Online detection method for trace impurities in benzene product

Abstract

The invention discloses an online detection method for trace impurities in a benzene product, and belongs to the field of online analysis of trace impurities in high-purity chemical organic products. The method comprises the following steps: collecting Raman spectrums of a benzene product mixture with known content and a pure substance benzene component based on the same measurement condition; carrying out pretreatment on the collected Raman spectrum; performing spectrum peak decomposition based on a Lorentz Gaussian convolution model on the preprocessed Raman spectrum to obtain Raman spectrum signals of the benzene component, the toluene and the non-aromatic hydrocarbon component; constructing a quantitative analysis model based on the effective peak height of the impurity peak and the relative concentration of the impurity; and calculating the mass percent of the impurities in the to-be-detected sample through the quantitative analysis model. The method effectively solves the problem of mutual overlapping of benzene peaks and impurity component signals in the Raman spectrum by utilizing a spectrum analysis technology, can remarkably improve the detection precision of trace impurities and reduce the detection limit, and has the advantages of low demand quantity of training samples, strong model extrapolation, short analysis time, accurate prediction result and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of online analysis of trace impurities in high-purity chemical organic products, in particular to an online quantitative detection method for trace impurities in benzene products by combining Raman spectroscopy and spectral peak decomposition technology, and specifically relates to an online detection method for trace impurities in benzene products. BACKGROUND

[0002] Benzene is the simplest aromatic hydrocarbon, which is difficult to dissolve in water and easy to dissolve in organic solvents, and can also be used as an organic solvent. As a basic raw material for petrochemical industry, benzene is widely used in the production of dyes, synthetic rubber, synthetic resin, synthetic fiber and other petrochemical processes, and the production and technical level thereof directly reflect the development level of national petrochemical industry.

[0003] The benzene product is currently mainly separated from mixed aromatic hydrocarbons by extraction distillation technology. After the feed de-pentane oil is separated from heavy components by pre-fractionation column, it is introduced into the middle of the extraction distillation column (ED column), and the circulating solvent is introduced into the top of the column; after extraction distillation, non-aromatic hydrocarbons (NA) are separated at the top of the column, and the benzene-rich solvent at the bottom of the column is introduced into the solvent recovery column. The benzene from the top of the recovery column is recovered, and the poor solvent is recycled after heat exchange; the benzene from the top of the recovery column is introduced into the white clay tower to remove trace olefins to obtain qualified benzene products.

[0004] The national standard "GB / T 3405-2011 Petroleum Benzene" puts forward clear requirements for the purity and impurity content of benzene products: for superior products, the purity of benzene is ≥99.90 wt%, the impurity non-aromatic hydrocarbons (NA) is ≤0.10 wt%, and the impurity toluene is ≤0.05 wt%. In order to ensure the purity of benzene products, the content of each impurity component in benzene product liquid (recovery column top liquid) needs to be detected in time and accurately, so as to provide help and guidance for process operation.

[0005] As a common online analyzer, online gas chromatograph is widely used. However, for the above-mentioned mixed liquid, online chromatography has the following disadvantages: long measurement period, complex sample pretreatment system, and the need for regular replacement of carrier gas, chromatographic column, and switching valve, etc., and large amount of daily maintenance work. Unlike online chromatography, online Raman spectroscopy has the advantages of fast detection speed, simple operation, field maintenance-free, clean and environmentally friendly, and is especially suitable for real-time rapid analysis of aromatic hydrocarbon mixtures, and has been successfully applied to catalytic reforming and PX devices. However, due to the low content of impurities in benzene products, for in-situ online Raman instrument, how to improve the detection precision and reduce the detection limit is a challenging task. SUMMARY

[0006] In view of the deficiencies of the prior art, the purpose of the present application is to provide an online detection method for trace impurities in benzene products.

