CN109724938B - Method for predicting properties of lubricating oil base oil by near infrared spectrum - Google Patents
Method for predicting properties of lubricating oil base oil by near infrared spectrum Download PDFInfo
- Publication number
- CN109724938B CN109724938B CN201711042673.3A CN201711042673A CN109724938B CN 109724938 B CN109724938 B CN 109724938B CN 201711042673 A CN201711042673 A CN 201711042673A CN 109724938 B CN109724938 B CN 109724938B
- Authority
- CN
- China
- Prior art keywords
- sample
- spectrum
- sub
- near infrared
- fitting
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 238000002329 infrared spectrum Methods 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 70
- 239000002199 base oil Substances 0.000 title claims abstract description 68
- 239000010687 lubricating oil Substances 0.000 title claims abstract description 38
- 238000001228 spectrum Methods 0.000 claims abstract description 100
- 230000003595 spectral effect Effects 0.000 claims abstract description 26
- 238000002835 absorbance Methods 0.000 claims abstract description 18
- 238000010561 standard procedure Methods 0.000 claims abstract description 15
- 239000000203 mixture Substances 0.000 claims description 17
- 239000000314 lubricant Substances 0.000 claims description 11
- 150000001924 cycloalkanes Chemical class 0.000 claims description 3
- 230000004069 differentiation Effects 0.000 claims description 3
- 239000012188 paraffin wax Substances 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims description 2
- 238000004497 NIR spectroscopy Methods 0.000 claims 1
- 239000003921 oil Substances 0.000 abstract description 5
- 239000000126 substance Substances 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 230000000052 comparative effect Effects 0.000 description 5
- 238000012937 correction Methods 0.000 description 4
- 230000001050 lubricating effect Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000002156 mixing Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000003647 oxidation Effects 0.000 description 1
- 238000007254 oxidation reaction Methods 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000002904 solvent Substances 0.000 description 1
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
A method for predicting the properties of basic oil of lubricating oil by near infrared spectrum includes collecting basic oil samples of lubricating oil, measuring the property data of each sample by standard method, measuring the near infrared spectrum of each sample, performing second-order differential treatment, and taking 7000-4000 cm‑1And (3) the absorbance of the spectrum area corresponds to the property data of the sample measured by a standard method, a near infrared spectrum database is established, a plurality of sub-libraries are randomly selected from the near infrared spectrum database, and the property data predicted value of the base oil sample of the lubricating oil to be measured is obtained by adding the average value of the property data values of the sample to be measured obtained by spectral fitting of each sub-library and the property data predicted value obtained by spectral fitting of the whole library according to a proper proportion.
Description
Technical Field
The invention relates to a method for predicting oil properties by using near infrared spectrum, in particular to a method for predicting the properties of lubricating oil base oil by using near infrared spectrum.
Background
The chemical composition of the base oil of the lubricating oil is closely related to various properties such as oxidation stability, viscosity index, pour point and the like, and the determination of the chemical composition of the base oil not only can control the production quality of the base oil, but also has important significance for the development and the preparation of the lubricating oil. The existing method for determining the detailed chemical composition of the lubricating oil base oil adopts an MS method, and the method has long analysis time and cannot meet the rapid analysis requirements of production and scientific research. The viscosity index is an important parameter for representing the viscosity-temperature performance of the lubricating oil and is also one of important quality indexes in the API classification. At present, the viscosity index is calculated by measuring the viscosity at 40 ℃ and 100 ℃, and the method is complicated. Wangbin et al (near infrared analysis method for determining chemical composition of lube base oil), petrochemical engineering, 2001, 30 (3): 224 and 227, near infrared analysis method for determining lube base oil viscosity index, lube oil, 2001, 16 (6): 53-56) adopt near infrared spectrum combined with partial least square method to establish a rapid analysis method for predicting lube base oil chemical composition and viscosity index. However, the calibration model established by the partial least square method has a certain application range, and if the composition of the sample to be measured is different from the calibration set sample collected by establishing the calibration model, the calibration model needs to be expanded and updated to be applicable. The establishment of the partial least squares quantitative correction model requires operations such as spectrum preprocessing, spectrum interval optimization and the like according to specific application objects, and the establishment of the model is usually completed by trained professionals for the reasons of multiple selected parameters, difficult understanding and mastering of a multivariate correction method and the like, so that the bottleneck problem of restricting the wide popularization and application of the technology is caused, and many projects cannot play the due role of the correction model due to untimely maintenance of the correction model.
