CN115248193A - On-line multi-channel measuring method for oil generated by catalytic reforming process - Google Patents
On-line multi-channel measuring method for oil generated by catalytic reforming process Download PDFInfo
- Publication number
- CN115248193A CN115248193A CN202110456527.5A CN202110456527A CN115248193A CN 115248193 A CN115248193 A CN 115248193A CN 202110456527 A CN202110456527 A CN 202110456527A CN 115248193 A CN115248193 A CN 115248193A
- Authority
- CN
- China
- Prior art keywords
- spectrum
- sample
- cosine
- samples
- catalytic reforming
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 117
- 238000001833 catalytic reforming Methods 0.000 title claims abstract description 64
- 230000008569 process Effects 0.000 title claims abstract description 41
- 238000001228 spectrum Methods 0.000 claims abstract description 174
- 238000002329 infrared spectrum Methods 0.000 claims abstract description 31
- 238000004458 analytical method Methods 0.000 claims abstract description 29
- 238000000691 measurement method Methods 0.000 claims abstract description 20
- 238000005259 measurement Methods 0.000 claims abstract description 18
- 230000005540 biological transmission Effects 0.000 claims abstract description 13
- 239000000523 sample Substances 0.000 claims description 149
- 239000013598 vector Substances 0.000 claims description 61
- 238000012937 correction Methods 0.000 claims description 44
- 150000004945 aromatic hydrocarbons Chemical class 0.000 claims description 36
- 230000003595 spectral effect Effects 0.000 claims description 34
- 239000011159 matrix material Substances 0.000 claims description 29
- 239000000203 mixture Substances 0.000 claims description 22
- 230000003287 optical effect Effects 0.000 claims description 21
- 239000013307 optical fiber Substances 0.000 claims description 20
- 238000012795 verification Methods 0.000 claims description 20
- TVMXDCGIABBOFY-UHFFFAOYSA-N octane Chemical compound CCCCCCCC TVMXDCGIABBOFY-UHFFFAOYSA-N 0.000 claims description 19
- 238000010200 validation analysis Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 14
- 238000000862 absorption spectrum Methods 0.000 claims description 13
- 238000006243 chemical reaction Methods 0.000 claims description 10
- 125000003118 aryl group Chemical group 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000002407 reforming Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 238000010986 on-line near-infrared spectroscopy Methods 0.000 claims description 8
- 230000017105 transposition Effects 0.000 claims description 8
- 239000012468 concentrated sample Substances 0.000 claims description 6
- 238000010561 standard procedure Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000002835 absorbance Methods 0.000 claims description 4
- 229910052736 halogen Inorganic materials 0.000 claims description 4
- 150000002367 halogens Chemical class 0.000 claims description 4
- 239000004575 stone Substances 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000009365 direct transmission Effects 0.000 claims description 2
- 238000002203 pretreatment Methods 0.000 claims description 2
- 238000000513 principal component analysis Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000002349 favourable effect Effects 0.000 abstract description 2
- 238000012546 transfer Methods 0.000 description 10
- 238000004519 manufacturing process Methods 0.000 description 6
- 239000002994 raw material Substances 0.000 description 3
- 238000011425 standardization method Methods 0.000 description 3
- 230000032683 aging Effects 0.000 description 2
- 230000003197 catalytic effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- BASFCYQUMIYNBI-UHFFFAOYSA-N platinum Chemical compound [Pt] BASFCYQUMIYNBI-UHFFFAOYSA-N 0.000 description 2
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 1
- 241001372564 Piona Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003054 catalyst Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004821 distillation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 229910052739 hydrogen Inorganic materials 0.000 description 1
- 239000001257 hydrogen Substances 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 229910052697 platinum Inorganic materials 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/01—Arrangements or apparatus for facilitating the optical investigation
- G01N21/03—Cuvette constructions
- G01N21/031—Multipass arrangements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
Abstract
The invention relates to an on-line multi-channel measurement method for oil generated by a catalytic reforming process, which comprises the steps of establishing a sample set of the oil generated by the catalytic reforming process, performing spectrum pretreatment, establishing a directly transferable multivariate calibration model and the like. Compared with the existing method, the model transmission method used in the method does not need to establish a mathematical relation among different channels through a group of standard sample sets, does not need to establish a mixed model by utilizing the spectrums of all the channels, can carry out model transmission, realizes simultaneous online measurement of a plurality of channels, and has small workload and easy popularization. The method can also be used for model transmission among different instruments, and is favorable for popularization and application of the near infrared spectrum analysis technology.
Description
Technical Field
The invention relates to a spectral on-line analysis method, in particular to an on-line multi-channel measurement method for oil composition and property generated by a catalytic reforming process based on an angular spectrum conversion direct model transfer method.
Background
The catalytic reforming process is an important process for producing aromatic hydrocarbons by reforming naphtha as a raw material with platinum, and has an extremely important position in the technical field of petrochemical industry. During the production process, the yield and composition of the product (the produced oil) are affected by changes in the raw materials, catalyst activity, hydrogen content, process conditions, and the like. Therefore, the method has important significance for improving the yield and quality by adjusting the raw materials and the process conditions in time according to the quality change of the generated oil. The main properties of the resulting oil include: octane number (RON, MON), PIONA composition, BTX composition, density, vapor pressure, viscosity, distillation range, etc. The measurement of the indexes involves the use of various standard methods, which not only have high cost, but also have long measurement period and cannot meet the requirements of optimization and control of the production process of the reforming process. The near infrared spectrum has the advantages of rapid and simultaneous determination of various properties, can be used for on-line analysis in the production process by combining with an optical fiber probe, and is widely applied to various fields such as agriculture, food, biochemistry, petroleum and the like. The composition and the property of the oil generated in the catalytic reforming process are analyzed in real time by utilizing a near infrared online analysis technology, so that the optimization and the control of the reforming process can be realized. Near-infrared online analysis techniques typically rely on multivariate calibration models for quantitative analysis of composition and properties. The multivariate calibration model is a mathematical relationship between the spectra of a set of samples and their corresponding compositions and properties established by multivariate calibration methods. The number of samples used for modeling is large, and composition or property data for modeling samples need to be analyzed by adopting a standard method, so that the modeling workload is huge, and a large amount of manpower and material resources are required. An actual production process typically has multiple production channels, requiring multiple instruments or the same instrument to simultaneously detect multiple compositions or properties of multiple channels of product. Near infrared analyzers vary from one instrument to another due to processing accuracy, and spectral acquisition accessories such as fiber couplers or flow cells mounted on different channels also vary from one processing accuracy to another. Therefore, the spectral data collected on the multiple channels are different, and the models established between different instruments and channels cannot be used universally, which further increases the workload of modeling. For example, analyzing 5 compositions or properties of a product for 8 channels requires building 40 models. In addition, aging of the instrument, performance changes of the instrument due to replacement of key parts, and changes of measurement environment conditions also cause large deviation of the prediction results of the previously established model, and even the model cannot be used.