[0007] The objective of this invention is achieved through the following technical solution: an online detection method for trace impurities in benzene products, comprising the following steps: S1. Under the same measurement conditions, use an online Raman spectrometer to collect the original Raman spectra of a mixture of benzene products with known mass percentages (i.e., a liquid mixture containing benzene, toluene, and non-aromatic hydrocarbons) and the pure benzene component. S2. Perform digital signal processing on the original Raman spectrum, including fluorescence background spectrum removal, smoothing filtering, and baseline correction. S3. Perform peak decomposition on the Raman spectrum after digital signal processing to extract the Raman spectral signals of benzene, toluene and non-aromatic components from the spectrum of the benzene product mixture; S4. Subtract the spectral signals of the corresponding components of benzene in the benzene product mixture to obtain effective peak signals in two spectral bands; wherein, for spectral band 1, obtain the characteristic peak height of toluene, and establish a quantitative analysis model of toluene using multiple regression based on the characteristic peak height of toluene and the mass ratio of toluene to benzene; for spectral band 2, obtain the characteristic peak height of non-aromatic hydrocarbons; establish a quantitative analysis model of non-aromatic hydrocarbons using multiple regression based on the characteristic peak height of non-aromatic hydrocarbons and the mass ratio of non-aromatic hydrocarbons to benzene. S5. For the benzene mixture sample to be tested, the Raman spectrum of the sample is acquired, and after digital signal processing in step S2, Raman peak decomposition is performed on spectral bands 1 and 2 respectively. The spectral signals of the corresponding components of the benzene component in the sample are subtracted to obtain the effective peak signals in the corresponding spectral bands, and then the peak heights of the characteristic peaks of toluene and non-aromatic hydrocarbons are obtained. The peak heights are substituted into the quantitative analysis model of toluene and the quantitative analysis model of non-aromatic hydrocarbons respectively to calculate the mass ratio of toluene and non-aromatic hydrocarbons relative to benzene. Thus, the mass percentages of benzene, toluene and non-aromatic hydrocarbons are derived.

[0008] Furthermore, the baseline correction employs an iterative polynomial baseline correction method.

[0009] Furthermore, the smoothing filter employs a polynomial smoothing filter.

[0010] Furthermore, the peak decomposition of the Raman spectrum after digital signal processing in step S3 includes: The Raman spectral signal of the benzene product mixture consists of three parts: the Raman spectral signal of the known component, the Raman spectral signal of the unknown component, and a noise signal. The measured Raman spectrum of the benzene product mixture is expressed as follows: (1); in, Represents the spectral signals of all known components. This represents the Lorentz peak signal for an unknown component. This indicates the instrument's broadened signal. This indicates the measurement of noise signals.

[0011] Furthermore, the peak decomposition of the Raman spectrum after digital signal processing in step S3 also includes: Through the digital signal processing in step S2, ignoring the interference of noise signals, and assuming that the mixture contains M known components, equation (1) is specifically expressed as: (2); Where v is the Raman shift; the Lorentz peak signal of the Raman spectrum is described by the sum of N Lorentz functions, M is the number of component species, and the k-th Lorentz function is... The center, half-height half-width (HFHM), and area are respectively , and ; This represents the parameter vector of the k-th Lorentz function, i.e. , This represents the parameter vector of all N Lorentz functions, i.e. Gaussian function normalized by area This represents the instrument broadened signal, used to characterize the Gaussian function. The parameters for half-width are ; Represents the spectrum of the j-th known component. This is the weighting coefficient for that component; Based on equation (2), the Raman spectra of the benzene product mixture are separated into peak signals. For each Raman spectrum of the mixture, the following objective function is minimized: (3); in, This indicates the measured Raman spectrum. This represents the estimated values ​​of the spectral model; the parameters to be optimized include... Weight vector of known components Parameter vector of the Lorentz spectral peak signal and the half-width parameter of the instrument broadening function ; This represents the Raman shift interval for separating the spectral peak signals.

[0012] Furthermore, the spectral peak decomposition is divided according to the following wavenumber range: spectral band 1 is 765-900 cm⁻¹. -1 Used to obtain the characteristic peak height of toluene; spectral band 2 is 1390-1500 cm⁻¹. -1 Used to obtain the characteristic peak heights of non-aromatic hydrocarbons.