CN102374975A discloses a Method for predicting oil property data by using near infrared spectrum, which proposes a Library spectrum Fitting Method (Library spectrum Fitting Method) as a new property prediction Method, based on the near infrared spectrum Library of oil and spectrum Fitting technology, based on the principles of spectrum similarity and property similarity, Fitting the spectrum of an unknown sample to be tested by one or more Spectra in the spectrum Library, and then calculating the property of the sample to be tested according to the property participating in Fitting the spectrum, the chemical essence of which is that the unknown sample can be formed by mixing a group of Library samples according to a certain proportion, therefore, the property to be tested of the unknown sample can be calculated according to the property of the Library samples according to the mixing proportion. However, the method relies on limited sample property data participating in spectrum fitting in a spectrum library to predict the properties of the sample to be measured, thereby influencing the accuracy and stability of prediction.
Disclosure of Invention
The invention aims to provide a method for predicting the properties of lubricating oil base oil by near infrared spectrum, which can improve the prediction accuracy and stability of the property data of a sample to be detected.
The invention provides a method for predicting the properties of lubricating oil base oil by near infrared spectrum, which comprises the following steps:
(1) collecting at least 300 lubricant base oil samples, determining property data for each sample using standard methods,
(2) measuring the near infrared spectrum of each sample, performing second-order differential treatment on the near infrared spectrum, and taking 7000-4000 cm-1The absorbance of the spectrum region corresponds to the property data of the sample measured by a standard method, a near infrared spectrum database is established,
(3) measuring the near infrared spectrum of a lubricating oil base oil sample to be measured, carrying out second-order differentiation, and selecting 7000-4000 cm-1Absorbance of the spectral region;
(4) randomly selecting spectrum data of a plurality of samples from a near-infrared spectrum database, establishing sub-databases of the near-infrared spectrum database, wherein the number of the samples in each sub-database is 50-70% of the number of the samples in the near-infrared spectrum database, establishing n sub-databases of the near-infrared spectrum database, and the samples in each sub-database are different,
(5) (I) for a near infrared spectral database library ApFitting the spectrum of the sample to be tested according to the following method:
a) expressing the spectrum of the sample to be measured according to the formula (I), solving the fitting coefficient of the library spectrum,
wherein x is the spectrum of the sample to be measured, viFor near infrared spectral database division ApThe ith library spectrum is shown in the specification, and k is a spectrum database sub-library ApNumber of spectra in (A)pNumber of samples contained, aiFor spectral database banking ApThe fitting coefficient corresponding to the ith library spectrum,
fitting coefficient aiThe method is solved by adopting a classical non-negative constraint least square method, namely the following objective functions are satisfied:
b) fitting coefficient a obtained by the aboveiAll non-zero fitting coefficients in the data are extracted and normalized according to the formula II to obtain a normalized fitting coefficient bi：
Wherein g is the number of non-zero fitting coefficients,
c) calculating the fitting spectrum of the sample to be measured according to the formula III:
d) calculating the fitting degree of the spectrum of the sample to be measured according to the formula,
wherein x isjIs the absorbance of the jth wavelength point of the spectrum of the sample to be measured,
to fit the absorbance at the jth wavelength point of the spectrum, m is the number of wavelength points of the spectrum,
if s is larger than the set threshold value, obtaining the prediction value of the sub-library property data of the sample to be tested according to the formula,
wherein,
for database partitioning by near infrared spectrum ApFitting to obtain the predicted value of the sub-library property data, wherein q is the property data corresponding to the spectral data participating in fitting,
if s is smaller than the set threshold, not adopting the near infrared spectrum database to calculate the property data of the sample to be measured,
(II) according to the method of the step (I), fitting the spectrum of the sample to be detected by using other (n-1) near infrared spectrum database sub-libraries, wherein t near infrared spectrum database sub-libraries with s larger than a set threshold value are provided, and t is larger than 60% multiplied by n,
(6) using the database-dividing property data predicted value of t samples to be measured, calculating the database-dividing property data predicted value (P) of the samples to be measured according to the formulaSeparate warehouse)：
Wherein,
a near infrared spectrum database sub-library A with s larger than a set threshold value1、A2、…、AtAnd fitting to obtain the predicted value of each sub-library property data of the sample to be tested.
(7) Fitting the near infrared spectrum of the sample to be tested by using all the spectra in the near infrared spectrum database according to the method in the step (I) in the step (5), wherein s is not less than a set threshold value, and calculating the property data corresponding to the spectra participating in fitting to obtain a total library property data predicted value (P) of the sample to be testedWhole storehouse) And then the predicted value of the property data of the sample to be tested is calculated by the formula (viii):
P＝(60％-90％)Pseparate warehouse+(10％-40％)PWhole storehouse ⑧
In the formula, P is a predicted value of the property data of the sample to be tested, and P isSeparate warehouseFor the prediction value, P, of the sub-library property data of the sample to be testedWhole storehouseAnd predicting the property data of the whole library of the sample to be tested.
