CN109724937A - By the method for near infrared spectrum prediction reduced pressure distillate oil nature - Google Patents
By the method for near infrared spectrum prediction reduced pressure distillate oil nature Download PDFInfo
- Publication number
- CN109724937A CN109724937A CN201711036826.3A CN201711036826A CN109724937A CN 109724937 A CN109724937 A CN 109724937A CN 201711036826 A CN201711036826 A CN 201711036826A CN 109724937 A CN109724937 A CN 109724937A
- Authority
- CN
- China
- Prior art keywords
- library
- sample
- spectrum
- near infrared
- tested
- 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.)
- Granted
Links
Abstract
A method of reduced pressure distillate oil nature is predicted by near infrared spectrum, including collects vacuum distillate sample, the property data of each sample is measured with standard method, measures the near infrared spectrum of each sample, second-order differential processing is carried out to it, takes 7000~4000cm‑1Compose the absorbance in area, it is corresponding with the property data that the sample is measured with standard method, establish near infrared spectrum data library, multiple points of libraries are randomly selected near infrared spectrum data library, the property data predicted value that the average value and full library spectrum simulation of the sample to be tested property data value obtained with each point of library spectrum simulation obtain obtains the property data predicted value of reduced pressure distillate oil samples to be measured by proper proportion adduction.
Description
Technical field
The present invention is a kind of method for predicting oil property near infrared spectrum, specifically, being a kind of to use near infrared light
The method of spectrum prediction reduced pressure distillate oil nature.
Background technique
Currently, the production technology of domestic high quality base oil is mainly full hydrogenation technique, process is related to kinds of processes
And the material and product of multiple types, the variation of a link may be affected to the production of entire base oil.Reduced pressure distillate
Oil is the starting material of full hydrogenation technique, and the quality quality of vacuum distillate can largely influence the quality of final base oil,
Viscosity index (VI) is to investigate the important indicator of basic oil quality, and the hydrocarbon system of vacuum distillate composition then will have a direct impact on subsequent basis
The viscosity index (VI) of oil.Research shows that the viscosity index (VI) highest of n-alkane, viscosity temperature characteristic are best；Isoparaffin with Long carbon chain
Take second place；Viscosity temperature characteristic it is worst be heavy aromatics, polycyclic ring alkane and cycloalkanes aromatic hydrocarbons.So accurately obtaining the race of vacuum distillate
Form information, it will directly help subsequent production technology to select and optimization, also can the final base oil of aid forecasting quality.
Race's composition of vacuum distillate is different with crude oil variation, and crude oil type that refinery itself processes change compared with
Fastly, it is therefore desirable to real-time monitoring be carried out quickly to obtain relevant information to its property, referred in time for the adjustment of technological parameter
It leads, it is more preferable to control basic oil quality, bigger benefit is brought for enterprise's production.Domestic production base oil enterprise is all based at present
Traditional analysis the pour point of vacuum distillate and the hydrocarbon system such as forms and is measured, time-consuming, not environmentally, cannot supervise online
It surveys, is no longer satisfied the demand of continual and steady efficiently production high quality base oil.Therefore, have the characteristics that quickly to detect close red
Outer technology shows huge advantage, by can be reliably to vacuum distillate pour point and hydro carbons in conjunction with Chemical Measurement by it
Composition is quickly measured.Yang Su etc. is in " with the chemical group composition of Oils by NIR heavy fraction of oil "
(INFRARED, 2006,27 (4): 20-24.) establish prediction heavy using near infrared spectrum combination deflected secondary air and evaporate
The method of the oily chemical group composition of part.But have certain scope of application using the calibration model that deflected secondary air is established, if to
There are larger differences with calibration set sample for the composition of test sample sheet, then need to carry out expanding and updating to calibration model.It establishes partially minimum
Two multiply quantitative calibration models need to carry out that pretreatment and spectrum range are preferably equal to be operated according to specific application to spectrum, because of institute
Reasons, the foundation of model such as parameter is more, multivariate calibration methods are more difficult and grasps is selected to usually require trained professional people
Member completes, this, which becomes, restricts the bottleneck problem that the technology is widely used to promote, many projects all because calibration model maintenance not
Its due effect cannot be played in time.