[0013] The beneficial effects of this invention are as follows: For liquid organic products, compared with online gas chromatographs, online Raman spectrometers offer faster detection speeds (Raman spectrometer detection cycle ≤ 1 min), require no on-site maintenance, provide good repeatability, and exhibit significant dynamic change trends. Furthermore, for high-purity benzene products, the detection limit of the online Raman spectrometer can reach 0.01% (i.e., 100 ppm), meeting process detection requirements and providing a new detection method for the online detection of constant or trace amounts in chemical production processes. Moreover, to improve the detection accuracy and reduce the detection limit of trace impurities, this invention introduces a peak decomposition technique based on the Lorentz-Gaussian convolution model, significantly reducing the interference of benzene peaks on impurity peaks. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of the present invention; Figure 2 This is the Raman spectrum of the benzene product liquid after pretreatment according to the present invention (P1 and P2 are the Raman characteristic peaks of impurities toluene and NA). Figure 3 This is a schematic diagram of signal separation in Raman band 1# of Samp#10 of the present invention; Figure 4 This is a schematic diagram of signal separation in Raman band 1# of Samp#02 of the present invention; Figure 5 This is the Raman spectrum (segment 1#) of all samples after deducting the fitted component of the benzene component in this invention. Figure 6 This is the Raman spectrum (segment 2#) of all samples after deducting the fitted component of the benzene component in this invention. Detailed Implementation

[0016] To enable those skilled in the art to better understand the technical solutions in the embodiments of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art should fall within the protection scope of the embodiments of this application.

[0017] like Figure 1 As shown in the figure, this embodiment of the invention provides an online detection method for trace impurities in benzene products, including the following steps: S1. Under the same measurement conditions, use an online Raman spectrometer to collect the original Raman spectra of a mixture of benzene products with known mass percentages (i.e., a liquid mixture containing benzene, toluene, and non-aromatic hydrocarbons) and the pure benzene component.

[0018] S2. Perform digital signal processing on the original Raman spectrum, including fluorescence background spectrum removal, smoothing filtering, and baseline correction.

[0019] S3. Perform peak decomposition on the Raman spectrum after digital signal processing to extract the Raman spectral signals of benzene, toluene and non-aromatic components from the spectrum of the benzene product mixture; S4. Subtract the spectral signals of the corresponding components of benzene in the benzene product mixture to obtain effective peak signals in two spectral bands; wherein, for spectral band 1, obtain the characteristic peak height of toluene, and establish a quantitative analysis model of toluene using multiple regression based on the characteristic peak height of toluene and the mass ratio of toluene to benzene; for spectral band 2, obtain the characteristic peak height of non-aromatic hydrocarbons; establish a quantitative analysis model of non-aromatic hydrocarbons using multiple regression based on the characteristic peak height of non-aromatic hydrocarbons and the mass ratio of non-aromatic hydrocarbons to benzene.

[0020] S5. For the benzene mixture sample to be tested, the Raman spectrum of the sample is acquired, and after digital signal processing in step S2, Raman peak decomposition is performed on spectral bands 1 and 2 respectively. The spectral signals of the corresponding components of the benzene component in the sample are subtracted to obtain the effective peak signals in the corresponding spectral bands, and then the peak heights of the characteristic peaks of toluene and non-aromatic hydrocarbons are obtained. The peak heights are substituted into the quantitative analysis model of toluene and the quantitative analysis model of non-aromatic hydrocarbons respectively to calculate the mass ratio of toluene and non-aromatic hydrocarbons relative to benzene. Thus, the mass percentages of benzene, toluene and non-aromatic hydrocarbons are derived.

[0021] Preferably, the smoothing filter employs a polynomial smoothing filter, and the baseline correction employs an iterative polynomial baseline correction method.

[0022] Preferably, the spectral peak decomposition is divided according to the following wavenumber range: spectral band 1 is 765-900 cm⁻¹. -1 Used to obtain the characteristic peak height of toluene; spectral band 2 is 1390-1500 cm⁻¹. -1 Used to obtain the characteristic peak heights of non-aromatic hydrocarbons.

