污染源解析方法
CN108008048A (zh) * 2017-12-08 2018-05-08 环境保护部华南环境科学研究所 一种地下水系统中新型污染物的污染源识别方法与应用
CN110412006A (zh) * 2018-04-26 2019-11-05 北京化工大学 一种实现水污染在线溯源的方法
CN109711674A (zh) * 2018-12-03 2019-05-03 北京师范大学 一种用于湖库水体污染溯源的指纹图谱库构建方法及装置
CN109785912A (zh) * 2019-02-13 2019-05-21 中国科学院大气物理研究所 一种用于目标污染物源解析的因子快速识别方法及装置
CN110083585B (zh) * 2019-03-15 2021-02-26 清华大学 一种水污染排放源数据库及其建立方法
CN110163290A (zh) * 2019-05-28 2019-08-23 北华航天工业学院 一种基于快速聚类和化学质量平衡对污染源进行解析的方法
CN110850020B (zh) * 2019-11-11 2022-03-29 中国药科大学 一种基于人工智能的中药识别方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150198573A1 (en) * 2014-01-14 2015-07-16 Bio-Rad Laboratories, Inc. Method and system for liquid chromatography data analysis
CN109470667A (zh) * 2018-11-14 2019-03-15 华东理工大学 一种结合水质参数和三维荧光光谱进行污染物溯源的方法
CN111426668A (zh) * 2020-04-28 2020-07-17 华夏安健物联科技(青岛)有限公司 一种利用三维荧光光谱特征信息对污染水体溯源分类识别的方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WANG ZHIRUI, YU JINGJIE;WANG SHAOPO;NIE YINGJIN;ZHANG YIFAN: "Research Progresseson Application of Three-dimensional Fluorescence Technology in Water Environment Monitoring", ENVIRONMENTAL PROTECTION OF CHEMICAL INDUSTRY, HUA GONG HUAN BAO BIAN JI BU, CN, vol. 40, no. 2, 21 February 2020 (2020-02-21), CN , pages 125 - 130, XP093006197, ISSN: 1006-1878, DOI: 10.3969/j.issn.1006-1878.2020.02.003 *

Also Published As

Publication number Publication date
CN113311081B (zh) 2023-08-11
CN113311081A (zh) 2021-08-27

Similar Documents

Publication Publication Date Title
CN106779418B (zh) 基于神经网络和证据理论的水污染事件智能决策方法
CN107577910B (zh) 一种基于深层次神经网络的车辆尾气浓度反演方法
WO2022241883A1 (zh) 基于三维液相色谱指纹的污染源识别方法及装置
CN113132399A (zh) 一种基于时间卷积网络和迁移学习的工业控制系统入侵检测方法
CN111105389A (zh) 融合Gabor滤波器与卷积神经网络的路面裂缝的检测方法
CN112926767A (zh) 基于粒子群bp神经网络的环雾状流气相表观流速预测方法
CN109784390B (zh) 一种人工智能嗅觉动态响应图谱气体检测识别方法
Yin et al. Identification and classification of atmospheric particles based on SEM images using convolutional neural network with attention mechanism
CN115659195A (zh) 一种大气污染在线识别方法
Cui et al. Deep learning methods for atmospheric PM2. 5 prediction: A comparative study of transformer and CNN-LSTM-attention
CN116858822A (zh) 一种基于机器学习和拉曼光谱的水体中磺胺嘧啶定量分析方法
CN116644278A (zh) 基于样本筛选和特征提取的双模型轴承异常检测算法
CN113033623A (zh) 基于紫外-可见吸收光谱的污染源识别方法及系统
CN115508322A (zh) 一种水污染源多维度溯源方法及装置
CN115880472A (zh) 一种电力红外图像数据智能诊断分析系统
CN115277159A (zh) 一种基于改进随机森林的工业互联网安全态势评估方法
CN114186625A (zh) 一种基于图像特征融合的木材鉴定方法及系统
CN114878543A (zh) 一种痕量物质的拉曼光谱探测识别方法及系统
CN111007220B (zh) 一种生猪养殖污水水质敏捷监测的方法
CN113807431A (zh) 一种基于多源信息融合的智能主轴状态评估方法和系统
CN102707025A (zh) 一种湖库营养状态的智能判别方法及装置
Yin et al. Open-set recognition for unknown organic pollutants in drinking water with three-dimensional fluorescence spectroscopy
CN115329862A (zh) 基于dbscan和随机森林算法的半监督气体识别方法
Zhao et al. Research on Transformer Oil Multi-frequency Ultrasonic Monitoring Technology Based on Convolutional Neural Network
Yang et al. Prediction of pollutant diffusion trend using CNN-GRU hybrid network in IoT air monitoring system

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21940332

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21940332

Country of ref document: EP

Kind code of ref document: A1