In order to solve the problem that the models cannot be used universally, many model transfer methods have been studied. The instrument for which the model is established is called a master machine, and the instrument to be transmitted is called a slave machine. The current model transfer methods mainly comprise three methods, namely a model coefficient updating method, a prediction result correcting method and a spectrum data standardization method. Model coefficient update the model coefficients are recalculated by adding some new samples to the old sample set. This method is often used for model transfer problems in the same instrument after aging of the instrument or changes in the environment, sometimes requiring the addition of large amounts of new samples in order to obtain good results. The predicted result correcting method is to establish a mathematical relation between the predicted values of the host machine and the submachine to realize the correction of the host machine model. The method is suitable for the situation that the reason of the spectrum difference is simple and systematic, and the application range is narrow, wherein the most typical method is the SBC method. The most ideal model transmission method is a spectrum data standardization method, and the method corrects the spectrum on the submachine by establishing a mathematical relationship between the spectra collected on the mainframe and the submachine so that the spectra can be close to the spectrum on the mainframe to the maximum extent, thereby directly adopting the spectrum of the submachine for prediction without changing a mainframe model. Among them, the PDS algorithm is most typical. The method of standardization of spectral data can be further divided into methods with and without standards depending on whether standard samples are required. The standard sample method needs to select a certain number of representative samples from the samples used for modeling by the host computer to form a transfer standard sample set, and a spectrum is measured on a transferred submachine so as to establish the relationship between the spectra of the host computer and the spectra of the submachine. In practice, however, it may not be possible to obtain a standard sample or to measure the spectrum on each subset, and the selection and storage of the standard sample is also a problem. The standard-free transmission does not need to use a sample which is modeled on the host computer as a transmission standard sample set, but also needs a certain new sample to measure the spectrum on the sub computer so as to establish the relationship between the host computer and the sub computer. The selection and number of these new samples determines how good the model is to be transferred, and also brings about a relatively large amount of work. There are also methods of building a hybrid model, i.e. modeling multiple instruments or samples on individual channels together. The method is applied to multi-channel on-line detection of the composition of RON and aromatic hydrocarbon in the medium-sized device for catalytic reforming, and the spectral difference between different channels is obviously improved. However, this method still requires collection of spectral and property data of a large number of samples for each channel, and is also labor intensive. These methods are more difficult for on-line analysis because the on-line production process is fixed, the product composition and properties do not fluctuate much, and it is more difficult to obtain representative samples that meet modeling requirements that differ in composition or properties. Therefore, the current model transfer still needs to be completed by consuming large manpower and material resources, and the development of the near infrared spectrum analysis technology is seriously hindered.
Disclosure of Invention
In order to solve the problems existing in the on-line analysis, the invention provides an on-line multi-channel measurement method for oil composition and property generated by a catalytic reforming process based on a spectral standardization model transfer method of angular spectrum conversion.
Therefore, the invention provides an online multi-channel measurement method for oil generated by a catalytic reforming process, which comprises the following steps:
s1: creating a sample set
Collecting reformate produced by a catalytic reforming device under different process parameters as a sample, and measuring the octane number and the aromatic hydrocarbon content of the sample by adopting a standard method as reference data for modeling;
collecting a near infrared spectrum of a sample by adopting a certain channel in an online near infrared analysis system, wherein the near infrared spectrum is used as an original spectrum for modeling, and the near infrared spectrum is an absorbance spectrum;
creating a sample set using the reference data and the raw spectra;
s2: spectral pre-treatment and conversion
Selecting a spectrum interval, and preprocessing the original spectrum by using a discrete wavelet transform method to remove noise;
taking a unit vector a as a reference vector, performing spectral measurement conversion on the original spectrum by using a moving window method, converting an absorbance spectrum into a corresponding cosine spectrum, wherein each original spectrum of a sample correspondingly obtains a cosine spectrum vector H = (cos (theta 1), cos (theta 2), \ 8230;, cos (theta m)), m represents the number of wavelength points of the cosine spectrum, and all samples in a sample set form a cosine spectrum matrix H = [ cos (theta i 1), cos (theta i 2), \8230;, cos (theta im) ], i = (1 k), and k represents the number of measured samples;
s3: building a model capable of direct transmission
Dividing a sample set into a correction set and a verification set according to the proportion of 6-9 to 4-1 by using a Rank-kennard-Stone method, selecting a range between cosine spectrum regions, respectively associating a cosine spectrum matrix of the correction set with reference data by using a partial least square method, and establishing an octane number model and an aromatic hydrocarbon model;
the optimal modeling principal component number f is determined by a method of drawing the principal component number by a PRESS method;
the model performance is represented by a correction correlation coefficient Rc, a correction standard deviation SEC, a prediction correlation coefficient Rp and a prediction standard deviation SEP;
s4: on-line analytical processing
Carrying out spectrum pretreatment on original spectra of samples collected on product channels of all catalytic reforming devices according to the step S2, and substituting the original spectra into the model which is built in the step S3 and can be directly transmitted for prediction, so that the samples on all the channels can be predicted by adopting the model built on one channel;
the prediction effect of the directly transferable model can be represented by the relative prediction error mean.
The reference data for modeling is not limited to RON and aromatic content, and can be the composition and properties of the oil produced by other catalytic reforming processes.
In step S1, preferably, the online near-infrared analysis system uses a fourier near-infrared spectrometer, the fourier near-infrared spectrometer uses a halogen lamp light source, light emitted from the light source passes through an interferometer and then through an optical switch, and is guided into a flow cell on a product channel of the catalytic reforming apparatus by using an optical fiber, and after passing through the flow cell, the light is sent back to the optical switch by using the optical fiber and then enters a detector, so as to obtain a near-infrared spectrum; preferably, the optical switch is provided with a plurality of optical fibers, and the flow cell on each product channel of the catalytic reforming device is connected with one optical fiber; preferably, the on-line near-infrared analysis system is also provided with a reference channel which is not connected with the catalytic reforming device, and the flow cell arranged on the channel is also connected with an optical fiber on the optical switch and is used for collecting a reference signal;
when the catalytic reforming device starts to produce, the on-line near-infrared analysis system also starts to start, and near-infrared light sequentially passes through the flow cell on the product channel of the catalytic reforming device by controlling the optical switch, so that the on-line acquisition of the near-infrared spectrum of each product channel of the catalytic reforming device is realized.