[0023] As a preferred embodiment, in order to improve the detection accuracy of trace impurities and reduce the detection limit, the present invention designs a peak decomposition technique based on the Lorentz Gaussian convolution model through the following embodiments, so as to significantly reduce the interference of benzene peak on impurity peaks.

[0024] In principle, whether it is a pure component or a mixture, the peak signal of its Raman spectrum exhibits a Lorentz line shape. That is, a single peak in the Raman spectrum can theoretically be approximated by the following Lorentz function: (1); Where v represents the Raman shift, c is the center of the Lorentz function, and w L Let HWHM be the half-width at half-maximum (HWHM) of the Lorentz function, and S be the area of ​​the Lorentz function across the entire spectrum.

[0025] However, in actual measurements, due to the influence of instrument-broadened signals, the measured Raman spectrum peak signals no longer possess the standard Lorentz function line shape. The actually measured Raman spectrum signal can be described as the convolution of the Lorentz peak signal and the instrument-broadened signal. In practical applications, the factors causing instrument broadening are often multifaceted, such as the characteristics of optical devices like lasers, spectrometers, and Raman probes. Under the combined influence of various instrument broadening factors, the comprehensive effect of the broadened signals can be approximated by a Gaussian function. Therefore, the measured Raman spectrum signal can be approximated by the result of the convolution of the original Lorentz peak signal and a Gaussian function.

[0026] This invention uses an area-normalized Gaussian function to represent the instrument broadening signal, with the function form as follows: (2); in, This represents the half-height and half-width (HWHM) of the Gaussian function.

[0027] make Equation (2) can be expressed as: (3); A single measured Raman spectrum typically contains multiple Raman peaks. After spectral preprocessing to separate the background baseline, the measured Raman spectral signal can be approximated by the convolution of several Lorentz peak signals and the instrument broadened signal, plus the measurement noise (referred to as the Raman spectral Voigt convolution model), which takes the following form: (4); Among them, L k This represents a single Lorentz peak, G represents instrument broadening, N represents measurement noise, and "*" represents convolution operation.

[0028] Substituting equations (1) and (3) into equation (4), and ignoring noise signals, we can obtain the Voigt convolution model of the measured Raman spectrum signal, which is expressed as follows: (5); Where v is the Raman shift; the Lorentz peak signal of the Raman spectrum is described by the sum of N Lorentz functions, M is the number of component species, and the k-th Lorentz function is... The center, HFHM and area are respectively , and . This represents the parameter vector of the k-th Lorentz function, i.e. , This represents the parameter vector of all N Lorentz functions, i.e. Gaussian function normalized by area This represents the instrument broadened signal, used to characterize the Gaussian function. The parameters for half-width are .

[0029] In a preferred embodiment, the peak decomposition of the Raman spectrum after digital signal processing in step S3 includes: The Raman spectral signal of the benzene product mixture consists of three parts: the Raman spectral signal of the known component, the Raman spectral signal of the unknown component, and a noise signal. The measured Raman spectrum of the benzene product mixture is expressed as follows: (6); in, Represents the spectral signals of all known components. This represents the Lorentz peak signal for an unknown component. This indicates the instrument's broadened signal. This indicates the measurement of noise signals.

[0030] Through the digital signal processing in step S2, ignoring the interference of noise signals, and assuming that the mixture contains M known components, equation (6) is specifically expressed as: (7); in, Represents the spectrum of the j-th known component. This is the weighting coefficient for this component.

[0031] Based on equation (7), the Raman spectra of the benzene product mixture are separated into peak signals. For each Raman spectrum of the mixture, the following objective function is minimized: (8); in, This indicates the measured Raman spectrum. This represents the estimated values ​​of the spectral model; the parameters to be optimized include... Weight vector of known components Parameter vector of the Lorentz spectral peak signal and the half-width parameter of the instrument broadening function ; This represents the Raman shift interval for separating the spectral peak signals.

[0032] As a preferred embodiment, the following experimental verification was conducted based on the method of the present invention, including the following steps: (1) The percentage content of benzene and toluene in 54 benzene mixture samples with known concentrations was determined by manual sampling laboratory chromatography.