CN102374975A discloses a kind of method using near infrared spectrum predicting physical property data of oil product, proposes one kind
New property prediction technique-library spectrum simulation method (Library Spectra Fitting Method), this method is based on
The Near-infrared spectrum database and spectrum simulation technology of oil product, based on spectrum, similar, the similar principle of property, passes through one in library of spectra
It opens or multiple spectrum is fitted the spectrum of unknown sample to be tested, then calculated according to the property for participating in fit-spectra to be measured
The property of sample, chemistry essence are that unknown sample can be mixed in a certain ratio by one group of library sample, therefore, unknown sample
This property to be measured can be obtained by the property of library sample according to contribution calcutation.But the method depends in library of spectra and participates in light
The property of limited several sample properties data prediction samples to be tested of spectrum fitting, to influence the accuracy and stabilization of prediction
Property.
Summary of the invention
The object of the present invention is to provide a kind of method by near infrared spectrum prediction reduced pressure distillate oil nature, which be can be improved
The forecasting accuracy and stability of sample to be tested property data.
Method provided by the invention by near infrared spectrum prediction reduced pressure distillate oil nature, includes the following steps:
(1) at least 300 vacuum distillate samples are collected, the property data of each sample is measured with standard method,
(2) near infrared spectrum for measuring each sample, carries out second-order differential processing to it, takes 7000~4000cm-1Compose area
Absorbance, it is corresponding with the property data that the sample is measured with standard method, establish near infrared spectrum data library,
(3) near infrared spectrum of vacuum distillate sample to be measured is measured, and carries out second-order differential, selection 7000~
4000cm-1Compose the absorbance in area；
(4) near infrared spectrum data library, the spectroscopic data of multiple samples is randomly selected, near infrared spectrum data is established
Library Fen Ku, each dividing the sample number in library is the 50~70% of near infrared spectrum data library sample number, vertical n near infrared spectrum of building together
Database divides library, and the sample in each point library is all different,
(5) (I) divides library A for some near infrared spectrum data libraryp, the spectrum of sample to be tested is carried out as follows
Fitting:
A) spectrum that sample to be tested is 1. indicated by formula, acquires the fitting coefficient of library spectrum,
Wherein, x is the spectrum of sample to be tested, viDivide library A near infrared spectrum data librarypIn i-th of library spectrum, k is spectrum
Database divides library ApIn spectrum number, aiDivide library A for spectra databasepIn the corresponding fitting coefficient of i-th of library spectrum,
Fitting coefficient aiIt is found out using classical nonnegativity restrictions least square method, that is, meets following objective function:
B) by the above-mentioned fitting coefficient a acquirediIn all non-zero fitting coefficient extract, be 2. normalized by formula
Processing, obtains normalized fitting coefficient bi:
Wherein, g is the number of non-zero fitting coefficient,
C) fit-spectra of sample to be tested is 3. calculated by formula:
D) degree of fitting of sample to be tested spectrum is 4. calculated by formula,
Wherein, xjFor the absorbance of j-th of wavelength points of sample to be tested spectrum,For the suction of j-th of wavelength points of fit-spectra
Luminosity, m are that the wavelength of spectrum is counted,
If s is greater than the threshold value of setting, this point of library property data predicted value of sample to be tested is 5. obtained by formula,
Wherein,To divide library A by near infrared spectrum data librarypIt is fitted this point of obtained library property data predicted value, qiFor
The corresponding property data of spectroscopic data of fitting is participated in,
If s is less than the threshold value of setting, the near infrared spectrum data library point library is not used to calculate the property number of sample to be tested
According to,
(II) method for pressing (I) step, with a near infrared spectrum data library other (n-1) divide library to the spectrum of sample to be tested into
Row fitting shares the near infrared spectrum data library Fen Ku that t s is greater than the threshold value of setting, and t > 60% × n,
(6) divide library property data predicted value with t sample to be tested, point library property data of sample to be tested is 6. calculated by formula
Predicted value (PDivide library):
Wherein,Respectively it is greater than the near infrared spectrum data library point of the threshold value of setting by s
Library A1、A2、…、AtIt is fitted each point of library property data predicted value of obtained sample to be tested.