The method adopts a method for establishing a sub-library for a near-infrared spectrum database, the spectrum of a sample to be tested is fitted by using the library spectrum of the sub-library, the property data of the sub-library spectrum participating in the fitting is further used for calculating to obtain the property data predicted value of the sample to be tested in the sub-library, the average value of the property data predicted values of the sub-libraries is used as the sub-library property data predicted value of the sample to be tested, and the sub-library property data predicted value of the sample to be tested is combined with the whole library property data predicted value to obtain the property data predicted. The method can fully utilize the resources of the existing near-infrared spectrum database, and effectively improve the prediction accuracy of the properties of the sample to be detected.
Detailed Description
The method comprises the steps of establishing a plurality of sub-base of lubricating oil base oil near-infrared spectrum databases, fitting the spectrum of a sample to be tested by the sub-base spectrum to obtain the predicted value of the property data of the sample to be tested of the sub-base, obtaining the predicted value of the sub-base property data of the sample to be tested by the average value of all sub-base prediction results capable of being fitted, combining the predicted value of the whole-base property data of the sample to be tested obtained by the whole-base spectrum fitting, and adding the predicted value and the predicted value according to a proper weight ratio to obtain the predicted value of the property data of the sample to be tested. Compared with CN102374975A, the method can eliminate the limitation of absolutely depending on limited sample data in a lubricating oil base oil spectrum database to a certain extent, and expand the number of samples participating in fitting to a greater extent, thereby improving the prediction accuracy and stability of the lubricating oil base oil sample property.
The lubricant base oil properties that can be predicted by the method of the present invention include at least one of the group composition, viscosity index and pour point of the lubricant base oil
The group composition may be at least one of paraffin, monocycloalkane, dicycloalkane, tricycloalkane, tetracycloalkane and cycloalkane content
The method comprises the steps of (1) collecting representative lubricating oil base oil samples, collecting sample data of at least 300 samples, properly collecting 300-500 samples, and measuring the property data of each sample by using a standard method. The standard method for determining the composition of a lubricating oil base oil sample group is SH/T0659, the standard method for determining the viscosity index is GB/T1995, and the standard method for determining the pour point is GB/T3535.
The lubricating base oil of the present invention comprises a group II base oil, a group III base oil or a group III base oil+Base oil of the group III+The similar base oil is base oil with the viscosity index not less than 130.
The method comprises the step (2) of measuring the near infrared spectrum of each sample, carrying out second-order differential treatment on the near infrared spectrum, and selecting 7000-4000 cm-1The absorbance of the spectral region (as its spectrum) corresponds to the property data of the sample measured by standard methods, and a near infrared spectral database is established. The property data may include the above group composition, viscosity index and pour point, or several of them, if there are other property data, it may be added into the near infrared spectrum database.
And (3) measuring the near infrared spectrum of the base oil sample of the lubricating oil to be measured, performing second-order differentiation, and selecting 7000-4000 cm-1Absorbance of the spectral region (as its spectrum); to facilitate fitting it with the library spectra. And (3) determining the near infrared spectrum of the lubricating oil base oil sample to be determined by the same method as the method for establishing the near infrared spectrum database to determine the near infrared spectrum of the sample.
And (4) randomly selecting spectrum data from the near-infrared spectrum database to establish sub-databases, wherein the spectrum included in each sub-database should cover all types of data in the near-infrared spectrum database as much as possible, and the number of samples included in each sub-database should be 50-70%, preferably 55-70%, of the number of samples in the near-infrared spectrum database. The samples contained in each constructed sub-library are different, namely, no same sub-library exists. The number n of the built sub-databases of the near infrared spectrum database is preferably 50-100.
And (5) fitting the spectrum of the base oil sample of the lubricating oil to be detected by using the library-divided spectrum, and calculating to obtainThe fitting degree s of (2) is compared with a threshold value, and whether the spectrum fitting is complete or not is judged. For a certain near infrared spectrum database, dividing a database ApAnd (b) fitting by using the method of steps a) to d) listed in step (I), and finally obtaining the fitting degree s of the spectrum of the sample to be measured, wherein the specific algorithm of the non-negative-constraint least square method in step b) is disclosed in the literature: L.Lawson and R.J.Hanson, solvent Least Squares reports, Prentice-Hall, Englewood Cliffs, NJ (1974); 160-165.