[0033] (2) Raman spectra of 54 samples were measured using an Ocean Optics QE65000 grating spectrometer with an optical resolution of [missing value]. -1 A laser with a center wavelength of 532 nm was used as the excitation source, and the detection range was 0-4000 cm⁻¹. -1 For all samples, measurements were taken three times consecutively, and the average of the three measurements was taken as the Raman spectrum of that sample. The integration time for each measurement was 30 seconds.

[0034] (3) To avoid the influence of interference factors such as measurement noise, background fluorescence, and changes in the intensity of the excitation light source, the measured Raman spectrum is preprocessed, mainly including: Smoothing Filtering: A polynomial smoothing filtering method is used, with the filter window width set to 150 cm. -1 .

[0035] Baseline correction: An iterative polynomial baseline correction method is employed, with the polynomial degree set to 1. The parameters of the linear baseline are iteratively optimized to obtain a fitted baseline. This fitted baseline is then subtracted from the Raman spectrum to eliminate fluorescence background interference.

[0036] (4) The spectrum after preprocessing is as follows Figure 2 As shown, for the Raman spectra of 54 samples, the first 20 samples were selected as modeling samples, and the last 34 samples were selected as validation samples. For spectral band 1: 765-900 cm⁻¹ -1 ; Obtain the Raman spectrum after subtracting the fitted component of benzene. Based on the characteristic peaks of the modeled sample, i.e., spectral band 1: 765-900 cm⁻¹. -1 The peak height at the highest point, combined with a multiple regression algorithm, is used to establish a quantitative analysis model for the mass ratio of impurity toluene to benzene.

[0037] For the 34 test samples, Raman peak separation was performed on spectral bands 1 and 2 respectively. A comparison of peak decomposition in spectral band 1 for sample #10 (sample number 10) and sample #02 (sample number 2) is shown below. Figure 3 and Figure 4As shown, due to the relatively large amount of impurities (i.e., toluene) artificially added to Samp#02, the decomposition effect corresponding to spectral segment 1 is significant, meaning that the spectral component of toluene is relatively higher than that of Samp#10. This indicates that by separating the benzene component spectrum and the impurity component spectrum through peak decomposition and subtracting the fitted component of the benzene component, the effective peak signal within the corresponding spectral segment of toluene can be obtained, thereby obtaining the peak height of the characteristic peak of toluene. Similarly, for non-aromatic hydrocarbons (NA), it is necessary to select appropriate spectral segments for characteristic description, such as... Figure 5 and Figure 6 As shown, for the spectrum of low-content toluene, spectral band 1 needs to be selected as the characteristic spectral band for quantitative model establishment, while the characteristic spectral band of NA needs to be described by spectral band 2, thus establishing the quantitative model corresponding to NA. By substituting the characteristic peaks in the characteristic spectral bands of toluene and NA into the above multiple regression model, the mass ratio of impurity toluene and NA relative to benzene can be calculated. Furthermore, the mass percentages of benzene, toluene, and NA can be calculated.

[0038] Method validation results: Table 1: Evaluation of Raman Quantitative Analysis Model for Benzene Product Composition Table 2: Raman spectral values ​​and corresponding chromatographic values ​​of each component in the test samples Based on modeling samples with known impurity content, a mathematical model was established between the effective peak height of impurities and their relative concentration (i.e., the mass ratio of impurities to benzene content). For any sample, the corresponding Raman spectrum can be acquired in real time using an online Raman analyzer, and the mass percentage of toluene and NA can be calculated using the mathematical model. The experimental results of the model's quantitative analysis are shown in Table 1, demonstrating that the detection limit of Raman spectroscopy for trace impurities in benzene product liquid can reach 0.006% (i.e., 60 ppm), which meets the process detection requirements (the mass percentage of each component in all 34 test samples is shown in Table 2).

[0039] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only.