(7) method of (I) step in (5) step is pressed to the close red of sample to be tested with all spectrum near infrared spectrum data library
External spectrum is fitted, and s not less than setting threshold value, and by participate in fitting the corresponding property data of spectrum be calculated to
The full library property data predicted value (P of test sample sheetQuan Ku), then 8. the property data predicted value of sample to be tested is calculated by formula:
P=(60%-90%) PDivide library+ (10%-40%) PQuan Ku ⑧
Formula 8. in, P be sample to be tested property data predicted value, PDivide libraryDivide library property data predicted value for sample to be tested,
PQuan KuFor the full library property data predicted value of sample to be tested.
The method of the present invention uses the method established near infrared spectrum data library and divide library, treats test sample with the library spectrum in point library
This spectrum is fitted, and then sample to be tested is calculated in this point of library with the property data for point Kuku spectrum for participating in fitting
Property data predicted value, then divide library property data using the average value of the property data predicted value in each point of library as sample to be tested
A point library property data predicted value for sample to be tested is obtained the property of sample to be tested by predicted value in conjunction with full library property data predicted value
Prime number is it is predicted that value.The method can make full use of the resource in existing near infrared spectrum data library, effectively improve sample to be tested property
Forecasting accuracy.
Specific embodiment
Method of the method for the present invention by establishing several vacuum distillate near infrared spectrum datas library point library, by dividing Kuku light
Spectrum is fitted the spectrum of sample to be tested, obtains the property data predicted value for dividing the sample to be tested in library, then the institute by that can be fitted
There is point average value of library prediction result to obtain point library property data predicted value of sample to be tested, in conjunction with by full library spectrum simulation
The two is obtained the property number of sample to be tested by appropriate weight ratio adduction by the full library property data predicted value of obtained sample to be tested
It is predicted that value.Compared with CN102374975A, absolute dependence vacuum distillate spectra database can be eliminated to a certain extent
In limited sample data limitation, to a greater extent extension participate in fitting sample size, to improve vacuum distillate
The forecasting accuracy and stability of sample properties.
The foreseeable reduced pressure distillate oil nature of the method for the present invention include vacuum distillate race composition and pour point in extremely
Few one kind.
Race's composition can be alkane, a ring cycloalkane, bicyclic ring alkane, tricyclic naphthenes hydrocarbon, Fourth Ring alkane, cycloalkanes
At least one of hydrocarbon, mononuclear aromatics and arene content
The method of the present invention (1) step is to collect representational vacuum distillate sample, and the sample data of collection is at least 300
A, suitable collection sample number is 300~500, then the property data of each sample is measured with standard method.The measurement subtracts
The standard method for pressing distillate sample race composition is SH/T0659, and the standard method for measuring viscosity index (VI) is GB/T1995.
The vacuum distillate that the optional boiling range of vacuum distillate is 350~540 DEG C.
The method of the present invention (2) step is the near infrared spectrum for measuring each sample, after carrying out second-order differential processing to it, is chosen
7000~4000cm-1The absorbance (for its spectrum) in area is composed, it is corresponding with the property data that the sample is measured with standard method, it establishes
Near infrared spectrum data library.The property data may include above-mentioned race composition and pour point, can also be it is therein several, if there is it
Its property data can also be added near infrared spectrum data library.
(3) step is the near infrared spectrum for measuring vacuum distillate sample to be measured, and carries out second-order differential, chooses 7000
~4000cm-1Compose the absorbance (for its spectrum) in area；In order to be fitted with library spectrum to it.Measure vacuum distillate sample to be measured
Near infrared spectrum method it is identical as the method for near infrared spectrum that (2) establish near infrared spectrum data library institute sample.