And s is an index for judging the fitting degree of the spectrum of the sub-library to the sample to be detected, and the larger the value is, the higher the fitting degree is, so that the more accurate the calculated prediction property is. If s is smaller than the set threshold, the spectrum fitting is not complete, that is, the sample to be measured cannot be completely represented by the library spectrum fitting, so that the property data of the sub-library spectrum cannot be accurately predicted.
And (4) calculating the predicted value of the property data of the sub-library with the formula (v) for the sub-library with the s larger than the threshold value.
The method for judging the fitting degree by using the s, the applicable conditions and the property calculation method are also suitable for calculating the property data predicted value of the sample to be measured by using the full-base spectrum fitting.
(5) The method for determining the threshold comprises the following steps: selecting a lubricating oil base oil sample, repeatedly measuring the three-time near infrared spectrum, performing second-order differential treatment on the spectrum measured each time, and taking 7000-4000 cm-1Calculating the pseudo fitting degree (sr) value between every two spectra according to formula (c), taking the maximum sr value, multiplying the maximum sr value by the coefficient 0.75 to obtain the threshold value,
in formula (c), x'jAnd x ″)jIs the absorbance at the j-th wavelength point of the two spectra, and m is the number of wavelength points of the spectra. The wavelength points of the spectrum are the wavelength points of absorbance collected in the spectral interval.
And (II) in the step 5) of the invention is to calculate the spectrum fitting degrees of the samples to be measured of the rest (n-1) built sub-libraries according to the method (I), wherein t sub-libraries with s larger than a set threshold value are provided in total, and t is larger than 60% multiplied by n. And t is more than 60 percent multiplied by n, which is the condition for calculating the property of the sample to be measured by using the sub-library spectrum fitting, and the sub-library property data predicted value of the sample to be measured can be calculated by using the step (6) only if the condition is met.
The step (6) of the invention is to calculate the average value of the database-partitioned property data predicted values of the t samples to be tested which meet the conditions of the steps (5) and (II) by a formula (I) to obtain the database-partitioned property data predicted value of the samples to be tested.
The step (7) of the invention is to calculate the predicted value of the property data of the sample to be tested by a formula (III), PWhole storehouseThe method for calculating the property of the sample to be measured by fitting the spectrum of the sub-library is the same as that of the sub-library in the fitting calculation method of the predicted value of the property data obtained by fitting calculation of all the spectra of the built near-infrared spectrum database on the sample to be measured.
(5) And (II) when t is less than or equal to 60% multiplied by n, the condition of calculating the property of the sample to be detected by using the sub-library spectrum fitting is not met, the sub-library is randomly established according to the method in the step (4), the property of the sample to be detected is calculated by using the sub-library spectrum fitting according to the method in the step (5) until the value of the sub-library of the near infrared spectrum database with s larger than the set threshold value meets t larger than 60% multiplied by n, and the property data predicted value of the sample to be detected is obtained according to the methods in the steps (6) and (7). Preferably, the number of pools, or the number of samples in each pool, is increased when the pools are re-established randomly, or both.
The method is suitable for rapidly predicting the property data of the lubricating oil base oil sample by using the near infrared spectrum, can be used for on-site monitoring of the properties of the lubricating oil base oil, controlling the quality monitoring of the production of the lubricating oil base oil and can be used for blending the lubricating oil.
The present invention is further illustrated by the following examples, but the present invention is not limited thereto.
In the examples and comparative examples, the near infrared spectrum of the lubricant base oil was measured by a Fourier transform near infrared spectrometer (ANTARIS II) manufactured by Thermo corporation, and the spectrum range was 3500 to 10000cm-1Resolution of 8cm-1The scanning times are accumulated for 128 times, and the sample is stabilized for 5min before the spectrum is collected.
Example 1
Group compositions and physical property data of the lubricant base oil are predicted.
(1) Establishing near infrared spectrum database of lubricating oil base oil
423 representative lubricant base oil samples were collected, containing 162 group II base oil samples and 157 group III base oil samples, III+104 base oil-like samples. The group composition data for each sample was determined using the SH/T0659 method, including paraffin, monocycloalkane, dicyclo-cycloalkane, tricyclo-cycloalkane, tetracycloalkane and cycloalkane contents, and the viscosity index and pour point data were determined using GB/T1995 and GB/T3535, respectively.