The invention relates to an online multi-channel measurement method for oil generated by a catalytic reforming process, wherein the method preferably comprises the following steps: the flow cell is arranged at the outlet position of a product pipeline of the catalytic reforming device, the pipeline direction of the product pipeline at the outlet position is vertical downward, and the flowing direction of a sample in the flow cell is vertical downward;
it is further preferred that the entire system of the on-line near-infrared analysis system shares 1 light source, 1 interferometer, 1 detector and 1 reference channel.
In step S1, the method for on-line multi-channel measurement of oil produced by a catalytic reforming process according to the present invention preferably employs a transmission method for collection of the original spectrum for modeling, and comprises the following steps:
dropwise adding a sample into the flow cell, and collecting the near infrared spectrum of the sample after the flow cell is filled; collecting the near infrared spectrum of each sample for 3 times, averaging the near infrared spectra to serve as the original spectrum for modeling;
wherein the spectrum acquisition parameters are: spectral range 10000-4000cm -1 Optical path of 0.2-4 mm, resolution of 2-32 cm -1 The scanning times are 3-6 times, and the reference scanning interval is 1-6 h/time by taking internal air as reference.
In the on-line multi-channel measurement method for the oil produced by the catalytic reforming process, in the step S2, the interval for performing the spectrum pretreatment is preferably 4800-9000cm -1 A wave band; the suction cupThe step of converting the photometric spectrum into the cosine spectrum comprises the following steps: moving a moving window with a set window width d point by point from a wavelength starting point, calculating cos theta between a reference vector a and a sample spectrum vector b in the window with the width d according to the following formula (1), and converting the absorbance value at the wavelength into a cosine value cos theta; reference vector a is a vector [1, 1., 1 ] composed of 1 s]The dimension of which is the same as the width d of the moving window;
wherein, a is a reference vector, and b is a sample spectrum vector in a d window after discrete wavelet processing.
The invention relates to an online multi-channel measurement method for oil generated by a catalytic reforming process, wherein the method preferably comprises the following steps: the method for determining the width d of the moving window comprises the following steps: randomly selecting 3 or more than 3 samples, collecting the spectrums of the samples in all the reforming device product channels according to the method in the step S1, and converting each absorbance spectrum into a cosine spectrum by using a formula (1) according to different window widths; calculating the correlation between cosine spectrum vectors h of different channels obtained by the same sample under the same window width according to the following formula (2), and selecting the width of a moving window with the maximum correlation as the width of a window for converting a spectrum into an angular spectrum;
wherein h is i And h j Respectively the cosine spectral vectors, gamma, of the ith and jth channels obtained by the same sample under the same window width ij Is h i And h j Angle between two spectral vectors, cos gamma ij Closer to 1 indicates higher spectral similarity.
The invention relates to an online multi-channel measurement method for oil generated by a catalytic reforming process, wherein the method preferably comprises the following steps: the expression of the model which can be directly transferred and is established in the step S3 is as follows:
Y’＝X’P T (T T T) -1 T T Y
wherein Y 'is aromatic hydrocarbon or octane number of a sample to be detected, X' is cosine spectrum of the sample to be detected, Y is aromatic hydrocarbon or octane number reference data vector of a correction concentrated sample, and T and P are front f (f) obtained by performing principal component analysis on X matrix of cosine spectrum of the correction concentrated sample<m) score matrix and load matrix of principal components, T T Is the transposition of T, P T Is the transposition of P, and the relation of T and P satisfies the formula X = TP T +E X Wherein, E X A residual matrix of X.
In the on-line multi-channel measuring method for the oil generated by the catalytic reforming process, in the step S3, the preferable step of selecting the range between cosine spectrums is as follows: respectively calculating the correlation R of a cosine vector X corresponding to each wavelength point in a cosine spectrum matrix X of the correction set and RON and an aromatic hydrocarbon content reference value vector y of a sample of the correction set according to the following formula (3); the interval range of the RON model of the correction concentrated sample is | R | > 0.2, and the interval range of the aromatic hydrocarbon model of the sample is | R | > 0.1;
wherein x is i Is the cosine value at a certain wavelength of the ith calibration sample,
for the average cosine value, y, of all samples at that wavelength i For the reference value of the i-th calibration sample,
the average reference value of all correction samples is referred to, and n is the number of samples in a correction set;
PRESS in the PRESS method is calculated by formula (4),
wherein, y i To correct the RON or aromatic content reference value for the ith sample in the set,
and (3) correcting the RON or aromatic hydrocarbon content model predicted value of the ith sample in the set, wherein n is the number of samples in the set.
In step S3, the calculation formulas of the corrected correlation coefficient Rc, the predicted correlation coefficient Rp, the corrected standard deviation SEC and the predicted standard deviation SEP are preferably as follows:
in the formula:
predicted value, y, obtained for the spectrum of the ith calibration sample i Is the reference value of the i-th correction sample,
predicted value, v, from the ith validation sample spectrum i For the reference value corresponding to the i-th verification sample, d is the degree of freedom of the correction model, d v Is the total number of reference values used for all v validation samples,
refers to the average reference value of all calibration samples,
mean reference values for all validation samples; n is the number of the calibration set samples; m represents the number of samples in the validation set.
In the on-line multi-channel measurement method for composition and properties of oil produced by catalytic reforming process of the present invention, in step S4, preferably, the relative error mean value calculation formula is as shown in the following formula (9):
in the formula:
(iv) RON or aromatic content prediction value, v, obtained for the ith verification sample spectrum in a certain channel i And k is the RON or arene reference data corresponding to the ith verification sample in a certain channel, and the number of the measurement samples.