Spectroscopic data is randomly selected in (4) Bu Weicong near infrared spectrum data library and establishes point library, and each a point library includes
Spectrum should cover all types of data near infrared spectrum data library as far as possible, the sample size for including in point library should be close red
The 50~70% of sample size, preferably 55~70% in external spectrum database.Sample contained in each point of library is built to be all different,
I.e. no identical point of inventory exists.The quantity n preferably 50~100 in library is divided in built near infrared spectrum data library.
(5) step is to be fitted with spectrum of point library spectrum to vacuum distillate sample to be measured, and will be calculated
Degree of fitting s compared with threshold value, whether judging spectrum simulation completely.Divide library A to some near infrared spectrum data libraryp, with (I)
Listed a)-d of step) method of step is fitted, and finally obtains the degree of fitting s of sample to be tested spectrum, and it is wherein b) non-negative described in step
The specific algorithm of Constraint least square algorithm is referring to document: C.L.Lawson and R.J.Hanson, Solving Least
Squares Problems,Prentice-Hall,Englewood Cliffs,NJ(1974)；160~165.
S is judgement point library spectrum to the index of sample to be tested degree of fitting, and the value is bigger to illustrate that degree of fitting is higher, thus calculates
Obtained prediction property is more accurate.If s is less than the threshold value of setting, illustrate that spectrum simulation is incomplete, i.e., sample to be tested cannot be complete
Quan Youku spectrum simulation indicates, thus can not carry out Accurate Prediction to its property data with point property data of library spectrum.
Divide library greater than threshold value to s, the property data predicted value in this point of library is 5. calculated with formula.
Above-mentioned s judge degree of fitting, applicable elements and property calculation method it is similarly suitable with full library spectrum simulation calculate to
The property data predicted value of test sample sheet.
(5) threshold value determination method described in step are as follows: choose a vacuum distillate sample, be repeated three times near-infrared
Spectrum carries out second-order differential processing to the spectrum measured every time, takes 7000~4000cm-1Compose area absorbance, by formula 7. based on
Pseudo- degree of fitting (sr) value between spectrum two-by-two is calculated, maximum sr value is taken, is threshold value multiplied by coefficient 0.75,
Formula 7. in, x 'jWith x "jFor the absorbance of j-th of wavelength points of two spectrum, m is that the wavelength of spectrum is counted.It is described
Spectrum wavelength points for spectrum range acquire absorbance wavelength points number.
(II) step in (5) step of the present invention is calculates remaining (n-1) a sample to be tested for building point library by (I) method
Spectrum simulation degree shares the near infrared spectrum data library Fen Ku that t s is greater than the threshold value of setting, and t > 60% × n.T > 60% × n
For the available condition for dividing library spectrum simulation to calculate sample to be tested property, only meet above-mentioned condition, just can be used (6) step to calculate to be measured
Sample divides library property data predicted value.
(6) step of the present invention be with meet (5) (II) step condition t sample to be tested divide library property data prediction
6. value, its average value is calculated by formula, obtains point library property data predicted value of sample to be tested.
8. (7) Bu Weiyong formula of the present invention calculates the property data predicted value of sample to be tested, formula 8. in, PQuan KuFor with institute
The whole spectrum for building near infrared spectrum data library are fitted the property data predicted value being calculated to sample to be tested, fitting
Calculation method is the same as the method for dividing library spectrum simulation to calculate sample to be tested property.
(5) it in step (II) step, as t≤60% × n, is unsatisfactory for calculating the item of sample to be tested property with point library spectrum simulation
Part is established at random by (4) one step process again and divides library, then the property for dividing library spectrum simulation to calculate sample to be tested is used by (5) one step process,
Until the numerical value that s is greater than the near infrared spectrum data library point library of the threshold value of setting meets t > 60% × n, then press (6), (7) step side
Method obtains the property data predicted value of sample to be tested.Preferably, when random foundation divides library again, increase divides library quantity or each point
The quantity of sample in library, can also both carry out simultaneously.
The method of the present invention is suitable for the property data near infrared spectrum quick predict vacuum distillate sample, can be used for subtracting
The field monitoring of fraction oil nature is pressed, to be adjusted in real time to vacuum distillate processing parameter.
