Rapid screening method for identifying attributes of crude oil and fuel oil
By constructing an infrared fingerprint database of crude oil and fuel oil and using CLUSTER cluster analysis method, the problem of difficult to distinguish crude oil from fuel oil in the existing technology is solved, and rapid screening and effective supervision of crude oil and fuel oil attributes are achieved.
Patent Information
- Application Number
- CN202411501496.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2025-02-11
AI Technical Summary
The existing technology is difficult to effectively distinguish between crude oil and fuel oil, which makes it difficult to identify fake goods, affecting regulatory capabilities and customs clearance efficiency.
Data were collected through Fourier transform infrared spectroscopy and attenuation total reflection technology, combined with stoichiometrics and characteristic chemical bond infrared absorption spectroscopy, and infrared fingerprint database of crude oil and fuel oil was constructed, and attribute identification was performed using CLUSTER clustering analysis method.
It has achieved rapid screening of crude oil and fuel oil attributes, can effectively identify crude oil and fuel oil, and improves regulatory capabilities and customs clearance efficiency.
Smart Images
Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of petroleum product detection, and in particular, relates to a rapid screening method for identifying properties of crude oil and fuel oil. Background Art
[0002] Crude oil has obvious advantages in production technology conditions and production costs as a refining raw material. However, due to quota restrictions on crude oil imports, the crude oil import quotas of some refineries cannot meet demand. Crude oil is often falsely reported as fuel oil or other imported products to evade supervision. In the import and export tariff commodities and item notes, the distinction between crude oil and fuel oil is mainly focused on the source and use. It is difficult to directly identify the commodity classification attributes based on the test results. Therefore, it restricts the supervision ability and customs clearance time of imported refining raw materials. Not only can it not effectively distinguish falsely reported goods, but it also affects the customs clearance efficiency of normal trade. Since refining raw materials usually have large customs clearance volumes and high transportation costs, customs clearance efficiency has had an adverse impact on the production of related companies and has attracted concern from many parties.
[0003] Therefore, it is urgent to establish a rapid screening method for identifying the properties of such commodities in order to achieve effective and rapid supervision of related products. Summary of the invention
[0004] The purpose of the present invention is to provide a rapid screening method for identifying the properties of crude oil and fuel oil, so as to make up for the deficiencies of the prior art.
[0005] To achieve the above object, the present invention is achieved through the following technical solutions:
[0006] The present invention uses Fourier transform infrared spectroscopy and attenuated total reflection as data acquisition means, combines chemometrics and characteristic chemical bond infrared absorption spectra, extracts 23 infrared absorption spectrum characteristics and 3 background ranges, and constructs a crude oil and fuel oil infrared fingerprint database with infrared absorption spectrum characteristics as fingerprint characteristics; after standardizing and converting the infrared fingerprint data, on the basis of obtaining a similarity matrix, CLUSTER cluster analysis is used to perform multivariate correlation analysis on the infrared fingerprint maps of crude oil and fuel oil and the infrared fingerprint database, and a method for identifying the commodity attributes of crude oil and fuel oil based on CLUSTER cluster analysis is realized.
[0007] A rapid screening method for identifying crude oil and fuel oil properties comprises the following steps:
[0008] (1) Select a variety of known crude oils and fuel oils and collect infrared spectrum information;
[0009] (2) based on the collected infrared spectrum data, the absorption characteristics in the mid-infrared wavelength range are screened as candidate characteristic absorption peaks as candidate characteristic values;
[0010] (3) constructing an infrared fingerprint database based on the candidate feature values, standardizing the infrared fingerprint database, and then calculating the similarity matrix;
[0011] (4) Based on the similarity of fingerprint data of different oil products, CLUSTER clustering analysis is used to establish a CLUSTER analysis model to intuitively reflect the similarity of different oil products;
[0012] (5) Perform infrared spectroscopy on the unknown oil sample to be tested to obtain an infrared absorption spectrum, and use the CLUSTER analysis model obtained in step (4) to analyze and obtain a screening result.