Specifically, the technical scheme of the on-line multi-channel measurement method for the composition and properties of the oil generated by the catalytic reforming process is as follows:
s1: establishing a sample set, comprising the following steps: collecting a sample of oil generated by a catalytic reforming process, and respectively measuring the octane value RON and the aromatic hydrocarbon content (wt%) of the sample by adopting GB/T5487 and SH/T0166-92 standard methods as reference data for modeling; collecting the near infrared spectrum of a sample in a certain catalytic reforming device product channel by using an online near infrared analysis system to serve as an original spectrum of the sample for modeling, wherein the collected near infrared spectrum is an absorbance spectrum; establishing a sample set by using the measured reference data and the original spectrum;
s2: spectrum pretreatment, comprising the following steps: selecting a spectrum interval, and processing an original spectrum of a sample by using a discrete wavelet transform method to remove noise; taking a unit vector a as a reference vector, performing spectral measurement conversion on a sample original spectrum by using a moving window method, converting a corresponding absorbance spectrum into a cosine spectrum, wherein each sample original spectrum correspondingly obtains a cosine spectrum vector H = (cos (theta 1), cos (theta 2), \ 8230;, cos (theta m), m represents the number of wavelength points of the cosine spectrum, and all samples in a sample set form a cosine spectrum matrix H = [ cos (theta i 1), cos (theta i 2), \ 8230;, cos (theta im) ], i = (1 k), and k represents the number of the samples;
s3: establishing a model capable of being directly transferred, comprising the following steps: dividing a sample set into a correction set and a verification set according to a set proportion by using a Rank-kennard-Stone method, selecting a range between cosine spectrum regions, and associating a cosine spectrum matrix of the correction set with reference data of octane number and aromatic hydrocarbon content by using a partial least square method to establish an octane number model and an aromatic hydrocarbon model; the optimal modeling principal component number f is determined by a method of plotting the principal component number by a PRESS method (interactive verification prediction residual square sum); the model performance is represented by a correction correlation coefficient Rc, a correction standard deviation SEC, a prediction correlation coefficient Rp and a prediction standard deviation SEP;
s4: performing on-line analysis processing, namely performing spectrum pretreatment on original spectra of samples collected on all product channels of the catalytic reforming device according to the step S2, and substituting the original spectra into the model which can be directly transmitted and is established in the step S3 for prediction, so that the samples on all the channels can be predicted by adopting the model established on one channel; the prediction effect of the directly transferable model can be represented by the relative prediction error mean.
Further, in step S1, the on-line near-infrared analysis system uses a fourier near-infrared spectrometer, the spectrometer uses a halogen lamp light source, light emitted from the light source passes through the interferometer and then the optical switch, and is guided into the flow cell on the product channel of the reforming device by using the optical fiber, and after passing through the flow cell, the light is sent back to the optical switch by using the optical fiber and then enters the detector, so as to obtain a near-infrared spectrum; the optical switch is provided with a plurality of optical fibers, and the flow cell on the product channel of each catalytic reforming device is connected with one optical fiber; the online near-infrared analysis system is also provided with a reference channel which is not connected with the reforming device, and the flow cell arranged on the channel is also connected with an optical fiber on the optical switch and is used for collecting a reference signal;
when the catalytic reforming device starts to produce, the on-line near-infrared analysis system also starts to be started, and near-infrared light sequentially passes through the flow cell on the product channel of the catalytic reforming device by controlling the light switch, so that the on-line acquisition of the near-infrared spectrum of each product channel of the catalytic reforming device is realized;
the flow cell on each catalytic reforming device product channel is arranged at the outlet of the channel, the channel direction is vertical downwards, and the flow direction of a sample in the flow cell is vertical downwards; the whole system shares 1 light source, 1 interferometer, 1 detector and 1 reference channel.
Further, in step S1, the acquisition of the original spectrum for modeling uses a transmission method, and the steps are as follows: dropwise adding a sample into the flow cell, and collecting the near infrared spectrum of the sample after the flow cell is filled; collecting the near infrared spectrum of each sample for 3 times, averaging the near infrared spectra to serve as the original spectrum for modeling; the spectral acquisition parameters were: spectral range 10000-4000cm -1 Optical path of 0.2-4 mm, resolution of 2-32 cm -1 The scanning times are 3-6 times, and the reference scanning interval is 1-6 h/time by taking internal air as reference.
Further, in step S2, the interval for performing the spectrum processing is 4800-9000cm -1 A wave band; the steps of converting the spectrum into the cosine spectrum are as follows: moving point by point from a wavelength starting point by using a moving window with a set window width d, calculating cos theta between a reference vector a and a sample spectrum vector b in the width window of d according to the following formula (1), and converting the absorbance value at the wavelength into a cosine value cos theta; the reference vector a is a vector [1, 1., 1 ] consisting of 1 s]The dimension is the same as the width d of the moving window;
wherein, a is a reference vector, and b is a sample spectrum vector in a d window after discrete wavelet processing.
Further, the method for determining the width d of the moving window comprises the following steps: randomly selecting 3 or more than 3 samples, collecting the spectrums of the samples in all the reforming device product channels according to the method in the step S1, and converting each spectrum into a cosine spectrum by using a formula (1) according to different window widths; calculating the correlation between cosine spectrum vectors h of different channels obtained by the same sample under the same window width according to the following formula (2), and selecting the width of a moving window with the maximum correlation as the window width for converting the spectrum into an angular spectrum;
wherein h is i And h j Respectively the cosine spectral vectors, gamma, of the ith and jth channels obtained by the same sample under the same window width ij Is h i And h j Angle between two spectral vectors, cos gamma ij A closer result to 1 indicates a higher spectral similarity.
Further, the expression of the directly transferable model established in step S3 is:
Y’＝X’P T (T T T) -1 T T Y，
wherein Y 'is the aromatic hydrocarbon or octane number of the unknown sample to be tested, X' is the cosine spectrum of the unknown sample, Y is the reference data vector of the aromatic hydrocarbon or octane number of the calibration set sample, T and P are respectively the front f (f) obtained by analyzing the principal components of the X matrix of the cosine spectrum of the calibration set sample<m) score matrix and load matrix of principal components, T T Is the transposition of T, P T Is the transposition of P, and the relation of T and P satisfies the formula X = TP T +E X Wherein E is X The residual matrix of X.
Further, in step S3, the step of selecting the range between cosine spectrum regions is: respectively calculating the correlation R of a cosine vector X corresponding to each wavelength point in a calibration set cosine spectrum matrix X, a calibration set sample RON and an aromatic hydrocarbon content reference value vector y according to the following formula (3); selecting a range with the value of | R | not less than 0.2 from the range of the RON model of the correction set sample, and selecting a range with the value of | R | not less than 0.1 from the range of the aromatic hydrocarbon model;
wherein x is i Is the cosine value at a certain wavelength of the ith calibration sample,
is the average cosine value, y, of all samples at that wavelength i For the reference value of the i-th calibration sample,
the average reference value of all correction samples is shown, and n is the number of samples in a correction set;
PRESS in the PRESS method is calculated by formula (4),
wherein, y i To correct the RON or aromatic content reference value for the ith sample in the set,
and (3) correcting the RON or aromatic hydrocarbon content model predicted value of the ith sample in the set, wherein n is the number of samples in the set.
Further, in step S3, the correlation coefficient R is corrected c Predicting the correlation coefficient R p The calculation formulas of the corrected standard deviation SEC and the predicted standard deviation SEP are respectively as follows:
in the formula:
predicted value, y, obtained for the spectrum of the ith calibration sample i Is the reference value for the i-th correction sample,
predicted value, v, from the ith validation sample spectrum i For the reference value corresponding to the i-th verification sample, d is the degree of freedom of the correction model, d v Is the total number of reference values used for all v validation samples,
refers to the average reference value of all calibration samples,
mean reference values for all validation samples; n is the number of the calibration set samples; m represents the number of samples in the validation set.