Below by example, present invention be described in more detail, but the present invention is not limited thereto.
In example and comparative example, in Fu of the instrument of measurement vacuum distillate near infrared spectrum using the production of Thermo company
Leaf transformation near infrared spectrometer (ANTARIS II), 3500~10000cm of spectral region-1, resolution ratio 8cm-1, accumulation scanning time
Number 128 times, sample stablizes 5min before adopting spectrum.
Example 1
Predict the race's composition and physical data of vacuum distillate.
(1) the near infrared spectrum data library of vacuum distillate is established
395, representational vacuum distillate sample is collected, is formed with the race that SH/T0659 method measures each sample
Data, including alkane, a ring cycloalkane, bicyclic ring alkane, tricyclic naphthenes hydrocarbon, Fourth Ring alkane, cycloalkane, mononuclear aromatics and virtue
Hydrocarbon content measures its pour point data with GB/T3535 method.
The near infrared spectrum for measuring each sample carries out second-order differential processing to it, chooses 7000~4000cm-1Spectrum model
The wavelength points of the absorbance enclosed, the acquisition absorbance of the spectrum range are 875.By 7000~4000cm-1Compose the absorbance in area
It is corresponding with the race's composition and pour point of standard method measurement, establish near infrared spectrum data library.
(2) the threshold value s of digital simulation degreev
1 reduced pressure distillate oil samples is taken, its near infrared spectrum is repeated three times, after carrying out second-order differential processing, is chosen
7000~4000cm-17. the absorbance of spectral region is calculated pseudo- degree of fitting (sr) value between spectrum two-by-two by formula, taken maximum
Sr value is threshold value s multiplied by coefficient 0.75v, sv=2.64.
(3) vacuum distillate near infrared spectrum data library Fen Ku is established at random
The near infrared spectrum data library of vacuum distillate is generated into 80 points of library A at random1、A2、…、A80, each a point library is selected
The sample number taken is 240.
(4) spectrum of sample to be tested is fitted with point library data, obtains its point of library property data predicted value
1 vacuum distillate sample to be tested A is taken, measures it by method identical with (2) this near infrared spectrum of pacing random sample
Near infrared spectrum carries out second-order differential, takes 7000~4000cm-1The absorbance of spectral region, to a near infrared spectrum data
Library A is divided in libraryp, by a described in (5) of the present invention (I)) and~d) method of step calculates the property data predicted value in this point of library,
I.e. by formula 1.~calculating 3. is fitted to the spectrum of sample to be tested A, obtain the fit-spectra of sample to be tested, 4. right back-pushed-type is counted
Degree of fitting s, the s=4.86 of sample to be tested spectrum are calculated, threshold value s is greater thanv, then by formula 5. by the corresponding property of spectrum of participation fitting
Data calculate the property data predicted value of this point of library sample to be tested.
Spectrum simulation is carried out to remaining 79 points of library in 80 points of libraries according to the above method, it is big to share 58 degree of fitting s
In threshold value svDivide library, i.e. t=58, be greater than (80 × 0.6=) 48, meet with dividing library spectrum simulation to calculate sample to be tested property
Condition.
(5) property of sample to be tested is predicted
With point library property data predicted value for the sample to be tested that 58 points of libraries are obtained by library spectrum, by formula 6. by 58 points
The sample to be tested property data predicted value in libraryCalculate sample to be tested divides library property data pre-
Measured value PDivide library。
Near infrared spectrum data library is established to (1) step, by formula 1.~calculating 3. is fitted to the spectrum of sample to be tested A,
The fit-spectra of sample to be tested is obtained, 4. right back-pushed-type calculates degree of fitting s, the s=3.58 of sample to be tested spectrum, be greater than threshold value
sv, then by formula the full library property data predicted value of sample to be tested is 5. calculated by the corresponding property data of spectrum for participating in fitting
PQuan Ku, the property data predicted value of sample to be tested is 8. calculated by formula, specifically, by P=70%PDivide library+ 30%PQuan KuIt calculates to test sample
The property data predicted value of product, the race's composition and pour point predicted value of obtained vacuum distillate sample A to be measured are shown in Table 1.