[0013] Furthermore, the infrared spectrum detection conditions in step (1) and step (5) are as follows: the infrared absorption spectrum of the sample is detected by Fourier transform infrared spectrometer; in the instrument parameter setting, the resolution is 2 cm -1 , wave number measurement range is 650~4000cm -1 The scanning number is 16 times; the instrument detection wavelength is calibrated by standard wavelength polystyrene film; the sample preparation method is attenuated total reflection (ATR) method, the refractive crystal is ZnSe, and the refractive angle is 45°.
[0014] Furthermore, in step (2), the candidate feature values are as shown in Table 1:
[0015] Table 1 Absorption peaks as alternative characteristic values
[0016]
[0017]
[0018] Furthermore, in step (2), since the background absorption intensities of different oil products in different wavenumber ranges are different, in order to reduce the source of error between samples, the average background absorption intensities of three wavenumber ranges are selected as the candidate characteristic value Table 2 based on the differences in the background absorption intensities of coal tar and fuel oil in different wavenumber ranges:
[0019] Table 2 Background ranges for alternative feature values
[0020]
[0021] Furthermore, in step (2), the candidate feature values are processed by using a principal component analysis method to group the contributions of the candidate feature values to sample classification, and to eliminate similar candidate feature values.
[0022] Compared with the prior art, the present invention has the following beneficial effects:
[0023] The present invention uses infrared spectroscopy technology and crude oil and fuel oil fingerprint database as the basis to establish a fast and effective crude oil and fuel oil attribute identification screening method. And through experimental verification, the method can effectively identify crude oil and normal fuel oil, and can be used for rapid screening of crude oil and fuel oil commodity attribute identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 This is the infrared absorption spectrum of crude oil (Hungo).
[0025] Figure 2 These are three typical infrared absorption spectra of fuel oils. The blue spectrum is from a fuel oil from a Chinese refinery; the green spectrum is from fuel oil originating from Russia; and the red spectrum is from fuel oil imported from Malaysia.
[0026] Figure 3 It is the data processing flow of the chemometric method for infrared fingerprint data.
[0027] Figure 4 It is the result of database CLUSTER analysis.
[0028] Figure 5 It is a CLUSTER cluster analysis of unknown samples. DETAILED DESCRIPTION
[0029] The technical solution of the present invention is further described below in conjunction with embodiments.
[0030] Example 1
[0031] In this embodiment, a Fourier transform infrared spectrometer (Necolet 380) was used to detect the infrared absorption spectrum of the sample. In the instrument parameter setting, the resolution was 2 cm -1 , wave number measurement range is 650~4000cm -1 , the number of scans was 16 times. The instrument detection wavelength was calibrated by a standard wavelength polystyrene film. The sample preparation method was the attenuated total reflection (ATR) method, the refractive crystal was ZnSe, and the refractive angle was 45°. Before each test, the background was collected and the background value was subtracted. Figure 1 This is the infrared absorption spectrum of crude oil (Hungo) collected. Figure 2 These are three typical infrared absorption spectra of fuel oil collected.
[0032] This embodiment provides a method for analyzing infrared fingerprint data of a known oil product, comprising the following steps:
[0033] 1. Selection of fingerprint database samples
[0034] The crude oil fingerprint database samples are derived from the normal import of crude oil, and a total of 48 crude oil varieties are collected, including 27 countries from Africa, the Middle East, America, Europe, Southeast Asia and Oceania. The crude oil varieties and origins are shown in Table 3.
[0035] Table 3 Crude oil varieties and origins in the infrared spectral fingerprint database
[0036]
[0037]
[0038] The samples of the fuel oil fingerprint database come from fuel oil that enters the country normally and the domestic market. A total of 13 types of fuel oil are collected, and their sources include China, Russia, Malaysia, Indonesia, Australia, etc. The sources and identification of the samples in the fingerprint database are shown in Table 4.