[0040] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for online detection of trace impurities in benzene products, characterized in that, Includes the following steps: S1. Under the same measurement conditions, use an online Raman spectrometer to collect the original Raman spectra of a benzene product mixture with a known mass percentage and a pure benzene component, wherein the benzene product mixture is a liquid mixture containing benzene, toluene and non-aromatic hydrocarbons; S2. Perform digital signal processing on the original Raman spectrum, including fluorescence background spectrum removal, smoothing filtering, and baseline correction. S3. Perform peak decomposition on the Raman spectrum after digital signal processing to extract the Raman spectral signals of benzene, toluene and non-aromatic components from the spectrum of the benzene product mixture; S4. Subtract the spectral signals of the corresponding components of benzene in the spectrum of the benzene product mixture to obtain the effective peak signals in two spectral bands; wherein, for spectral band 1, obtain the characteristic peak height of toluene, and establish a quantitative analysis model of toluene using multiple regression technology based on the characteristic peak height of toluene and the mass ratio of toluene to benzene; for spectral band 2, obtain the characteristic peak height of non-aromatic hydrocarbons, and establish a quantitative analysis model of non-aromatic hydrocarbons using multiple regression technology based on the characteristic peak height of non-aromatic hydrocarbons and the mass ratio of non-aromatic hydrocarbons to benzene. S5. For the benzene mixture sample to be tested, the Raman spectrum of the sample is acquired, and after digital signal processing in step S2, Raman peak decomposition is performed on spectral bands 1 and 2 respectively. The spectral signals of the corresponding components of the benzene component in the sample are subtracted to obtain the effective peak signals in the corresponding spectral bands, and then the peak heights of the characteristic peaks of toluene and non-aromatic hydrocarbons are obtained. The peak heights are substituted into the quantitative analysis model of toluene and the quantitative analysis model of non-aromatic hydrocarbons respectively to calculate the mass ratio of toluene and non-aromatic hydrocarbons relative to benzene. Thus, the mass percentages of benzene, toluene and non-aromatic hydrocarbons are derived.

2. The method according to claim 1, characterized in that, The baseline correction employs an iterative polynomial baseline correction method.

3. The method according to claim 1, characterized in that, The smoothing filter used is a polynomial smoothing filter.

4. The method according to claim 1, characterized in that, The peak decomposition of the Raman spectrum after digital signal processing in step S3 includes: The Raman spectral signal of the benzene product mixture consists of three parts: the Raman spectral signal of the known component, the Raman spectral signal of the unknown component, and a noise signal. The measured Raman spectrum of the benzene product mixture is expressed as follows: （1）； in, Represents the spectral signals of all known components. This represents the Lorentz peak signal for an unknown component. This indicates the instrument's broadened signal. This indicates the measurement of noise signals.

5. The method according to claim 4, characterized in that, The peak decomposition of the Raman spectrum after digital signal processing in step S3 further includes: Through the digital signal processing in step S2, ignoring the interference of noise signals, and assuming that the mixture contains M known components, equation (1) is specifically expressed as: （2）； Where v is the Raman shift; the sum of N Lorentz functions describes the Lorentz peak signal of the Raman spectrum, M is the number of component species; the k-th Lorentz function... The center, half-height and half-width, and area are respectively , and ; This represents the parameter vector of the k-th Lorentz function, i.e. ; This represents the parameter vector of all N Lorentz functions, i.e. Gaussian function normalized by area This represents the instrument broadened signal, used to characterize the Gaussian function. The parameters for half-width are ; Represents the spectrum of the j-th known component. This is the weighting coefficient for that component; Based on equation (2), the Raman spectra of the benzene product mixture are separated into peak signals. For each Raman spectrum of the mixture, the following objective function is minimized: （3）； in, This indicates the measured Raman spectrum. This represents the estimated values ​​of the spectral model; the parameters to be optimized include... Weight vector of known components Parameter vector of the Lorentz spectral peak signal and the half-width parameter of the instrument broadening function ; This represents the Raman shift interval for separating the spectral peak signals.

6. The method according to claim 1, characterized in that, The spectral peak decomposition is divided according to the following wavenumber range: spectral band 1 is 765-900 cm⁻¹. -1 Used to obtain the characteristic peak height of toluene; spectral band 2 is 1390-1500 cm⁻¹. -1 Used to obtain the characteristic peak heights of non-aromatic hydrocarbons.

Images