Further, in step S4, the relative error mean value calculation formula is shown as the following formula (9):
in the formula:
RON or aromatic hydrocarbon content prediction value obtained for the ith verification sample spectrum in a certain channel, v i And k is the RON or arene reference data corresponding to the ith verification sample in a certain channel, and the number of the measurement samples.
The invention has the following beneficial effects:
the on-line multi-channel measurement method for the composition and the property of the oil generated by the catalytic reforming process converts the spectrum into the angular spectrum, and models are built again, so that the model established on one instrument or channel can be directly applied to other instruments or channels, the use is convenient, the measurement and calculation are accurate, the operability is strong, the efficiency is improved, the time and the cost are saved, and the calculation resources are reduced.
Drawings
FIG. 1 is a raw spectrum of a sample taken in channel 1 of example 1;
FIG. 2 is an angular spectrum of a sample of channel 1 of example 1;
FIG. 3 is a correlation coefficient plot of angular spectra of channel 1 and channel 2 samples of example 1 after processing with different window widths;
FIG. 4 is a schematic process flow diagram of the process of the present invention.
Detailed Description
The following examples illustrate the invention in detail: the present example is carried out on the premise of the technical scheme of the present invention, and detailed embodiments and processes are given, but the scope of the present invention is not limited to the following examples, and the experimental methods without specific conditions noted in the following examples are generally performed according to conventional conditions.
Example 1
The on-line multi-channel measurement method for the composition and the property of the oil generated by the catalytic reforming process is shown in figure 4 and comprises the following steps:
s1: creating a sample set
42 reformate samples were collected to make up the sample set. The octane number RON and the aromatic content (wt%) of the sample were measured by GB/T5487 and SH/T0166-92 respectively. The GB/T5487 is a method for measuring the octane number of gasoline, and the SH/T0166 is a method for measuring the content of C6-C9 aromatic hydrocarbon in reforming raw oil and product oil. Through detection, the minimum value of the octane value RON is 80.55, the maximum value is 101.71, the average value is 94.86, and the standard deviation is 6.66; the aromatic content (wt%) had a minimum of 52.22wt% and a maximum of 81.86wt%, with an average of 69.01 and a standard deviation of 9.05.
Collecting near infrared spectrums of all samples in a product channel 1 of a catalytic reforming device by using a near infrared online analysis system to serve as original spectrums of samples for modeling, wherein the collected near infrared spectrums are absorbance spectrums; the measured reference data and the raw spectra were used to create a sample set, as shown in FIG. 1.
The following briefly introduces an online near-infrared analysis system:
the near-infrared online analysis system comprises a Fourier near-infrared spectrometer, wherein the spectrometer uses a halogen lamp light source, light emitted from the light source passes through an interferometer and then an optical switch, is guided into a flow cell on a product channel on each catalytic reforming device by using optical fibers, passes through the flow cell, then is sent back to the optical switch by using the optical fibers, and then enters a detector to obtain a near-infrared spectrum. The optical switch is provided with a plurality of optical fibers, and the flow cell on the product channel of each catalytic reforming device is connected with one optical fiber.
The flow cell is arranged at the position of the product channel outlet of the catalytic reforming device, the direction of the product channel outlet is vertical downward, and the flowing direction of the sample in the flow cell is vertical downward. And during on-line analysis, the near infrared is controlled by the optical switch to sequentially pass through the flow cell on each catalytic reforming device product channel, so that the near infrared spectrum of the sample in each catalytic reforming device product channel is acquired on line. The system is also provided with a reference channel, and a flow cell is also arranged in the reference channel and is used for acquiring a reference signal. The whole system shares 1 light source, 1 interferometer, 1 detector and 1 reference channel.
When the sample spectrum for modeling is collected, the flow cell is rinsed for 3 times by using a sample to be tested, then the sample to be tested is dripped into the flow cell, and after the flow cell is filled with the sample, the spectrum measurement is carried out. Install a tee bend on the flow-through cell of product passageway, through the operation of tee bend, allowed to add the flow-through cell with the sample from a passageway in the tee bend in the flow-through cell top, during the rinse, the three-way valve in flow-through cell below is opened, and the sample flows out, need fill up the flow-through cell and measure the time, close the three-way valve in flow-through cell below, and the sample just can not flow out like this, is full of the back when the flow-through cell, alright in order to carry out the spectrum collection.
Loading each sample for 3 times, collecting 3 spectra, and averaging the 3 spectraThe raw spectrum of the sample was modeled. The spectral acquisition parameters were: spectral range 10000-4000cm -1 Resolution of 16.0cm -1 The reference scan interval was 6 h/time with internal air as reference.
S2: spectrum pretreatment and conversion.
Selecting 4800-9000cm -1 And (4) preprocessing the spectrum interval of the near-infrared band. Carrying out noise filtering treatment on the original spectrum of the sample by using discrete wavelet transform, wherein the mother function is db4, 11 layers are decomposed, the first 3 layers are taken as noise to be removed, high-frequency noise in the spectrum is eliminated, and then the spectrum is reconstructed, so that the noise filtering effect is achieved;
and taking the unit vector a as a reference vector, performing spectral measurement conversion on the original spectrum of the sample by using a moving window method, and converting the corresponding absorbance spectrum into a cosine spectrum.
First, the width of the moving window used for cosine spectral translation is determined: 3 samples were randomly selected and spectra of the 3 samples in each catalytic reformer product channel were collected as in step S1. Each spectrum was converted to a cosine spectrum using equation (1) using different window widths of 5-40 (interval 5), respectively. And (3) calculating the correlation between cosine spectral vectors h of different channels obtained by the same sample under the same window width according to the formula (2), wherein the result is shown in figure 2.
Wherein, a is a reference vector, and b is a sample spectrum vector in a d window after discrete wavelet processing.
Wherein h is i And h j Respectively the cosine spectral vectors, gamma, of the ith and jth channels obtained by the same sample under the same window width ij Is h i And h j Angle between two spectral vectors, cos gamma ij The closer the result is to 1, the spectral similarity is shownThe higher the sex.