[0039] Table 4 Fuel oil sources in the infrared spectral fingerprint database
[0040]
[0041] 2. Alternative feature value and feature quantity selection method
[0042] According to the instrument parameter settings, each sample is between 650 and 4000 cm -1 There are 3481 data points within the wavenumber measurement range. If these data points are not selected, it will not only increase the difficulty of data processing, but also cover up the differences between the infrared absorption characteristic spectra of different crude oils due to accidental measurement errors. Therefore, based on the absorption characteristics of crude oil and fuel oil in the mid-infrared wavelength range, 23 alternative characteristic absorption peaks were determined as alternative characteristic values (see Table 1). In addition, since the background absorption intensity of different oil products in different wavenumber ranges is different, in order to reduce the source of error between samples, the average background absorption intensity of 3 wavenumber ranges was selected as the alternative characteristic value (see Table 2) according to the difference in background absorption intensity of coal tar and fuel oil in different wavenumber ranges.
[0043] Table 1 Absorption peaks as alternative characteristic values
[0044]
[0045] Table 2 Background ranges for alternative feature values
[0046]
[0047] Principal component analysis (PCA) was used to group the contributions of the above 26 features to sample classification. In order to compress the data volume and avoid excessive duplication of features affecting the statistical results, this study eliminated the approximate candidate feature values based on the PCA analysis results, the test sensitivity and anti-interference ability of the feature values.
[0048] 3. Establishment of infrared fingerprint database and chemometric method
[0049] Based on the extracted characteristic quantities, an infrared fingerprint database of crude oil, fuel oil and reference samples was constructed. After the fingerprint data was standardized, the similarity matrix was calculated, and cluster analysis was used to measure the fingerprint data of different oil products in order to intuitively reflect the similarity of different oil products. The data processing flow of the chemometric method of infrared fingerprint data is shown in Figure 3 .
[0050] Based on the extraction method of the infrared fingerprint database of crude oil and fuel oil, the proposed measurement method was used to qualitatively identify unknown oil samples, verifying the applicability of the infrared fingerprint database and the measurement method.
[0051] 4. Infrared fingerprint feature database and fingerprint recognition method
[0052] Table 5 Contribution of feature quantities to the composite dimensions in the principal component analysis (PCA) of the five-dimensional reconstruction
[0053]
[0054]
[0055] 5. Establish infrared fingerprint database
[0056] The infrared fingerprint data of each sample was extracted using the selected feature values, and the infrared fingerprint data was standardized to construct a database (Tables 6 and 7).
[0057]
[0058]
[0059]
[0060]
[0061]
[0062]
[0063] 6. Multivariate statistical analysis results
[0064] The characteristic peaks were extracted from the infrared spectrum data of crude oil and fuel oil in the database, and the absorbance of the characteristic peaks was normalized. The similarity between each group of data was calculated using the group average method with the Mahalanobis distance of the data after normalization to obtain the similarity matrix. The results of CLUSTER clustering analysis on the infrared spectrum data of crude oil and fuel oil in the database are shown in Figure 4 .
[0065] According to the analysis of the CLUSTER cluster analysis results, the similarity between fuel oils is higher than 92%, but the similarity between crude oils is large, and the similarity between some types of crude oil and other crude oils is less than 85%. In particular, the two heavy crude oils, Dar crude oil from the Republic of Sudan and Mars crude oil from Malaysia, have a similarity of nearly 92% with fuel oil, but only 89% with other crude oils. It can be seen that some types of heavy crude oil have a high similarity with fuel oil, and the CLUSTER cluster analysis using only infrared spectral data is easy to confuse with fuel oil. According to the crude oil types in this study, the possible misjudgment ratio of heavy crude oil is about 4%.
[0066] The above analysis results show that, through the infrared fingerprint database and multivariate statistical analysis method, only the CLUSTER cluster analysis of infrared spectrum data is used, heavy crude oil may be misjudged as crude oil, but the misjudgment ratio of different varieties of crude oil is about 4%. It should be noted that for heavy crude oil, which accounts for about 4% of the total crude oil varieties, this method may misjudge. However, since the CLUSTER cluster analysis requires little computing power, this method can be used to screen the commodity attributes of oil products when computing power is insufficient.