Comparative example 1
Vacuum distillate sample to be tested A used in example 1 is taken, predicts its property data by CN102374975A method, is tied
Fruit is shown in Table 1.
As shown in Table 1, the method for the present invention has lesser prediction deviation, illustrates this hair compared with CN102374975A method
Bright method has higher forecasting accuracy.
Table 1
* deviation 1: the deviation between 1 predicted value of example and standard method measured value；
* the deviation between deviation 2:CN102374975A method predicted value and standard method measured value.
Example 2
Take 1 vacuum distillate sample to be tested B, measure its near infrared spectrum by the method for example 1 (3)-(5) step, altogether with
Machine establishes 80 points of libraries, and each dividing library sample number is 240, and calculates its property data with a point library spectrum simulation, shares 56
Degree of fitting s is greater than threshold value svDivide library, t=56 is greater than (80 × 0.6=) 48, by the property data predicted value in this 56 points of libraries
6. P is obtained by formulaDivide library, then sample to be tested is fitted with full library spectrum, calculate PQuan Ku, 8. obtain sample to be tested B's by formula
Property data predicted value, specifically, by P=80%PDivide library+ 20%PQuan KuThe property data predicted value for calculating sample to be tested, is as a result shown in
Table 2.
Comparative example 2
The vacuum distillate sample to be tested B for taking example 2, predicts its property data by CN102374975A method, as a result sees
Table 2.
Table 2
Example 3
1 vacuum distillate sample to be tested C is taken, measures its near infrared spectrum by the method for example 1 (3)-(5) step, at random
80 points of libraries are established, each dividing library sample number is 240, and calculates its property data with a point library spectrum simulation, shares 40 and intends
Right s is greater than threshold value svDivide library, t=40 is less than (80 × 0.6=) 48, is unsatisfactory for point library spectrum simulation calculating to test sample
The condition of this property.The method for repeating example 1 (3)-(5) step, establishes 80 points of libraries at random, and each dividing library sample number is 270
It is a, and its property data is calculated with a point library spectrum simulation, 55 degree of fitting s are shared greater than threshold value svDivide library, t1=55, it is greater than
6. (80 × 0.6=) 48 obtains P by formula by the property data predicted value in this 55 points of librariesDivide library, then to the full library light of sample to be tested
Spectrum is fitted, and calculates PQuan Ku, the property data predicted value of sample to be tested C is 8. obtained by formula, specifically, by P=70%PDivide library+
30%PQuan KuThe property data predicted value for calculating sample to be tested, the results are shown in Table 3.
Comparative example 3
The vacuum distillate sample to be tested C for taking example 3, predicts its property data by CN102374975A method, as a result sees
Table 3.
Table 3
Claims (8)
1. a kind of method by near infrared spectrum prediction reduced pressure distillate oil nature, includes the following steps:
(1) at least 300 vacuum distillate samples are collected, the property data of each sample is measured with standard method,
(2) near infrared spectrum for measuring each sample, carries out second-order differential processing to it, takes 7000~4000cm-1Compose the suction in area
Luminosity, it is corresponding with the property data that the sample is measured with standard method, near infrared spectrum data library is established,
(3) near infrared spectrum of vacuum distillate sample to be measured is measured, and carries out second-order differential, chooses 7000~4000cm-1Spectrum
The absorbance in area；
(4) near infrared spectrum data library, the spectroscopic data of multiple samples is randomly selected, establishes near infrared spectrum data library point
Library, each dividing the sample number in library is the 50~70% of near infrared spectrum data library sample number, vertical n near infrared spectrum data of building together
Library Fen Ku, and the sample in each point library is all different,
(5) (I) divides library A for some near infrared spectrum data libraryp, the spectrum of sample to be tested is fitted as follows:
A) spectrum that sample to be tested is 1. indicated by formula, acquires the fitting coefficient of library spectrum,
Wherein, x is the spectrum of sample to be tested, viDivide library A near infrared spectrum data librarypIn i-th of library spectrum, k is spectroscopic data
Library A is divided in librarypIn spectrum number, aiDivide library A for spectra databasepIn the corresponding fitting coefficient of i-th of library spectrum,