Measuring the near infrared spectrum of each sample, performing second-order differential treatment on the near infrared spectrum, and selecting 7000-4000 cm-1Absorbance in the spectral range, and the number of the wavelength points of absorbance collected in the spectral interval is 875. 7000 to 4000cm-1The absorbance of the spectrum region corresponds to the family composition, viscosity index and pour point measured by a standard method, and a near infrared spectrum database is established.
(2) Calculating a threshold s of fitnessv
Taking 1 lubricating oil base oil sample as a III-type base oil, repeatedly measuring the near infrared spectrum of the III-type base oil for three times, performing second-order differential treatment, and selecting 7000-4000 cm-1Calculating the pseudo fitting degree (sr) value between every two spectra according to formula (c), taking the maximum sr value, and multiplying the maximum sr value by the coefficient 0.75 to obtain the threshold value sv， sv＝1.82。
(3) Randomly establishing near infrared spectrum database sub-database of lubricating oil base oil
Randomly generating 80 sub-libraries A from a near infrared spectrum database of the lubricating oil base oil1、A2、…、A80The number of samples selected for each sub-bank is 260.
(4) Fitting the spectrum of the sample to be tested by using the sub-library data to obtain the predicted value of the sub-library property data
Taking 1 sample A to be measured of the III-class lubricating oil base oil, measuring the near infrared spectrum of the sample A according to the method same as the method for measuring the near infrared spectrum of the sample in the step (2), and carrying out second order differentialTaking 7000-4000 cm-1Absorbance in the spectral range, banking A against a near infrared spectral databasepCalculating the predicted value of the property data of the sub-library according to the methods a) to d) in the invention (5) and (I), namely performing fitting calculation according to the formulas (i) to (iii) to obtain a fitting spectrum of the sample to be measured, and then calculating the fitting degree s of the spectrum of the sample to be measured according to the formula (i), wherein s is greater than a threshold value s, and s is 6.25vAnd calculating the predicted value of the property data of the sample to be tested according to the formula (v).
Performing spectrum fitting on the other 79 sub-libraries in the 80 sub-libraries according to the method, wherein the fitting degree s of 58 sub-libraries is larger than the threshold value svI.e., t-58.
(5) Predicting property data of a sample to be tested
Obtaining the prediction values of the property data of the to-be-detected sample by 58 sub-libraries according to a formula
Calculating the prediction value P of the sub-library property data of the sample to be detectedSeparate warehouse。
Establishing a near infrared spectrum database in the step (1), performing fitting calculation on the spectrum of the sample A to be measured according to formulas I to III to obtain a fitting spectrum of the sample A to be measured, and then calculating the fitting degree s of the spectrum of the sample to be measured according to a formula II, wherein the s is 4.56 and is larger than a threshold value svCalculating the property data corresponding to the spectrum participating in fitting to obtain the total library property data predicted value P of the sample to be tested according to the formulaWhole storehouseCalculating the predicted value of the property data of the sample to be tested according to the formula (phi), and concretely, calculating the predicted value of the property data of the sample to be tested according to the formula (phi) that P is 70% PSeparate warehouse+30％PWhole storehouseThe group composition, viscosity index and pour point prediction value of the lubricating oil base oil sample A to be tested are obtained and are shown in Table 1.
Comparative example 1
The sample A used in example 1 was used to predict the properties according to CN102374975A, and the results are shown in Table 1.
As can be seen from Table 1, the method of the present invention has smaller prediction deviation than the method of CN102374975A, which indicates that the method of the present invention has higher prediction accuracy.
TABLE 1
Deviation 1: example 1 deviation between predicted values and standard method measurements;
deviation 2: deviation between the CN102374975A method predicted value and the standard method measured value.
Example 2
Taking a lubricating oil base oil II type base oil sample B, measuring the near infrared spectrum thereof according to the method of steps (1), (3) and (5) of the example, randomly establishing 80 sub-libraries, wherein the number of the samples in each sub-library is 260, calculating the property data thereof by using spectral fitting of the sub-libraries, and totally 60 fitting degrees s are larger than a threshold value svT 60 is greater than (80 × 0.6 ═ 48), and predicted values of the property data of the 60 sub-libraries are obtained by the formula (I)Separate warehouseFitting the sample to be tested by using the full-library spectrum, and calculating PWhole storehouseObtaining a predicted value of the property data of the sample B to be tested by the formula (80% P)Separate warehouse+20％PWhole storehouseThe predicted value of the property data of the sample to be tested is calculated, and the result is shown in table 2.
Comparative example 2
The lubricating base oil sample B of example 2 was taken and its property data were predicted by the method of CN102374975A, and the results are shown in Table 2.