Secondly, selecting a moving window width 5 with the maximum correlation to perform spectral measurement conversion on the original spectra of the samples in the sample set established in the step S1, and converting the corresponding absorbance spectra into cosine spectra, wherein the result is shown in fig. 3, and the specific steps are as follows: starting from one end of the spectrum, moving point by point from a wavelength starting point by using a moving window with the window width range of 5, calculating cos theta between a reference vector a and a sample spectrum vector b within the window width range of 5 according to a formula (1), converting the absorbance value at the wavelength into a cosine value cos theta, and converting each sample original spectrum into a cosine spectrum vector h = [ cos (theta) (theta) consisting of a group of cos theta values 1 ),cos(θ 2 ),…,cos(θ m )]All samples obtained a cosine matrix H = [ cos (θ) i1 ),cos(θ i2 ),…,cos(θ im )]I = (1. The reference vector a is a vector [1, 1., 1 ] consisting of 1 s]The dimension is the same as the width of the moving window.
S3: and establishing a direct transitive model.
Dividing a sample set into a correction set and a verification set according to a set proportion by using a Rank-kennard-Stone method (Rank-KS), selecting a range between cosine spectrum regions, and using a partial least square method (PLS) to enable a cosine spectrum matrix H = [ cos (theta) ] obtained in the step S2 to be processed i1 ),cos(θ i2 ),…,cos(θ im )]I = (1.
The step of selecting the range between cosine spectrum regions is as follows: respectively calculating the correlation R of a cosine vector X corresponding to each wavelength point in a calibration set cosine spectrum matrix X and a calibration set sample RON and an aromatic hydrocarbon content reference value vector y according to the following formula (3); selecting a range with the value of | R | not less than 0.2 from the range of the RON model of the correction set sample, and selecting a range with the value of | R | not less than 0.1 from the range of the aromatic hydrocarbon model;
wherein x is i Is the ith calibration sample at a certain pointThe cosine values at each wavelength are determined,
for the average cosine value, y, of all samples at that wavelength i For the reference value of the i-th calibration sample,
the average reference value of all correction samples is referred to, and n is the number of samples in a correction set;
and (4) calculating a correlation coefficient between the cosine matrix H and the reference data vector according to a formula (3), and selecting an interval larger than a threshold value to establish a direct transfer model.
The expression of the directly transferable model is:
Y’＝X’P T (T T T) -1 T T Y，
wherein Y 'is the aromatic hydrocarbon or octane number of unknown sample to be tested, X' is the cosine spectrum of unknown sample, Y is the reference data vector of the aromatic hydrocarbon or octane number of sample in correction set, T and P are respectively the front f (f) obtained by analyzing the principal component of X matrix of cosine spectrum of sample in correction set<m) score matrix and load matrix of principal components, T T Is the transposition of T, P T Is the transposition of P, the relation of T and P satisfies the formula X = TP T +E X Wherein E is X The residual matrix of X.
The modeling optimal principal component number f is determined by a method of plotting the principal component number by a PRESS method (sum of squares of cross-validation prediction residuals).
The PRESS is calculated by equation (4),
wherein, y i To correct the RON or aromatic content reference value for the ith sample in the set,
for correcting the RON or aromatic hydrocarbon content model predicted value of the ith sample in the set, n is the number of samples in the setAnd (4) counting.
And determining the performance of the model by comparing the corrected correlation coefficient (Rc) and the corrected root mean Square Error (SEC) of the model established by different principal component numbers, the predicted correlation coefficient (Rp) and the predicted root mean Square Error (SEP).
The corrected correlation coefficient, the predicted correlation coefficient, the corrected root mean square error and the predicted root mean square error are respectively Rc, rp, SEC and SEP, and the formulas are respectively shown in (5) to (8):
in the formula:
predicted value, y, obtained for the spectrum of the ith calibration sample i Is the reference value for the i-th correction sample,
predicted value, v, from the ith validation sample spectrum i For the reference value corresponding to the i-th verification sample, d is the degree of freedom of the correction model, d v Is the total number of reference values used for all v validation samples,
refers to the average reference value of all calibration samples,
mean reference values for all validation samples; n is the number of the calibration set samples; m represents the number of samples in the validation set.
The performance parameters of the optimal model obtained in this example are shown in table 1.
TABLE 1 RON, aromatics value optimal model Performance parameters for channel 1
S4: performing on-line analysis processing, namely performing spectrum pretreatment on original spectra of samples collected on all product channels of the catalytic reforming device according to the step S2, and substituting the original spectra into the model which can be directly transmitted and is established in the step S3 for prediction, so that the samples on all the channels can be predicted by adopting the model established on one channel; the prediction effect of the directly transferable model can be represented by the relative prediction error mean.
During operation, sample spectrums measured on the channel 2 and the channel 3 on line are converted into cosine spectrums, and cosine spectrum bands are input into the model which can be directly transmitted and is established in the step S3 for prediction, so that model transfer is realized.
The relative error mean value calculation formula is shown in the following formula (9):
in the formula:
RON or aromatic hydrocarbon content prediction value obtained for the ith verification sample spectrum in a certain channel, v i And k is the RON or arene reference data corresponding to the ith verification sample in a certain channel, and the number of the measurement samples.
The effect of channel 2 and channel 3 predicted using the transitive model was calculated using equation (9), and the results are shown in table 2.
TABLE 2 transfer Effect of the angular Spectrum-based model
The results in table 2 show that the prediction error mean values of the channel 2 and the channel 3 predicted by the normalized transferable model based on the angular spectrum transformation spectrum established on the channel 1 are obviously reduced compared with the prediction error mean values of the model without angular spectrum processing, and the prediction accuracy of the model is greatly improved.
The composition and properties of the catalytic reformate described in this invention are not limited to RON and aromatics content. The invention carries out model transmission by a spectrum standardization method for converting the absorbance spectrum into the cosine spectrum, and can be directly applied to other channels by only utilizing a model established by the spectrum collected on one product channel, thereby realizing the on-line multi-channel measurement of the composition and the property of the oil generated by the catalytic reforming process. Compared with the existing method, the model transmission method used in the method has the advantages that the mathematical relation among different channels does not need to be established through a group of standard sample sets, the model transmission can be carried out without establishing a mixed model by utilizing the spectrums of all the channels, the simultaneous online measurement of a plurality of channels is realized, the workload is small, and the popularization is easy. The method can also be used for model transmission among different instruments, and is favorable for popularization and application of the near infrared spectrum analysis technology.
The present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof, and it is therefore intended that all such changes and modifications as fall within the true spirit and scope of the invention be considered as within the following claims.