Fitting coefficient aiIt is found out using classical nonnegativity restrictions least square method, that is, meets following objective function:
B) by the above-mentioned fitting coefficient a acquirediIn all non-zero fitting coefficient extract, be 2. normalized by formula,
Obtain normalized fitting coefficient bi:
Wherein, g is the number of non-zero fitting coefficient,
C) fit-spectra of sample to be tested is 3. calculated by formula:
D) degree of fitting of sample to be tested spectrum is 4. calculated by formula,
Wherein, xjFor the absorbance of j-th of wavelength points of sample to be tested spectrum,For the absorbance of j-th of wavelength points of fit-spectra,
M is that the wavelength of spectrum is counted,
If s is greater than the threshold value of setting, this point of library property data predicted value of sample to be tested is 5. obtained by formula,
Wherein,To divide library A by near infrared spectrum data librarypIt is fitted this point of obtained library property data predicted value, qiTo participate in
The corresponding property data of the spectroscopic data of fitting,
If s is less than the threshold value of setting, the near infrared spectrum data library point library is not used to calculate the property data of sample to be tested,
(II) method for pressing (I) step, intends the spectrum of sample to be tested with other a near infrared spectrum data library point libraries (n-1)
It closes, shares the near infrared spectrum data library Fen Ku that t s is greater than the threshold value of setting, and t > 60% × n,
(6) divide library property data predicted value with t sample to be tested, point library property data prediction of sample to be tested is 6. calculated by formula
It is worth (PDivide library):
Wherein,Library A is divided in the near infrared spectrum data library for being respectively greater than the threshold value of setting by s1、
A2、…、AtIt is fitted each point of library property data predicted value of obtained sample to be tested,
(7) near infrared light of the method to sample to be tested of (I) step in (5) step is pressed with all spectrum near infrared spectrum data library
Spectrum is fitted, and s is calculated not less than the threshold value of setting, and by the corresponding property data of spectrum of participation fitting to test sample
This full library property data predicted value (PQuan Ku), then 8. the property data predicted value of sample to be tested is calculated by formula:
P=(60%-90%) PDivide library+ (10%-40%) PQuan Ku ⑧
Formula 8. in, P be sample to be tested property data predicted value, PDivide libraryDivide library property data predicted value, P for sample to be testedQuan Ku
For the full library property data predicted value of sample to be tested.
2. according to the method for claim 1, it is characterised in that (5) in step (II) step, as t≤60% × n, press again
(4) one step process is established at random divides library, then by (5) the one step process property for dividing library spectrum simulation to calculate sample to be tested, until s is greater than
The numerical value in the near infrared spectrum data library point library of the threshold value of setting meets t > 60% × n, then obtains by (6), (7) one step process to be measured
The property data predicted value of sample.
3. according to the method for claim 1, it is characterised in that the property includes race's composition of vacuum distillate and inclines
At least one of point.
4. according to the method for claim 3, it is characterised in that race's composition is selected from alkane, a ring cycloalkane, two
At least one of ring cycloalkane, tricyclic naphthenes hydrocarbon, Fourth Ring alkane, total cycloalkane, mononuclear aromatics and arene content.
5. according to the method for claim 1, it is characterised in that the standard side of measurement vacuum distillate sample race composition
Method is SH/T0659.
6. according to the method for claim 1, it is characterised in that (4) the quantity n in library is divided in Bu Suojian near infrared spectrum data library
It is 50~100.
7. according to the method for claim 1, it is characterised in that (5) threshold value determination method described in step are as follows: choose one
Vacuum distillate sample, is repeated three times near infrared spectrum, carries out second-order differential processing to the spectrum measured every time, takes
7000~4000cm-17. the absorbance for composing area, pseudo- degree of fitting (sr) value between spectrum two-by-two is calculated by formula, takes maximum sr
Value is threshold value multiplied by coefficient 0.75,
Formula 7. in, x 'jWith x "jFor the absorbance of j-th of wavelength points of two spectrum, m is that the wavelength of spectrum is counted.