[0067] Embodiment 2:
[0068] This embodiment uses the detection method and clustering model provided in Embodiment 1 to detect unknown oil products.
[0069] Four samples of unknown crude oil and four samples of fuel oil of unknown origin were selected as blind samples (see Table 8). Based on the infrared fingerprint database established in this study and the proposed measurement method, qualitative discriminant analysis was performed on the blind samples to verify the applicability of the infrared fingerprint database and measurement method. The results of CLUSTER cluster analysis are shown in Figure 5 .
[0070] Table 8 Blind sample sources
[0071] Sample identification Actual species Unknown1 Crude oil (Arabian Heavy) Unknow2 Crude oil (Basrah light) Unknow3 Crude oil(Merey) Unknow4 Crude Oil (SAXI) Unknow5 Fuel oil 1 Unknow6 Fuel oil 2 Unknow7 Fuel oil 3 Unknow8 Fuel oil 4
[0072] Depend on Figure 5It can be seen that the similarity between the four unknown varieties of crude oil and the crude oil samples in the database is higher than 95%, and the similarity with the fuel oil in the database is lower than 90%; the similarity between the four unknown sources of fuel oil and the fuel oil samples in the database is close to 95%, and the similarity with the crude oil in the database is lower than 90%. This shows that the CLUSTER clustering analysis of infrared spectral data can identify the above eight different varieties of crude oil and fuel oil from different sources.
[0073] The above analysis shows that in the result verification, all unknown crude oil and fuel oil samples of unknown origin can be correctly distinguished. Among them, CLUSTER cluster analysis can also infer the source of crude oil to some extent.
[0074] Finally, although this specification is described according to implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.
Claims
1. A rapid screening method for identifying crude oil and fuel oil properties, characterized in that: The following steps are involved: (1) Select a variety of known crude oils and fuel oils and collect infrared spectrum information; (2) based on the collected infrared spectrum data, the absorption characteristics in the mid-infrared wavelength range are screened as candidate characteristic absorption peaks as candidate characteristic values; (3) constructing an infrared fingerprint database based on the candidate feature values, standardizing the infrared fingerprint database, and then calculating the similarity matrix; (4) Based on the similarity of fingerprint data of different oil products, CLUSTER clustering analysis is used to establish a CLUSTER analysis model to intuitively reflect the similarity of different oil products; (5) Perform infrared spectroscopy on the unknown oil sample to be tested to obtain an infrared absorption spectrum, and use the CLUSTER analysis model obtained in step (4) to analyze and obtain a screening result.
2. The rapid screening method for identifying crude oil and fuel oil properties according to claim 1, characterized in that: The infrared spectrum detection conditions in step (1) and step (5) are as follows: the infrared absorption spectrum of the sample is detected by Fourier transform infrared spectrometer; in the instrument parameter setting, the resolution is 2 cm -1 , wave number measurement range is 650~4000cm -1 The scanning number is 16 times; the instrument detection wavelength is calibrated by standard wavelength polystyrene film; the sample preparation method is attenuated total reflection method, the refractive crystal is ZnSe, and the refractive angle is 45°.
3. The rapid screening method for identifying crude oil and fuel oil properties according to claim 1, characterized in that: In the step (2), the candidate feature values are shown in Table 1: Table 1 Absorption peaks as alternative characteristic values 4. The rapid screening method for identifying crude oil and fuel oil properties according to claim 1, characterized in that: In the step (2), since different oil products have different background absorption intensities in different wavenumber ranges, in order to reduce the source of error between samples, the average background absorption intensities of three wavenumber ranges are selected as the candidate characteristic value table 2 according to the difference in background absorption intensities of coal tar and fuel oil in different wavenumber ranges: Table 2 Background ranges for alternative feature values 5. The rapid screening method for identifying crude oil and fuel oil properties according to claim 1, characterized in that: In the step (2), the candidate feature values are processed by using a principal component analysis method to group the contributions of the candidate feature values to sample classification, and to eliminate similar candidate feature values.