Claims (10)
1. An on-line multi-channel measuring method for oil generated by a catalytic reforming process is characterized by comprising the following steps:
s1: establishing a sample set
Collecting reformate produced by a catalytic reforming device under different process parameters as a sample, and measuring the octane number and the aromatic hydrocarbon content of the sample by adopting a standard method as reference data for modeling;
collecting a near infrared spectrum of a sample by adopting a certain channel in an online near infrared analysis system, wherein the near infrared spectrum is used as an original spectrum for modeling, and the near infrared spectrum is an absorbance spectrum;
creating a sample set using the reference data and the raw spectra;
s2: spectral pre-treatment and conversion
Selecting a spectrum interval, and preprocessing the original spectrum by using a discrete wavelet transform method to remove noise;
performing spectral measurement conversion on the spectrum after the noise is removed by using a unit vector a as a reference vector and using a moving window method, converting an absorbance spectrum into a corresponding cosine spectrum, wherein each sample spectrum correspondingly obtains a cosine spectrum vector H = (cos (theta 1), cos (theta 2), \8230;, cos (theta m)), m represents the number of wavelength points of the cosine spectrum, and all samples in a sample set form a cosine spectrum matrix H = [ cos (theta i 1), cos (theta i 2), \ 8230; cos (theta im) ], i = (1 k), and k represents the number of measured samples;
s3: building a model capable of direct transmission
Dividing a sample set into a correction set and a verification set according to the proportion of 6-9 to 4-1 by using a Rank-kennard-Stone method, selecting a range between cosine spectrum regions, respectively associating a cosine spectrum matrix of the correction set with reference data by using a partial least square method, and establishing an octane number model and an aromatic hydrocarbon model; and verifying the sample performance by using a verification set.
The optimal modeling principal component number f is determined by a method of drawing the principal component number by a PRESS method;
the model performance is represented by a correction correlation coefficient Rc, a correction standard deviation SEC, a verification correlation coefficient Rp and a verification standard deviation SEP;
s4: on-line analytical processing
Carrying out spectrum pretreatment on original spectra of samples collected on product channels of all catalytic reforming devices according to the step S2, and substituting the original spectra into the model which is built in the step S3 and can be directly transmitted for prediction, so that the samples on all the channels can be predicted by adopting the model built on one channel;
the prediction effect of the directly transferable model can be represented by the relative prediction error mean.
2. The on-line multi-channel measurement method for oil produced by the catalytic reforming process, according to claim 1, is characterized in that in step S1, the on-line near-infrared analysis system uses a fourier near-infrared spectrometer, the fourier near-infrared spectrometer uses a halogen lamp light source, light emitted from the light source is guided into a flow cell on a product channel of the catalytic reforming device through an interferometer, then through an optical switch and using an optical fiber, and after passing through the flow cell, the light is sent back to the optical switch using the optical fiber, and then enters a detector, so that a near-infrared spectrum is obtained; preferably, the optical switch is provided with a plurality of optical fibers, and the flow cell on each product channel of the catalytic reforming device is connected with one optical fiber; preferably, the on-line near-infrared analysis system is also provided with a reference channel which is not connected with the catalytic reforming device, and the flow cell arranged on the channel is also connected with an optical fiber on the optical switch and is used for collecting a reference signal;
when the catalytic reforming device starts to produce, the on-line near-infrared analysis system also starts to start, and near-infrared light sequentially passes through the flow cell on the product channel of the catalytic reforming device by controlling the optical switch, so that the on-line acquisition of the near-infrared spectrum of each product channel of the catalytic reforming device is realized.
3. The on-line multi-channel measurement method for oil produced by the catalytic reforming process according to claim 2, wherein the flow cell is installed at an outlet position of a product pipe of the catalytic reforming device, the pipe direction of the product pipe at the outlet position is vertically downward, and the flow direction of the sample in the flow cell is vertically downward;
preferably, the entire system of the online near-infrared analysis system shares 1 light source, 1 interferometer, 1 detector and 1 reference channel.
4. The on-line multi-channel measurement method for oil produced by the catalytic reforming process according to claim 2, wherein in the step S1, the raw spectrum for modeling is collected in a transmission mode, and the steps are as follows:
dropwise adding a sample into the flow cell, and collecting the near infrared spectrum of the sample after the flow cell is filled; collecting the near infrared spectrum of each sample for 3 times, averaging the near infrared spectra to serve as the original spectrum for modeling;
wherein the spectrum acquisition parameters are: spectral range 10000-4000cm -1 Optical path of 0.2-4 mm, resolution of 2-32 cm -1 The scanning times are 3-6 times, and the reference scanning interval is 1-6 h/time by taking internal air as reference.
5. The on-line multi-channel measurement method for oil produced by catalytic reforming process according to claim 1, wherein in step S2, the interval for spectral pretreatment is 4800-9000cm -1 A wave band; the step of converting the absorbance spectrum into the cosine spectrum comprises the following steps: moving point by point from a wavelength starting point by using a moving window with a set window width d, calculating cos theta between a reference vector a and a sample spectrum vector b in the width window of d according to the following formula (1), and converting the absorbance value at the wavelength into a cosine value cos theta; reference vector a is a vector [1, 1., 1 ] composed of 1 s]The dimension of which is the same as the width d of the moving window;
wherein, a is a reference vector, and b is a sample spectrum vector in a d window after discrete wavelet processing.
6. The method of claim 5, wherein the width d of the moving window is determined by: randomly selecting 3 or more than 3 samples, collecting the spectrums of the samples in all the reforming device product channels according to the method in the step S1, and converting each absorbance spectrum into a cosine spectrum by using a formula (1) according to different window widths; calculating the correlation between cosine spectrum vectors h of different channels obtained by the same sample under the same window width according to the following formula (2), and selecting the width of a moving window with the maximum correlation as the width of a window for converting a spectrum into an angular spectrum;
wherein h is i And h j Respectively the cosine spectral vectors, gamma, of the ith and jth channels obtained by the same sample under the same window width ij Is h i And h j The angle between the two spectral vectors.
7. The on-line multi-channel measurement method for oil produced by the catalytic reforming process according to claim 1, wherein the expression of the direct transferable model established in step S3 is:
Y’＝X’P T (T T T) -1 T T Y
wherein Y 'is aromatic hydrocarbon or octane number of a sample to be detected, X' is cosine spectrum of the sample to be detected, Y is aromatic hydrocarbon or octane number reference data vector of a correction concentrated sample, and T and P are front f (f) obtained by performing principal component analysis on X matrix of cosine spectrum of the correction concentrated sample<m) score matrix and load matrix of principal components, T T Is the transposition of T, P T Is the transposition of P, and the relation of T and P satisfies the formula X = TP T +E X Wherein E is X The residual matrix of X.