8. according to the method for claim 1, it is characterised in that it is 350~540 DEG C that the vacuum distillate, which is selected from boiling range,
Vacuum distillate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711036826.3A CN109724937B (en) | 2017-10-30 | 2017-10-30 | Method for predicting vacuum distillate oil property by near infrared spectrum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711036826.3A CN109724937B (en) | 2017-10-30 | 2017-10-30 | Method for predicting vacuum distillate oil property by near infrared spectrum |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109724937A true CN109724937A (en) | 2019-05-07 |
| CN109724937B CN109724937B (en) | 2021-05-14 |
Family
ID=66291848
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711036826.3A Active CN109724937B (en) | 2017-10-30 | 2017-10-30 | Method for predicting vacuum distillate oil property by near infrared spectrum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109724937B (en) |
Citations (4)
| 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 |
| 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 |
-
2017
- 2017-10-30 CN CN201711036826.3A patent/CN109724937B/en active Active
Patent Citations (4)
| 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 |
| 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 |
|---|
| MAZLINA M. SAID ET AL.: "Near-infrared spectroscopy (NIRS) and chemometric analysis of Malaysian and UK paracetamol tablets: A spectral database study", 《INTERNATIONAL JOURNAL OF PHARMACEUTICS》 * |
| XIAO-LI CHU ET AL.: "Rapid identifi cation and assay of crude oils based on moving-window correlation coeffi cient and near infrared spectral library", 《CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS》 * |
| 梁双等: "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", 《计算机应用》 * |
| 王艳斌等: "近红外分析方法测定润滑油基础油的化学族组成", 《石油化工》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109724937B (en) | 2021-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| 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 | |
| CN107976419B (en) | Method for predicting properties of oil product by near infrared spectrum | |
| CN102539326B (en) | Method for carrying out quantitative evaluation on soup hue quality of tea | |
| CN111721715B (en) | Tobacco shred blending uniformity measuring method based on combination of chromaticity value and entropy weight method | |
| CN105466884B (en) | It is a kind of by near infrared light spectrum discrimination crude oil species and its method for property | |
| CN104990894B (en) | A kind of gasoline property detection method based on weighting absorbance and similar sample | |
| CN104390927B (en) | The quick determination method of ash content in coal sample | |
| Wiesinger et al. | Pigment and binder concentrations in modern paint samples determined by IR and Raman spectroscopy | |
| CN101900672A (en) | Method for quickly identifying class and viscosity grade of lubricating oil | |
| CN105987886B (en) | The method of near infrared ray hydrocracking tail oil property | |
| CN108760789A (en) | A kind of crude oil fast evaluation method | |
| US6477516B1 (en) | System and method for predicting parameter of hydrocarbon with spectroscopy and neural networks | |
| CN105319179B (en) | A kind of method using middle infrared spectrum prediction hydrogen sulfide content in desulfurized amine | |
| CN103398971A (en) | Chemometrics method for determining cetane number of diesel oil | |
| CN106770005B (en) | A kind of division methods of calibration set for near-infrared spectrum analysis and verification collection | |
| CN109724938A (en) | By the method for near infrared spectrum prediction lube base oil nature | |
| US11415568B2 (en) | Systems and methods for implicit chemical resolution of vacuum gas oils and fit quality determination | |
| CN109724939A (en) | By the method for near infrared spectrum prediction hydrogenation tail oil property | |
| CN109724937A (en) | By the method for near infrared spectrum prediction reduced pressure distillate oil nature | |
| JP2012246490A5 (en) | ||
| CN107976416A (en) | A kind of method by infrared spectrum prediction oil property | |
| CN107356556A (en) | A kind of double integrated modelling approach of Near-Infrared Spectra for Quantitative Analysis | |
| CN104568811B (en) | The new quick determination method of caloric value in coal sample | |
| CN107976420B (en) | Method for predicting composition of mixed crude oil 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 |