TABLE 2
Example 3
Taking a lubricating oil base oil III+Measuring the near infrared spectrum of the base oil-like sample C according to the method of the steps from (1), (3) to (5) in the example 1, randomly establishing 80 sub-libraries, wherein the number of the samples in each sub-library is 260, calculating the property data of the base oil-like sample C by using spectral fitting of the sub-libraries, and totally 55 fitting degrees s are larger than a threshold value svT 55, which is greater than (80 × 0.6) 48, from these 55The predicted value of the property data of each library is obtained by the formula (I)Separate warehouseFitting the sample to be tested by using the full-library spectrum, and calculating PWhole storehouseObtaining a predicted value of the property data of the sample B to be tested by the formula (phi), and concretely, obtaining the predicted value of the property data of the sample B to be tested according to the formula (phi) that P is 70% PSeparate warehouse+30％PWhole storehouseThe predicted value of the property data of the sample to be tested is calculated, and the result is shown in table 3.
Comparative example 3
The lubricating base oil sample C of example 3 was used, and its property data were predicted by the method of CN102374975A, and the results are shown in Table 3.
TABLE 3
Example 4
Taking a lubricating oil base oil II type base oil sample D, measuring the near infrared spectrum thereof according to the method of the steps from (1), (3) to (5) of the example, randomly establishing 80 sub-libraries, wherein the number of the samples in each sub-library is 260, calculating the property data thereof by using spectral fitting of the sub-libraries, and totally 44 fitting degrees s are larger than a threshold value svT is 44 and is less than 48, and the condition of calculating the property of the sample to be measured by using spectral fitting of the sub-library is not satisfied. Repeating the steps 1, 3 and 5, randomly establishing 80 sub-libraries with 290 samples in each sub-library, calculating the property data by spectral fitting of the sub-libraries, and totally 56 fitting degrees s larger than a threshold value svSub-bank of (t)156 is larger than (80 multiplied by 0.6) 48, and the predicted value of the property data of the 56 sub-libraries is obtained by the formula (I)Separate warehouseFitting the sample to be tested by using the full-library spectrum, and calculating PWhole storehouseObtaining a predicted value of the property data of the sample C to be tested by the formula (V), wherein P is 70% PSeparate warehouse+30％PWhole storehouseThe predicted value of the property data of the sample to be tested is calculated, and the result is shown in table 4.
Comparative example 4
The lubricating base oil sample D of example 3 was taken and its property data were predicted according to the method of CN102374975A, and the results are shown in Table 4.
TABLE 4
Claims (8)
1. A method for predicting the properties of a lubricant base oil by near infrared spectroscopy comprises the following steps:
(1) collecting at least 300 lubricant base oil samples, determining property data for each sample using standard methods,
(2) measuring the near infrared spectrum of each sample, performing second-order differential treatment on the near infrared spectrum, and taking 7000-4000 cm-1The absorbance of the spectrum region corresponds to the property data of the sample measured by a standard method, a near infrared spectrum database is established,
(3) measuring the near infrared spectrum of a lubricating oil base oil sample to be measured, carrying out second-order differentiation, and selecting 7000-4000 cm-1Absorbance of the spectral region;
(4) randomly selecting spectrum data of a plurality of samples from a near-infrared spectrum database, establishing sub-databases of the near-infrared spectrum database, wherein the number of the samples in each sub-database is 50-70% of the number of the samples in the near-infrared spectrum database, establishing n sub-databases of the near-infrared spectrum database, and the samples in each sub-database are different,
(5) (I) for a near infrared spectral database library ApFitting the spectrum of the sample to be tested according to the following method:
a) expressing the spectrum of the sample to be measured according to the formula (I), solving the fitting coefficient of the library spectrum,
wherein x is the spectrum of the sample to be measured, viFor near infrared spectral database division ApThe ith library spectrum is shown in the specification, and k is a spectrum database sub-library ApNumber of spectra in (a)iFor spectral database banking ApThe fitting coefficient corresponding to the ith library spectrum,
fitting coefficient aiUsing classical non-negative constraintsThe product of the first and second multiplications satisfies the following objective function:
b) fitting coefficient a obtained by the aboveiAll non-zero fitting coefficients in the data are extracted and normalized according to the formula II to obtain a normalized fitting coefficient bi：
Wherein g is the number of non-zero fitting coefficients,
c) calculating the fitting spectrum of the sample to be measured according to the formula III:
d) calculating the fitting degree of the spectrum of the sample to be measured according to the formula,
wherein x isjIs the absorbance of the jth wavelength point of the spectrum of the sample to be measured,
to fit the absorbance at the jth wavelength point of the spectrum, m is the number of wavelength points of the spectrum,
if s is larger than the set threshold value, obtaining the prediction value of the sub-library property data of the sample to be tested according to the formula,
wherein,
for database partitioning by near infrared spectrum ApThe predicted value of the sub-library property data, q, is obtained by fittingiFor property data corresponding to spectral data participating in the fitting,
if s is smaller than the set threshold, not adopting the near infrared spectrum database to calculate the property data of the sample to be measured,
(II) according to the method of the step (I), fitting the spectrum of the sample to be detected by using other (n-1) near infrared spectrum database sub-libraries, wherein t near infrared spectrum database sub-libraries with s larger than a set threshold value are provided, and t is larger than 60% multiplied by n,
(6) using the database-dividing property data predicted value of t samples to be measured, calculating the database-dividing property data predicted value (P) of the samples to be measured according to the formulaSeparate warehouse)：
Wherein,
a near infrared spectrum database sub-library A with s larger than a set threshold value1、A2、…、AtAnd fitting to obtain the predicted value of each sub-library property data of the sample to be tested.