8. The on-line multi-channel measurement method for oil produced by the catalytic reforming process according to claim 1, wherein in step S3, the step of selecting the range between cosine spectrums is as follows: respectively calculating the correlation R of a cosine vector X corresponding to each wavelength point in a cosine spectrum matrix X of the correction set and RON and an aromatic hydrocarbon content reference value vector y of a sample of the correction set according to the following formula (3); the interval range of the RON model of the correction concentrated sample is | R | > 0.2, and the interval range of the aromatic hydrocarbon model of the sample is | R | > 0.1;
wherein x is i Is the cosine value at a certain wavelength of the ith calibration sample,
is the average cosine value, y, of all samples at that wavelength i Is the reference value of the ith calibration sample,
the average reference value of all correction samples is shown, and n is the number of samples in a correction set;
PRESS in the PRESS method is calculated by formula (4),
wherein, y i To correct the RON or aromatic content reference value for the ith sample in the set,
and (3) correcting the RON or aromatic hydrocarbon content model predicted value of the ith sample in the set, wherein n is the number of samples in the set.
9. The on-line multi-channel measurement method of oil produced by a catalytic reforming process according to claim 1, wherein: in step S3, the calculation formulas of the corrected correlation coefficient Rc, the predicted correlation coefficient Rp, the corrected standard deviation SEC, and the predicted standard deviation SEP are respectively as follows:
in the formula:
predicted value, y, obtained for the spectrum of the ith calibration sample i Is the reference value for the i-th correction sample,
predicted value, v, from the ith validation sample spectrum i For the reference value corresponding to the i-th verification sample, d is the degree of freedom of the correction model, d v Is the total number of reference values used for all v validation samples,
refers to the average reference value of all calibration samples,
mean reference values for all validation samples; n is the number of the calibration set samples; m represents the number of samples in the validation set.
10. The on-line multi-channel measurement method for the composition and properties of oil produced by catalytic reforming process according to claim 1, wherein in step S4, the relative error mean value calculation formula is shown as the following formula (9):
in the formula:
(iv) RON or aromatic content prediction value, v, obtained for the ith verification sample spectrum in a certain channel i And k is the RON or arene reference data corresponding to the ith verification sample in a certain channel, and the number of the measurement samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110456527.5A CN115248193A (en) | 2021-04-26 | 2021-04-26 | On-line multi-channel measuring method for oil generated by catalytic reforming process |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110456527.5A CN115248193A (en) | 2021-04-26 | 2021-04-26 | On-line multi-channel measuring method for oil generated by catalytic reforming process |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115248193A true CN115248193A (en) | 2022-10-28 |
Family
ID=83696553
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110456527.5A Pending CN115248193A (en) | 2021-04-26 | 2021-04-26 | On-line multi-channel measuring method for oil generated by catalytic reforming process |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115248193A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104089911A (en) * | 2014-06-27 | 2014-10-08 | 桂林电子科技大学 | Spectral model transmission method based on unary linear regression |
| CN107134499A (en) * | 2017-05-23 | 2017-09-05 | 电子科技大学 | Compound curved surface light trapping structure and preparation method thereof |
| WO2018010352A1 (en) * | 2016-07-11 | 2018-01-18 | 上海创和亿电子科技发展有限公司 | Qualitative and quantitative combined method for constructing near infrared quantitative model |
| CN107817223A (en) * | 2017-10-20 | 2018-03-20 | 华东理工大学 | The construction method of quick nondestructive real-time estimate oil property model and its application |
-
2021
- 2021-04-26 CN CN202110456527.5A patent/CN115248193A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104089911A (en) * | 2014-06-27 | 2014-10-08 | 桂林电子科技大学 | Spectral model transmission method based on unary linear regression |
| WO2018010352A1 (en) * | 2016-07-11 | 2018-01-18 | 上海创和亿电子科技发展有限公司 | Qualitative and quantitative combined method for constructing near infrared quantitative model |
| CN107134499A (en) * | 2017-05-23 | 2017-09-05 | 电子科技大学 | Compound curved surface light trapping structure and preparation method thereof |
| CN107817223A (en) * | 2017-10-20 | 2018-03-20 | 华东理工大学 | The construction method of quick nondestructive real-time estimate oil property model and its application |
Non-Patent Citations (1)
| Title |
|---|
| 陈洋等: "基于区间组合移动窗口法筛选近红外光谱信息", 应用光学, vol. 38, no. 01, 15 January 2017 (2017-01-15), pages 99 - 105 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6897071B2 (en) | Topological near infrared analysis modeling of petroleum refinery products | |
| CN102374975B (en) | Method for predicting physical property data of oil product by using near infrared spectrum | |
| AU603868B2 (en) | Method for the direct determination of octane number | |
| JP2851249B2 (en) | Method for analyzing hydrocarbons by near infrared spectroscopy | |
| AU709105B2 (en) | Method for preparing blend products | |
| US8735820B2 (en) | Direct match spectrographic determination of fuel properties | |
| CN107703097B (en) | Method for constructing model for rapidly predicting crude oil property by using near-infrared spectrometer | |
| US8481942B2 (en) | Direct match spectrographic determination of fuel properties | |
| JPH03504769A (en) | Method and apparatus for spectroscopic comparison of compositions | |
| CN104089911A (en) | Spectral model transmission method based on unary linear regression | |
| CN106680211A (en) | Method for evaluating online synchronous comparing on accuracy of online monitoring device for dissolved gas in transformer oil | |
| CN108760789A (en) | A kind of crude oil fast evaluation method | |
| WO2006126978A1 (en) | Topological near infrared analysis modeling of petroleum refinery products | |
| CN116559110A (en) | Self-adaptive near infrared spectrum transformation method based on correlation and Gaussian curve fitting | |
| CN100443883C (en) | Method for detecting hydrgenated tail-oil paraffin composition using near-infrared spectrum | |
| CN100470235C (en) | Method for measuring content of dialkene in gasoline through spectrum of near infrared light | |
| CA2154786A1 (en) | Method for prediction of physical property data of hydrocarbon products | |
| CN115248193A (en) | On-line multi-channel measuring method for oil generated by catalytic reforming process | |
| CN115436315A (en) | Near infrared spectrum-based cement additive detection method | |
| CN111103257B (en) | Method for rapidly predicting yield of monocyclic aromatic hydrocarbon in steam cracking liquid-phase oil product | |
| JP2007126663A (en) | Method for measuring conjugated diolefine content by analyzing near-infrared spectrum of sample, and application of the same to device control | |
| WO2024011687A1 (en) | Method and apparatus for establishing oil product physical property fast evaluation model | |
| CN113793652A (en) | Spectrogram chemometrics analysis method based on segmented intelligent optimization | |
| CN111829976A (en) | Method for predicting composition of gasoline fraction hydrocarbon group of crude oil by near infrared spectrum | |
| CN100451615C (en) | Method for detecting hydrogenated tail-oil cyclanes and arene composition using 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 |