(7) Fitting the near infrared spectrum of the sample to be tested by using all the spectra in the near infrared spectrum database according to the method in the step (I) in the step (5), wherein s is not less than a set threshold value, and calculating the property data corresponding to the spectra participating in fitting to obtain a total library property data predicted value (P) of the sample to be testedWhole storehouse) And then the predicted value of the property data of the sample to be tested is calculated by the formula (viii):
P＝(60％-90％)Pseparate warehouse+(10％-40％)PWhole storehouse ⑧
In the formula, P is a predicted value of the property data of the sample to be tested, and P isSeparate warehouseIs the property of the sub-library of the sample to be testedData prediction value, PWhole storehouseAnd predicting the property data of the whole library of the sample to be tested.
2. The method according to claim 1, wherein in the step (5) and the step (II), when t is less than or equal to 60% x n, the sub-library is randomly established according to the step (4), the properties of the sample to be tested are calculated by using spectral fitting of the sub-library according to the step (5) until the value of the sub-library of the near infrared spectrum database with s larger than the set threshold value meets t > 60% x n, and the predicted value of the property data of the sample to be tested is obtained according to the steps (6) and (7).
3. The method of claim 1, wherein the properties comprise at least one of group composition, viscosity index, and pour point of the lubricant base oil.
4. The method of claim 3, wherein said group composition is selected from at least one of paraffin, monocycloalkane, dicycloalkane, tricycloalkane, tetracycloalkane, and cycloalkane contents.
5. The method of claim 1, wherein the standard method for determining the composition of a sample group of lubricant base oils is SH/T0659.
6. The method according to claim 1, wherein the number n of the built near infrared spectrum database sub-banks in the step (4) is 50 to 100.
7. The method of claim 1, wherein the threshold value in step (5) is determined by: selecting a lubricating oil base oil sample, repeatedly measuring the three-time near infrared spectrum, performing second-order differential treatment on the spectrum measured each time, and taking 7000-4000 cm-1Calculating the pseudo fitting degree (sr) value between every two spectra according to formula (c), taking the maximum sr value, multiplying the maximum sr value by the coefficient 0.75 to obtain the threshold value,
in formula (c), x'jAnd x ″)jIs the absorbance at the j-th wavelength point of the two spectra, and m is the number of wavelength points of the spectra.
8. The method of claim 1, wherein the lubricant base oil comprises a group II base oil, a group III base oil, or a group III base oil+And (3) base oil.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711042673.3A CN109724938B (en) | 2017-10-30 | 2017-10-30 | Method for predicting properties of lubricating oil base oil by near infrared spectrum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711042673.3A CN109724938B (en) | 2017-10-30 | 2017-10-30 | Method for predicting properties of lubricating oil base oil by near infrared spectrum |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109724938A CN109724938A (en) | 2019-05-07 |
| CN109724938B true CN109724938B (en) | 2021-06-11 |
Family
ID=66293445
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711042673.3A Active CN109724938B (en) | 2017-10-30 | 2017-10-30 | Method for predicting properties of lubricating oil base oil by near infrared spectrum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109724938B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115290787B (en) * | 2022-08-03 | 2023-09-08 | 青岛海关技术中心 | Attribute identification method of imported lubricating oil and application of attribute identification method in commodity classification |
| CN116242799B (en) * | 2023-03-14 | 2023-08-18 | 合肥工业大学 | Base oil detection device and detection method based on deep learning infrared multidimensional fusion algorithm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102374975A (en) * | 2010-08-19 | 2012-03-14 | 中国石油化工股份有限公司 | Method for predicting physical property data of oil product by using near infrared spectrum |
| CN103398970A (en) * | 2013-07-24 | 2013-11-20 | 骆驰 | Method for qualitatively and quantitatively analyzing edible oil and further detecting hogwash oil |
| CN103761742A (en) * | 2014-01-24 | 2014-04-30 | 武汉大学 | Method for hyperspectral remote sensing image sparse mix-decomposition based on homogenous indexes |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120034844A1 (en) * | 2010-08-05 | 2012-02-09 | Applied Materials, Inc. | Spectrographic monitoring using index tracking after detection of layer clearing |
-
2017
- 2017-10-30 CN CN201711042673.3A patent/CN109724938B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102374975A (en) * | 2010-08-19 | 2012-03-14 | 中国石油化工股份有限公司 | Method for predicting physical property data of oil product by using near infrared spectrum |
| CN103398970A (en) * | 2013-07-24 | 2013-11-20 | 骆驰 | Method for qualitatively and quantitatively analyzing edible oil and further detecting hogwash oil |
| CN103761742A (en) * | 2014-01-24 | 2014-04-30 | 武汉大学 | Method for hyperspectral remote sensing image sparse mix-decomposition based on homogenous indexes |
Non-Patent Citations (4)
| Title |
|---|
| Near-infrared spectroscopy (NIRS) and chemometric analysis of Malaysian and UK paracetamol tablets: A spectral database study;Mazlina M. Said et al.;《International Journal of Pharmaceutics》;20110527;第415卷;102-109 * |
| Rapid identifi cation and assay of crude oils based on moving-window correlation coeffi cient and near infrared spectral library;Chu Xiaoli et al.;《Chemometrics and Intelligent Laboratory Systems》;20110122;第107卷;44-49 * |
| 基于聚类分析分库策略的社交网络数据库查询性能与数据迁移;梁双等;《计算机应用》;20170310;第37卷(第3期);673-679 * |
| 近红外分析方法测定润滑油基础油的化学族组成;王艳斌等;《石油化工》;20011231;第30卷(第3期);224-227 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109724938A (en) | 2019-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107976419B (en) | Method for predicting properties of oil product by near infrared spectrum | |
| CN105388123B (en) | A kind of method by near infrared spectrum prediction oil property | |
| CN102374975B (en) | Method for predicting physical property data of oil product by using near infrared spectrum | |
| CN104990894B (en) | A kind of gasoline property detection method based on weighting absorbance and similar sample | |
| CN105466884B (en) | It is a kind of by near infrared light spectrum discrimination crude oil species and its method for property | |
| CN103534578B (en) | Determine spectral device and the method for composition present in sample | |
| CN107748146A (en) | A kind of crude oil attribute method for quick predicting based near infrared spectrum detection | |
| EP0706050A1 (en) | Lubricant property determination | |
| WO2006023800A2 (en) | Improved method for analyzing an unknown material and predicting properties of the unknown based on calculated blend | |
| CN109724938B (en) | Method for predicting properties of lubricating oil base oil by near infrared spectrum | |
| CN101995389A (en) | Method for fast recognition of crude oil variety through near infrared spectrum | |
| CN101900672A (en) | Method for quickly identifying class and viscosity grade of lubricating oil | |
| US6317654B1 (en) | Control of crude refining by a method to predict lubricant base stock's ultimate lubricant preformance | |
| CN109060702A (en) | Infrared spectroscopy quantitative analysis of nonlinear method | |
| CN107966420B (en) | Method for predicting crude oil property by near infrared spectrum | |
| CN107976417B (en) | Crude oil type identification method based on infrared spectrum | |
| CN107966499B (en) | Method for predicting crude oil carbon number distribution by near infrared spectrum | |
| CN103398971A (en) | Chemometrics method for determining cetane number of diesel oil | |
| US11415568B2 (en) | Systems and methods for implicit chemical resolution of vacuum gas oils and fit quality determination | |
| CN107976416B (en) | Method for predicting crude oil property by infrared spectrum | |
| US6295485B1 (en) | Control of lubricant production by a method to predict a base stock's ultimate lubricant performance | |
| CN109724939B (en) | Method for predicting hydrogenated tail oil property by near infrared spectrum | |
| CN107976420B (en) | Method for predicting composition of mixed crude oil by near infrared spectrum | |
| CN113782112B (en) | Method and device for determining petroleum fraction composition model | |
| CN109724937B (en) | Method for predicting vacuum distillate oil property by near infrared spectrum |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |