Method, apparatus, electronic device and storage medium for directly predicting properties of reforming feedstock from crude oil

Abstract

The present invention relates to the technical field of near-infrared spectroscopy analysis, and is a method, device, electronic device and storage medium for directly predicting the properties of reforming feedstock from crude oil. By collecting the first near-infrared spectrum corresponding to the crude oil and inputting it into a pre-constructed spectral conversion model, the second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil is output, and then input into a pre-constructed reforming feedstock property detection model to output the property information of the reforming feedstock fraction. The present invention realizes that only by collecting the crude oil spectrum once, the spectrum of the reforming feedstock fraction can be obtained, and then the properties of the reforming feedstock fraction can be analyzed, through the pre-constructed spectral conversion model and reforming feedstock property detection model between the crude oil and the reforming feedstock fraction cut by true boiling point distillation. It can move the detection process of the feedstock properties of the reforming unit forward to before crude oil processing, quickly predict the properties of the reforming feedstock fraction after crude oil processing, and provide a huge space for optimizing and improving the operation of the reforming unit.

Description

Technical Field

[0001] The present invention relates to the technical field of near-infrared spectroscopy analysis, and is a method, device, electronic device and storage medium for directly predicting the properties of reforming feedstock from crude oil. Background Art

[0002] The catalytic reforming units of refining enterprises usually use the mixed oil of straight-run naphtha from the atmospheric and vacuum distillation unit (CDU) or hydrocracked secondary processed gasoline as feedstock. Understanding the properties of these feedstocks is very important, which directly affects the production efficiency and product quality of the reforming unit.

[0003] Under the action of the catalyst, these feedstocks undergo reactions such as dehydrogenation cyclization, hydrocracking and isomerization, so that hydrocarbon molecules are rearranged into new molecular structures, mainly producing C6 to C8 aromatic hydrocarbon products or high-octane gasoline. At the same time, the unit can use the hydrogen by-produced from reforming to supply the process units of secondary processed thermal cracking, delayed coking gasoline or diesel hydrofining.

[0004] In related technologies, some refining enterprises have begun to adopt a rapid detection method based on near-infrared spectroscopy. By scanning the sample spectrum, the key properties of the sample can be calculated by modeling (such as partial least squares regression or topology).

[0005] For example, the Chinese patent document with the publication number CN107367481A discloses a method for predicting the general properties of crude oil by line near-infrared spectroscopy. The steps include: installing a near-infrared detection system on the crude oil pipeline; selecting a crude oil sample set and establishing a calibration model for predicting the general properties of crude oil by partial least squares method; using the near-infrared detection system to measure the content data of the general properties of the crude oil to be measured in real time.

[0006] For example, the Chinese patent document with the authorization announcement number CN103364364B discloses a rapid detection method for crude oil properties based on a composite prediction technology. This method is for refining enterprises and is based on a near-infrared spectroscopy database and a corresponding crude oil property database. First, a topology modeling technology is used to determine whether the prediction by the topology modeling technology is credible. If it is credible, the topology modeling technology is used to predict the key property data of the crude oil; if it is not credible, the partial least squares method is used to predict the key property data of the crude oil to ensure the prediction accuracy of the key property data of the crude oil.

[0007] In the above-mentioned existing publicly disclosed methods, although the key properties of atmospheric and vacuum straight-run products or reforming feedstocks can be quickly detected, it is necessary to obtain the reforming feedstock sample first before analyzing its key properties. That is, before the reforming feedstock is obtained, the properties of the reforming feedstock cannot be directly predicted from crude oil, which affects production optimization. Therefore, it is necessary to study a new technology using near-infrared spectroscopy analysis and spectral transfer to construct a spectral conversion model from crude oil to reforming feedstock fractions and a reforming feedstock property detection model to directly predict the property information of reforming feedstock fractions from crude oil spectra. Summary of the Invention

[0008] The present invention provides a method, device, electronic device and storage medium for directly predicting the properties of reforming feedstock from crude oil, overcoming the deficiencies of the above-mentioned prior art, and effectively solving the problem that in the existing detection methods, it is necessary to obtain the reforming feedstock sample first before analyzing its key properties, which affects production optimization.

[0009] One of the technical solutions of the present invention is achieved by the following measures: A method for directly predicting the properties of reforming feedstock from crude oil includes:

[0010] Collect the first near-infrared spectrum corresponding to the crude oil;

[0011] Input the first near-infrared spectrum into a pre-constructed spectral conversion model to output the second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil. Among them, the spectral conversion model is constructed by using the near-infrared spectra of crude oil samples and corresponding reforming feedstock fraction samples by partial least squares method;

[0012] Input the second near-infrared spectrum into a pre-constructed reforming feedstock property detection model to output the property information of the reforming feedstock fraction. Among them, the reforming feedstock property detection model is constructed by using the near-infrared spectra and property information of reforming feedstock fraction samples and actual reforming feedstock fraction samples by partial least squares method.

[0013] The following is a further optimization or / and improvement of one of the above-mentioned technical solutions of the present invention:

[0014] If the above spectral conversion model includes sub-spectral conversion models corresponding to at least one target wavenumber point in the target wavenumber range, then inputting the first near-infrared spectrum into the pre-constructed spectral conversion model to output the second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil includes:

[0015] Extract the first spectral data related to the target wavenumber point from the first near-infrared spectrum;

[0016] Input the first spectral data into the sub-spectral conversion model corresponding to the target wavenumber points, and output the second spectral data corresponding to the reforming feedstock fraction at the target wavenumber points. The second spectral data corresponding to the reforming feedstock fraction at at least one target wavenumber point constitutes the second near-infrared spectrum.

[0017] If the above sub-spectral conversion model uses the moving window partial least squares method based on the preset number of windows for conversion, then the first spectral data related to the target wavenumber points is extracted from the first near-infrared spectrum, including:

[0018] Determine the adjacent wavenumber points centered on the target wavenumber point. The number of wavenumber points of the target wavenumber point and the adjacent wavenumber points is the same as the preset number of windows;

[0019] Extract the first spectral data corresponding to the target wavenumber point and the adjacent wavenumber points from the first near-infrared spectrum.

[0020] After collecting the first near-infrared spectrum corresponding to the crude oil, the following steps are also included:

[0021] Preprocess the target wavenumber segment of the first near-infrared spectrum. Among them, the preprocessing includes at least one of the following: wavelet transform processing, standard normal transform processing.

[0022] Before inputting the first near-infrared spectrum into the pre-constructed spectral conversion model, the following steps are also included:

[0023] Obtain at least one group of crude oil samples, and collect the third near-infrared spectrum corresponding to each group of crude oil samples;

[0024] Perform true boiling point distillation on each group of crude oil samples, and cut out the reforming feedstock fraction samples corresponding to the preset temperature range;

[0025] Collect the fourth near-infrared spectrum corresponding to each group of reforming feedstock fraction samples;

[0026] Construct a spectral conversion model according to the third near-infrared spectrum and the fourth near-infrared spectrum.

[0027] The above construction of the spectral conversion model according to the third near-infrared spectrum and the fourth near-infrared spectrum includes:

[0028] Determine the target wavenumber segment, and the target wavenumber segment includes at least one target wavenumber point;

[0029] For each target wavenumber point in the target wavenumber segment, extract the third spectral data related to the target wavenumber point from the third near-infrared spectrum, and extract the fourth spectral data corresponding to the target wavenumber point from the fourth near-infrared spectrum;

[0030] Construct a sub-spectrum conversion model corresponding to the target wavenumber points based on the third spectral data and the fourth spectral data, and the sub-spectrum conversion models corresponding to at least one target wavenumber point constitute the spectral conversion model.

[0031] The above-mentioned extraction of the third spectral data related to the target wavenumber points from the third near-infrared spectrum includes:

[0032] Determine the adjacent wavenumber points centered on the target wavenumber points, and the number of wavenumber points of the target wavenumber points and the adjacent wavenumber points is consistent with the preset window number.

[0033] Extract the third spectral data corresponding to the target wavenumber points and the adjacent wavenumber points from the third near-infrared spectrum;

[0034] The above-mentioned extraction of the fourth spectral data related to the target wavenumber points from the fourth near-infrared spectrum includes:

[0035] Determine the adjacent wavenumber points centered on the target wavenumber points, and the number of wavenumber points of the target wavenumber points and the adjacent wavenumber points is consistent with the preset window number.

[0036] Extract the fourth spectral data corresponding to the target wavenumber points and the adjacent wavenumber points from the fourth near-infrared spectrum;

[0037] Construct a sub-spectrum conversion model corresponding to the target wavenumber points based on the third spectral data and the fourth spectral data, including:

[0038] According to the third spectral data and the fourth spectral data, use the moving window least squares transformation method with the preset window number to construct the sub-spectrum conversion model.

[0039] The above-mentioned determination of the target wavenumber range includes:

[0040] Obtain the first property information corresponding to each group of reforming feedstock fraction samples;

[0041] According to the fourth near-infrared spectrum and the first property information of each group of reforming feedstock fraction samples, construct the relevant weight coefficients corresponding to each wavenumber point of the reforming feedstock fraction samples;

[0042] Determine the wavenumber range with the relevant weight coefficients greater than the preset value as the target wavenumber range.

[0043] Before the above-mentioned construction of the spectral conversion model based on the third near-infrared spectrum and the fourth near-infrared spectrum, it also includes:

[0044] Perform preprocessing on the target wavenumber range of the third near-infrared spectrum and the fourth near-infrared spectrum, and the preprocessing includes at least one of the following: wavelet transform processing, standard normal transform processing.

[0045] Before the above-mentioned input of the second near-infrared spectrum into the pre-constructed reforming feedstock property detection model, it also includes:

[0046] Obtain the first property information corresponding to each group of reforming feedstock fraction samples;

[0047] Collect the fifth near-infrared spectrum corresponding to at least one group of actual reforming feedstock fraction samples, and obtain the second property information corresponding to each group of actual reforming feedstock fraction samples;

[0048] Construct a reforming feedstock property detection model based on the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information.

[0049] The above-mentioned construction of the reforming feedstock property detection model based on the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information includes:

[0050] Construct a reforming feedstock property detection model using the partial least squares method based on the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information.

[0051] Before the above-mentioned construction of the reforming feedstock property detection model, it further includes:

[0052] Preprocess the target wavenumber range of the fourth near-infrared spectrum and the fifth near-infrared spectrum. The preprocessing includes at least one of the following: wavelet transform processing, standard normal transform processing.

[0053] The second technical solution of the present invention is achieved by the following measures: An apparatus for applying a method for directly predicting the properties of reforming feedstock from crude oil, including:

[0054] A collection module for collecting the first near-infrared spectrum corresponding to crude oil;

[0055] A conversion module for inputting the first near-infrared spectrum into a pre-constructed spectral conversion model and outputting the second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil;

[0056] A prediction module for inputting the second near-infrared spectrum into a pre-constructed reforming feedstock property detection model and outputting the property information of the reforming feedstock fraction.

[0057] The third technical solution of the present invention is achieved by the following measures: An electronic device includes a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus;

[0058] The memory is used to store a computer program;

[0059] When the processor is used to execute the program stored on the memory, it realizes the steps of the method for directly predicting the properties of reforming feedstock from crude oil.

[0060] The fourth technical solution of the present invention is achieved by the following measures: A storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a method for directly predicting the properties of a reforming feedstock from crude oil are realized.

[0061] Through a pre-constructed spectral conversion model and a reforming feedstock property detection model between crude oil and the reforming feedstock fractions cut by true boiling point, the present invention realizes that by collecting the spectrum of crude oil only once, the spectrum of the reforming feedstock fractions can be obtained, and then the properties of the reforming feedstock fractions can be analyzed. This method advances the process of detecting the properties of the feedstock of the reforming unit to before crude oil processing, can quickly predict the properties of the reforming feedstock fractions after crude oil processing, and provides a huge space for optimizing and improving the operation of the reforming unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Attached Figure 1 is a schematic flow chart of a method for directly predicting the properties of a reforming feedstock from crude oil provided by an embodiment of the present invention;

[0063] Attached Figure 2 is Figure 1 a detailed flow chart of step S102 in the illustrated embodiment;

[0064] Attached Figure 3 is a schematic flow chart of a method for constructing a spectral conversion model provided by an embodiment of the present invention;

[0065] Attached Figure 4 is Figure 3 a detailed flow chart of step S304 in the illustrated embodiment;

[0066] Attached Figure 5 is a schematic flow chart of a method for constructing a reforming feedstock property detection model provided by an embodiment of the present invention;

[0067] Figure 6 is a schematic flow chart of another method for directly predicting the properties of a reforming feedstock from crude oil provided by an embodiment of the present invention, where attached Figure 6-1 is a schematic flow chart of the method in the model construction stage, and attached Figure 6-2 is a schematic flow chart of the method in the model application stage;

[0068] Attached Figure 7 is a near-infrared spectrum diagram of 20 groups of crude oil samples provided by an embodiment of the present invention ;

[0069] Attached Figure 8 is a near-infrared spectrum diagram of reforming feedstock fraction samples cut from 20 groups of crude oil samples provided by an embodiment of the present invention ;

[0070] Attached Figure 9The spectrogram after preprocessing the near-infrared spectra of 20 groups of crude oil samples provided by the embodiments of the present invention ;

[0071] Appendix Figure 10 The spectrogram after preprocessing the near-infrared spectra of the reforming feedstock fraction samples cut from 20 groups of crude oil samples provided by the embodiments of the present invention ;

[0072] Appendix Figure 11 The structural schematic diagram of a device for directly predicting the properties of reforming feedstock from crude oil provided by the embodiments of the present invention;

[0073] Appendix Figure 12 The structural schematic diagram of an electronic device provided by the embodiments of the present invention. Specific embodiments

[0074] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0075] Figure 1 The flowchart of a method for directly predicting the properties of reforming feedstock from crude oil provided by the embodiments of the present invention, as Figure 1 shown, the method includes:

[0076] Step S101, collecting the first near-infrared spectrum corresponding to the crude oil.

[0077] Specifically, a near-infrared (NIR) spectroscopic instrument can be used to scan the crude oil to obtain the absorption characteristics of the crude oil in the near-infrared spectral range, that is, the first near-infrared spectrum. In addition, when collecting the spectrum of the crude oil, a crude oil automatic preprocessing system can be used to filter, keep the temperature constant, and keep the pressure constant for the crude oil, and the spectrum collection conditions are set as shown in Table 1.

[0078] Table 1

[0079] .

[0080] Step S102, inputting the first near-infrared spectrum into a pre-constructed spectral conversion model, and outputting the second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil.

[0081] Specifically, the pre-constructed spectral conversion model (or spectral transfer model) can convert the near-infrared spectrum of crude oil into the near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil. In this step, by inputting the first near-infrared spectrum of the crude oil into the spectral conversion model, the spectral information of the crude oil is converted into the second near-infrared spectrum corresponding to the reforming feedstock fraction.

[0082] Step S103: Input the second near-infrared spectrum into the pre-constructed reforming feedstock property detection model, and output the property information of the reforming feedstock fraction.

[0083] Specifically, the pre-constructed reforming feedstock property detection model can predict the property information of the reforming feedstock fraction based on the input near-infrared spectrum, such as density, distillation range, etc. In this step, the second near-infrared spectrum corresponding to the reforming feedstock fraction obtained in step S102 is input into this reforming feedstock property detection model to predict the property information of the reforming feedstock fraction.

[0084] In some embodiments, the spectral conversion model includes sub-spectral conversion models corresponding to at least one target wavenumber point in the target wavenumber range. Figure 2 For Figure 1 a detailed flowchart of step S102 in the illustrated embodiment, as Figure 2 shown, step S102 includes:

[0085] Step S1021: Extract the first spectral data related to the target wavenumber point from the first near-infrared spectrum.

[0086] Step S1022: Input the first spectral data into the sub-spectral conversion model corresponding to the target wavenumber point, and output the second spectral data corresponding to the reforming feedstock fraction at the target wavenumber point. The second spectral data corresponding to the reforming feedstock fraction at at least one target wavenumber point constitutes the second near-infrared spectrum.

[0087] Specifically, the target wavenumber range can be understood as a characteristic spectral range that has a greater impact on the key properties of the reforming feedstock fraction, such as 4000 cm -1 to 4800 cm -1 , 5600 cm -1 to 6200 cm -1 characteristic spectral ranges. Each characteristic spectral range includes multiple target wavenumber points, such as 5830 cm -1 , 5832 cm -1 , 5834 cm -1 , etc.; the sub-spectral conversion model can be understood as a spectral conversion model corresponding to one wavenumber point, such as the sub-spectral conversion model corresponding to 5832 cm -1 ; the first spectral data and the second spectral data can be understood as relevant parameters in the near-infrared spectrum, such as absorbance, etc.

[0088] In this embodiment, first, absorbance data related to the target wavenumber points is extracted from the first near-infrared spectrum of the crude oil. Herein, the absorbance data related to the target wavenumber points can be understood as the absorbance corresponding to the target wavenumber points and the associated (usually adjacent wavenumber points) wavenumber points related to the target wavenumber points. Then, the absorbance related to the target wavenumber points is input into the sub-spectrum conversion model to obtain the absorbance of the reforming feed fraction corresponding to the target wavenumber points. By analogy, the absorbance of the reforming feed fraction corresponding to multiple target wavenumber points is obtained, constituting the second near-infrared spectrum of the reforming feed fraction.

[0089] In some embodiments, if the sub-spectrum conversion model adopts the moving window partial least squares method conversion based on the preset window number, then step S1021 includes: determining the adjacent wavenumber points centered on the target wavenumber points, where the number of wavenumber points of the target wavenumber points and the adjacent wavenumber points is consistent with the preset window number; extracting the first spectral data corresponding to the target wavenumber points and the adjacent wavenumber points from the first near-infrared spectrum.

[0090] Specifically, the sub-spectrum conversion model is constructed by the moving window partial least squares method conversion based on the preset window number. Taking the sub-spectrum conversion model corresponding to the wavenumber point of 5832 cm as an example with the preset window number k = 7 -1 for illustration. First, centered on the wavenumber point of 5832 cm -1 its adjacent wavenumber points are determined to be 5826 cm -1 , 5828 cm -1 , 5830 cm -1 , 5834 cm -1 , 5836 cm -1 , 5838 cm -1 . There are a total of 7 wavenumber points for the target wavenumber points and the adjacent wavenumber points, which is consistent with the preset window number k = 7. The absorbance corresponding to these 7 wavenumber points is extracted from the first near-infrared spectrum of the crude oil and input into the sub-spectrum conversion model corresponding to the wavenumber point of 5832 cm -1 to output the absorbance of the reforming feed fraction at the wavenumber point of 5832 cm -1 . By analogy, the absorbance of the reforming feed fraction corresponding to the target wavenumber range can be output, constituting the second near-infrared spectrum of the reforming feed fraction.

[0091] In some embodiments, after step S101, it further includes: preprocessing the target wavenumber range of the first near-infrared spectrum, and the preprocessing includes at least one of the following: wavelet transform processing, standard normal transform processing.

[0092] Specifically, the first near-infrared spectrum of the crude oil is preprocessed. To improve efficiency, the target wavenumber band of the first near-infrared spectrum is preprocessed. First, wavelet transform (WT) processing is performed, which can be used to eliminate spectral noise. The general form of the wavelet function is shown in formula (1):

[0093]

[0094] where s represents the scale, which is used to control the stretching and shrinking of the wavelet function and corresponds to the frequency (inverse ratio); represents the translation amount, which is used to control the translation of the wavelet function and corresponds to time, represents the result of the wavelet transform, which is a function of s and ; represents the original signal to be analyzed, which is a function of time t; represents the wavelet basis function after scale and displacement transformation; represents the convolution operation; represents the integral operation from negative to positive infinity with respect to time t.

[0095] Then, based on the wavelet transform, standard normal variate (SNV) processing is performed, which can be used to remove spectral signal variations, as shown in formula (2):

[0096]

[0097] where represents the near-infrared spectrum at the j th target wavenumber point, , m is the number of target wavenumber points corresponding to the target wavenumber band, represents the near-infrared spectrum at the j th target wavenumber point after standard normal variate processing.

[0098] Finally, the preprocessed first near-infrared spectrum is input into the spectral conversion model to obtain the second near-infrared spectrum corresponding to the reforming feed fraction; the second near-infrared spectrum is input into the reforming feed property detection model to predict the properties of the reforming feed fraction.

[0099] The method for directly predicting the properties of reforming feedstock from crude oil provided by the embodiments of the present invention includes: collecting the first near-infrared spectrum corresponding to the crude oil; inputting the first near-infrared spectrum into a pre-constructed spectral conversion model to output the second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil; inputting the second near-infrared spectrum into a pre-constructed reforming feedstock property detection model to output the property information of the reforming feedstock fraction. That is, through the pre-constructed spectral conversion model and reforming feedstock property detection model between the crude oil and the reforming feedstock fraction cut by true boiling point distillation, it is realized that only by collecting the crude oil spectrum once, the spectrum of the reforming feedstock fraction can be obtained, and then the key properties of the reforming feedstock fraction can be predicted. This method advances the process of detecting the properties of the feedstock of the reforming unit to before the crude oil processing. After the crude oil is received and stored in the factory, the properties of the reforming feedstock fraction after the crude oil processing can be quickly predicted, providing a huge space for optimizing and improving the operation of the reforming unit.

[0100] Based on the foregoing embodiments, Figure 3 The flowchart of a method for constructing a spectral conversion model provided by the embodiments of the present invention is shown as Figure 3 shown, and the method includes:

[0101] Step S301: Obtain at least one group of crude oil samples and collect the third near-infrared spectrum corresponding to each group of crude oil samples.

[0102] Step S302: Perform true boiling point distillation on each group of crude oil samples to cut out the reforming feedstock fraction samples corresponding to a preset temperature range.

[0103] Step S303: Collect the fourth near-infrared spectrum corresponding to each group of reforming feedstock fraction samples.

[0104] Step S304: Construct a spectral conversion model according to the third near-infrared spectrum and the fourth near-infrared spectrum.

[0105] Specifically, before performing step S102, it is necessary to pre-construct a spectral conversion model by performing steps S301 - S304 first. In this embodiment, first, multiple groups of crude oil samples are obtained, and the near-infrared spectra of each group of crude oil samples are collected. Among them, the crude oil samples can be typical processed crude oils of refining enterprises, including incoming component crude oils, incoming tank mixed crude oils, incoming unit mixed crude oils, etc., which basically cover the quality changes during the current crude oil storage, transfer, and processing processes of the enterprise. When collecting the near-infrared spectra of the crude oil samples, a crude oil automatic pretreatment system can be used to filter, maintain constant temperature, and maintain constant pressure on the crude oil samples, and spectral acquisition is performed based on the spectral acquisition conditions shown in Table 1. Then, an enterprise laboratory true boiling point distillation apparatus can be used to perform true boiling point atmospheric distillation on each group of crude oil samples. According to the standard heating method, each group of crude oil samples distills out a reforming feed fraction at 15 - 140 °C under the same operating conditions. Then, based on the same spectral acquisition conditions as in Table 1, the near-infrared spectra of the reforming feed fractions cut by the enterprise laboratory are collected. Finally, a spectral conversion model is constructed based on the near-infrared spectra of the crude oil and the near-infrared spectra of the reforming feed fractions.

[0106] Figure 4 For Figure 3 a detailed flowchart of step S304 in the embodiment shown, as Figure 4 shown, step S304 includes:

[0107] Step S3041, determine the target wavenumber range, and the target wavenumber range includes at least one target wavenumber point.

[0108] Step S3042, for each target wavenumber point in the target wavenumber range, extract the third spectral data related to the target wavenumber point from the third near-infrared spectrum, and extract the fourth spectral data corresponding to the target wavenumber point from the fourth near-infrared spectrum.

[0109] Step S3043, construct a sub-spectral conversion model corresponding to the target wavenumber point according to the third spectral data and the fourth spectral data, and the sub-spectral conversion models corresponding to at least one target wavenumber point constitute the spectral conversion model.

[0110] Specifically, first determine the target wavenumber range that has a greater impact on the key properties of the reforming feed fraction, such as 4000 cm -1 to 4800 cm -1 、5600 cm -1 to 6200 cm -1Characteristic spectral band; then, a sub-spectral conversion model corresponding to each target wavenumber point in the target wavenumber band is constructed as follows: Third spectral data related to the target wavenumber point is extracted from the third near-infrared spectrum corresponding to each group of crude oil samples, and fourth spectral data corresponding to the target wavenumber point is extracted from the fourth near-infrared spectrum corresponding to each group of reforming feedstock fraction samples; a sub-spectral conversion model at the target wavenumber point is constructed based on each third spectral data and each fourth spectral data; by analogy, the sub-spectral conversion models of multiple target wavenumber points constitute a spectral conversion model.

[0111] In some embodiments, extracting the third spectral data related to the target wavenumber point from the third near-infrared spectrum in step S3042 includes: determining adjacent wavenumber points centered on the target wavenumber point, and the number of wavenumber points of the target wavenumber point and the adjacent wavenumber points is the same as the preset window number; extracting the third spectral data corresponding to the target wavenumber point and the adjacent wavenumber points from the third near-infrared spectrum; then step S3043 includes: constructing a sub-spectral conversion model according to the third spectral data and the fourth spectral data by using a moving window least squares conversion method with the preset window number.

[0112] Specifically, the spectral conversion model from crude oil to reforming feedstock fraction is optimized and calculated by using the moving window partial least squares method, and the window number is preset , and the established spectral conversion model corresponding coefficient matrix , as shown in formula (3):

[0113]

[0114] Among them, each column in represents the coefficient matrix of the sub-spectral conversion model corresponding to the j th target wavenumber point, that is , among which, j = 1, 2, …… m， For example, substituting j = 1 into , we get , which exactly corresponds to the first column coefficient of the coefficient matrix .

[0115] When the preset window number k = 7, converting the crude oil spectrum to the spectrum corresponding to the reforming feedstock fraction, as shown in formula (4), represents that according to the i th crude oil sample at the j -3rd wavenumber point, the j -2nd wavenumber point, the j -1st wavenumber point, the j th wavenumber point, the j +1st wavenumber point, the j +2nd wavenumber point, thej The sum of the products of the absorbances at three wavenumber points and the coefficient matrix of the sub-spectral conversion model corresponding to the j th target wavenumber point is obtained, and the absorbance of the reforming feed fraction sample cut from the i th crude oil sample at the j th wavenumber point is obtained ; When constructing the spectral conversion model, the objective function used is as shown in formula (5). When the objective function converges to a minimum value, it indicates that the spectral conversion model construction is completed.

[0116]

[0117]

[0118] Among them, n represents the number of crude oil samples, m represents the number of target wavenumber points in the target wavenumber range of the crude oil and reforming feed fraction, represents the absorbance of the i th reforming feed fraction sample at the j th target wavenumber point, represents the absorbance of the reforming feed fraction sample predicted after the i th crude oil sample passes through the spectral conversion model at the j th target wavenumber point;

[0119] represents the absorbance of the i-th crude oil sample at the to wavenumber points. When k = 7, it represents the absorbance of the i-th crude oil sample at the wavenumber points from the (j - 3)-th to the (j + 3)-th wavenumber points; represents the coefficient of the sub-spectral conversion model corresponding to the j-th target wavenumber point, corresponding to the j-th column in the coefficient matrix ; ; represents finding the n crude oil samples m the objective function of minimizing the sum of the squared differences of the absorbances before and after spectral conversion at

[0120] In some embodiments, before step S304, it further includes: preprocessing the target wavenumber ranges of the third near-infrared spectrum and the fourth near-infrared spectrum, and the preprocessing includes at least one of the following: wavelet transform processing, standard normal variate transformation processing.

[0121] Specifically, the third near-infrared spectrum of each group of crude oil samples is subjected to wavelet transform (WT) and standard normal variate transform (SNV) to obtain the preprocessed third near-infrared spectrum as shown in formula (6); the fourth near-infrared spectrum of each group of reforming feedstock fraction samples is subjected to wavelet transform and standard normal variate transform to obtain the preprocessed fourth near-infrared spectrum as shown in formula (7); the constructed spectral conversion model is as shown in formula (8), indicating that the preprocessed third near-infrared spectrum is input into the spectral conversion model to predict the near-infrared spectrum of each group of reforming feedstock fraction samples; when constructing the spectral conversion model using the preprocessed third near-infrared spectrum and fourth near-infrared spectrum, the objective function adopted can be further transformed from formula (5) to formula (9) as follows:

[0122]

[0123]

[0124]

[0125]

[0126] Among them, represents the absorbance matrix of the th target wavenumber point of the crude oil sample after preprocessing with the window number , respectively represent the absorbances of the to wavenumber points of the first crude oil sample after preprocessing. For example, when k = 7, it represents the absorbances of the j - 3 to j + 3 wavenumber points of the first crude oil sample after preprocessing. Similarly, respectively represent the absorbances of the to wavenumber points of the nth crude oil sample after preprocessing. For example, when k = 7, it represents the absorbances of the j - 3 to j + 3 wavenumber points of the nth crude oil sample after preprocessing; represents the absorbance of the th target wavenumber point of the reforming feedstock fraction actually cut from the crude oil after preprocessing, respectively represent the absorbances of the th target wavenumber points of the reforming feedstock fractions cut from the first to nth crude oil samples after preprocessing, is the absorbance matrix of the jth target wavenumber point of the reforming feedstock fraction predicted based on the spectral conversion model, where represents the coefficient matrix of the spectral conversion model, represents the constant matrix; represents obtaining a single crude oil sample mThe objective function for minimizing the sum of the squares of the differences in absorbance before and after spectral conversion at a target wavenumber point.

[0127] In some embodiments, step S3041 includes: obtaining first property information corresponding to each group of reforming feedstock fraction samples; constructing weight coefficients related to each wavenumber point corresponding to the reforming feedstock fraction samples according to the fourth near-infrared spectrum and the first property information of each group of reforming feedstock fraction samples; and determining the wavenumber segment with weight coefficients greater than a preset value as the target wavenumber segment.

[0128] Specifically, first, the key properties of each group of reforming feedstock fraction samples can be analyzed by standard methods, and then, based on the properties and near-infrared spectra of these multiple groups of reforming feedstock fraction samples, a curve of weight coefficients related to each wavenumber point is established, as shown in formula (10):

[0129]

[0130] where j represents the j th wavenumber point, p represents the pth property, is the covariance between and is the variance of is the variance of

[0131] Finally, the wavenumber segment with the correlation coefficient is determined as the target wavenumber segment. For example, it is determined that 4000 cm -1 to 4800 cm -1 and 5600 cm -1 to 6200 cm -1 are the target wavenumber segments.

[0132] Based on the foregoing embodiments, by obtaining at least one group of crude oil samples and collecting the corresponding third near-infrared spectrum of each group of crude oil samples; performing true boiling point distillation on each group of crude oil samples to cut out the corresponding reforming feedstock fraction samples within a preset temperature range; collecting the corresponding fourth near-infrared spectrum of each group of reforming feedstock fraction samples; and constructing a spectral conversion model according to the third near-infrared spectrum and the fourth near-infrared spectrum, a spectral conversion model between the crude oil spectrum and the reforming feedstock fraction spectrum is constructed, providing a basis for directly predicting the properties of reforming feedstock fractions from crude oil in the subsequent process.

[0133] Based on the foregoing embodiments, Figure 5 is a schematic flowchart of a method for constructing a reforming feedstock property detection model provided by an embodiment of the present invention. As Figure 5 shown, the method includes:

[0134] Step S501: Obtain the first property information corresponding to each group of reforming feedstock fraction samples.

[0135] Step S502: Collect the fifth near-infrared spectra corresponding to at least one group of actual reforming feedstock fraction samples, and obtain the second property information corresponding to each group of actual reforming feedstock fraction samples.

[0136] Step S503: Construct a reforming feedstock property detection model based on the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information.

[0137] Specifically, before performing step S103, it is necessary to first pre-construct a reforming feedstock property detection model by performing steps S501 - S503. First, the first property information of the reforming feedstock fraction samples cut from the true boiling point of each group of crude oil samples can be obtained based on standard analysis methods, such as density, distillation range, etc.; then collect the fifth near-infrared spectra of multiple groups of actual reforming feedstock fractions in the reforming unit, and the spectrum collection conditions are shown in Table 1, and obtain the second property information of each group of actual reforming feedstock fractions based on standard analysis methods; finally, establish a reforming feedstock property detection model based on the fourth near-infrared spectrum + the fifth near-infrared spectrum and their associated first property information + second property information.

[0138] In some embodiments, step S503 includes: constructing a reforming feedstock property detection model using the partial least squares method based on the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information.

[0139] In some embodiments, before step S503, it further includes: preprocessing the target wavenumber range of the fourth near-infrared spectrum and the fifth near-infrared spectrum, and the preprocessing includes at least one of the following: wavelet transform processing, standard normal variate transform processing.

[0140] Specifically, after obtaining the fourth near-infrared spectrum and the first property information corresponding to the reforming feedstock fraction samples cut from the true boiling point of the crude oil, and the fifth near-infrared spectrum and the second property information corresponding to the actual reforming feedstock fraction samples, perform the same preprocessing on the fourth near-infrared spectrum and the fifth near-infrared spectrum, that is, wavelet transform and standard normal variate transform processing; use the preprocessed fourth near-infrared spectrum and fifth near-infrared spectrum and the corresponding first property information and second property information to construct a reforming feedstock property detection model using the partial least squares method.

[0141] On the basis of the foregoing embodiments, by obtaining the first property information corresponding to each group of reforming feedstock fraction samples; collecting the fifth near-infrared spectrum corresponding to at least one group of actual reforming feedstock fraction samples, and obtaining the second property information corresponding to each group of actual reforming feedstock fraction samples; constructing a reforming feedstock property detection model according to the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information; the pre-construction of the reforming feedstock property detection model is realized, which provides a basis for subsequently directly predicting the properties of reforming feedstock fractions from crude oil.

[0142] To further understand the embodiments of the present invention, FIG. 6 is a schematic flowchart of another method for directly predicting the properties of reforming feedstock from crude oil provided by the embodiments of the present invention. As Figure 6-1 and Figure 6-2 shown, the method includes:

[0143] Step S601, obtain at least one group of crude oil samples, and collect the third near-infrared spectrum corresponding to each group of crude oil samples.

[0144] Specifically, collect 20 groups of typical processed crude oils from refining enterprises, including incoming component crude oils, in-tank mixed crude oils, in-unit mixed crude oils, etc. These crude oil samples basically cover the quality changes during the crude oil storage, transfer, and processing in refining enterprises; and use an automatic pre-treatment device for crude oil samples to filter, keep the temperature constant, and keep the pressure constant for the crude oil samples, and perform spectrum collection under the spectrum collection conditions shown in Table 1, as Figure 7 is a near-infrared spectrogram of 20 groups of crude oil samples provided by the embodiments of the present invention .

[0145] Step S602, perform true boiling point distillation on each group of crude oil samples, and cut out the reforming feedstock fraction samples corresponding to a preset temperature range.

[0146] Specifically, a laboratory true boiling point distillation apparatus of a refining enterprise can be used to perform true boiling point atmospheric distillation on 20 groups of crude oil samples. According to the standard heating method, each group of crude oil samples distills out fractions at 15°C to 140°C under the same operating conditions.

[0147] Step S603, collect the fourth near-infrared spectrum corresponding to each group of reforming feedstock fraction samples, and obtain the first property information corresponding to each group of reforming feedstock fraction samples.

[0148] Specifically, based on the same spectrum collection conditions as in step S601, collect the near-infrared spectrogram of the reforming feedstock fractions cut from the crude oil in the laboratory at 15°C to 140°C, as Figure 8 is a near-infrared spectrogram of reforming feedstock fraction samples cut from 20 groups of crude oil samples provided by the embodiments of the present invention ; use a standard method (shown in Table 2) to analyze and test the key properties of the reforming feedstock fractions , for example, to obtain the density of the reforming feed fraction, the standard method of GB / T 29617 is used; to obtain the normal paraffins, isoparaffins, naphthenes, and aromatics in the reforming feed fraction, the standard method of Q / SY DS 04.018 is used; to obtain the distillation temperatures of the reforming feed fraction, the standard method of GB / T 6536 is used.

[0149] Table 2

[0150] .

[0151] Step S604: According to the fourth near-infrared spectrum and the first property information of each group of reforming feed fraction samples, construct the relevant weight coefficients corresponding to each wavenumber point of the reforming feed fraction samples.

[0152] Specifically, based on the near-infrared spectra of 20 groups of reforming feed fraction samples and the key properties of the reforming feed fraction , establish the relevant weight coefficient curves corresponding to each wavenumber point, as shown in formula (10).

[0153] Step S605: Determine the wavenumber segments with relevant weight coefficients greater than the preset value as the target wavenumber segments, and the target wavenumber segments include at least one target wavenumber point.

[0154] Specifically, count the target wavenumber segments of the correlation coefficient , and finally determine that the wavenumber range from 4000 cm -1 to 4800 cm -1 , and the wavenumber range from 5600 cm -1 to 6200 cm -1 are the target wavenumber segments, and the target wavenumber segments include multiple target wavenumber points, such as 5830 cm -1 , 5832 cm -1 , 5834 cm -1 , etc.

[0155] Step S606: Preprocess the target wavenumber segments of the third near-infrared spectrum and the fourth near-infrared spectrum, and the preprocessing includes at least one of the following: wavelet transform processing, standard normal transform processing.

[0156] Specifically, perform wavelet transform (WT) and standard normal transform (SNV) on the target wavenumber segments of the third near-infrared spectra of 20 groups of crude oil samples successively, as Figure 9 is the spectrum diagram after preprocessing the near-infrared spectra of 20 groups of crude oil samples provided by an embodiment of the present invention ; perform wavelet transform (WT) plus standard normal transform (SNV) on the fourth near-infrared spectrum diagrams of 20 groups of laboratory-cut reforming feed fractions of crude oil successively, as Figure 10The spectrogram after preprocessing the near-infrared spectra of the reforming feedstock fraction samples cut from 20 groups of crude oil samples provided by the embodiments of the present invention .

[0157] Step S607: For each target wavenumber point in the target wavenumber band, determine the adjacent wavenumber points centered on the target wavenumber point. The number of wavenumber points of the target wavenumber point and the adjacent wavenumber points is consistent with the preset window number, and extract the third spectral data corresponding to the target wavenumber point and the adjacent wavenumber points from the third near-infrared spectrum.

[0158] Step S608: Extract the fourth spectral data corresponding to the target wavenumber point from the fourth near-infrared spectrum.

[0159] Step S609: According to the third spectral data and the fourth spectral data, adopt the moving window least squares conversion method with the preset window number to construct a sub-spectrum conversion model, and the sub-spectrum conversion models corresponding to at least one target wavenumber point constitute a spectrum conversion model.

[0160] Specifically, based on the preprocessed crude oil spectrum and its reforming feedstock fraction spectrum, a moving window partial least squares conversion model with a preset window number of 7 is established , and taking the establishment of the sub-spectrum conversion model at 5832 cm -1 as an example for illustration.

[0161] At 5832 cm -1 , the absorbance matrix after preprocessing the crude oil with a preset window number of 7 is shown in Table 3, where Crude01-Crude20 represent the 1st group of crude oil samples - the 20th group of crude oil samples respectively, and each group of crude oil samples includes 7 target wavenumber points of 5826, 5828, 5830, 5832, 5834, 5836, and 5838.

[0162] Table 3

[0163] .

[0164] The absorbance matrix corresponding to the wavenumber point at 5832 cm -1 after preprocessing the reforming feedstock fraction cut from each group of crude oil in the laboratory is shown in Table 4:

[0165] Table 4

[0166] .

[0167] According to Table 1 and Table 2, at 5832 cm -1 , the sub-spectrum conversion model established by adopting the moving window least squares conversion method with the preset window number is as follows:

[0168] .

[0169] And so on, thereby establishing a sub-spectral conversion model corresponding to each wavenumber point in the target wavenumber range, and constituting a spectral conversion model.

[0170] Step S610: Collect the fifth near-infrared spectrum corresponding to at least one set of actual reforming feedstock fraction samples, and obtain the second property information corresponding to each set of actual reforming feedstock fraction samples.

[0171] Specifically, collect the fifth near-infrared spectra of 50 sets of atmospheric and vacuum distillation unit primary overhead naphtha , and analyze its key properties using the standard method shown in Table 2 .

[0172] Step S611: Preprocess the target wavenumber range of the fifth near-infrared spectrum, and the preprocessing includes at least one of the following: wavelet transform processing, standard normal variate transform processing.

[0173] Specifically, perform the same preprocessing on the fifth near-infrared spectra of 50 sets of primary overhead naphtha as the fourth near-infrared spectra of 20 sets of reforming feedstock fractions to obtain .

[0174] Step S612: Construct a reforming feedstock property detection model using partial least squares method based on the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information.

[0175] Specifically, use the preprocessed , correlate the property set , and establish a reforming feedstock property detection model using partial least squares (PLS) .

[0176] Step S613: Collect the first near-infrared spectrum of the crude oil.

[0177] Step S614: Preprocess the target wavenumber range of the first near-infrared spectrum, and the preprocessing includes at least one of the following: wavelet transform processing, standard normal variate transform processing.

[0178] Step S615: Input the first near-infrared spectrum into the pre-constructed spectral conversion model, and output the second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil.

[0179] Step S616: Input the second near-infrared spectrum into the pre-constructed reforming feedstock property detection model, and output the property information of the reforming feedstock fraction.

[0180] Specifically, after the spectral conversion model and the reforming feedstock property detection model are constructed, the first near-infrared spectrum of the crude oil for predicting the properties of the reforming feedstock fraction can be collected, and the first near-infrared spectrum is input into the spectral conversion model to output the second near-infrared spectrum corresponding to the reforming feedstock fraction. Then, the second near-infrared spectrum is input into the reforming feedstock property detection model to obtain the properties of the reforming feedstock fraction.

[0181] In summary, this embodiment can be divided into a model construction stage (corresponding to steps S601 to S612) and a model application stage (corresponding to steps S613 to S616); by establishing a spectral conversion model between the crude oil and its true boiling point cut fixed distillation section (15°C to 140°C), determining the target wave number section through the relevant weight coefficients at each wave number point, and using wavelet transform and standard normal transform to preprocess the spectrum to eliminate spectral noise and spectral variation in the spectral signal, establishing a mixed feedstock property detection model for the true boiling point cut reforming feedstock fraction and the straight-run reforming feedstock fraction product of the distillation unit, and conducting key property prediction analysis on the spectrum of the reformed feedstock fraction after the crude oil is converted; this embodiment solves the problem of the long time-consuming true boiling point distillation of crude oil in traditional crude oil evaluation, can quickly and accurately obtain the key properties of the reforming feedstock fraction obtained by true boiling point cutting of crude oil, helps enterprises quickly obtain the physical property data of the reforming feedstock fraction of crude oil processing products, and timely guides the formulation of enterprise processing plans and the operation of distillation units.

[0182] Optionally, after the model is constructed, the model can also be evaluated, and the evaluation process is as follows:

[0183] Select 4 groups of detailed evaluation crude oils as test samples, collect the near-infrared spectra of the test samples, and after spectral preprocessing, convert the crude oil spectra to reforming feedstock fraction spectra through the constructed spectral conversion model, and then use the reforming feedstock property detection model to detect the converted reforming feedstock fraction spectra to obtain the properties of the reforming feedstock fraction cut from the crude oil, as shown in Table 5:

[0184] Table 5

[0185] 。

[0186] By comparing the predicted values and actual values of the key properties in Table 5, it can be seen that the error of the predicted values is relatively small. In other words, in this embodiment, only one spectral collection of the crude oil sample is required to accurately predict the properties of the reforming feedstock fraction cut from the crude oil. Through the spectral conversion model, the crude oil spectrum is converted into the spectrum of the reforming feedstock fraction, so as to obtain the component properties of the light fractions in the crude oil, and this analysis accuracy is sufficient to meet the requirements of production and processing.

[0187] Figure 11The structural schematic diagram of an apparatus for directly predicting the properties of reforming feedstock from crude oil provided by an embodiment of the present invention is as follows. Figure 11 As shown, the apparatus includes:

[0188] A collection module 1101, configured to collect a first near-infrared spectrum corresponding to crude oil; a conversion module 1102, configured to input the first near-infrared spectrum into a pre-constructed spectrum conversion model, and output a second near-infrared spectrum corresponding to the reforming feedstock fraction cut from the crude oil; a prediction module 1103, configured to input the second near-infrared spectrum into a pre-constructed reforming feedstock property detection model, and output property information of the reforming feedstock fraction.

[0189] In some embodiments, the spectrum conversion model includes sub-spectrum conversion models corresponding to at least one target wavenumber point in a target wavenumber segment. The conversion module 1102 is specifically configured to: extract first spectral data related to the target wavenumber point from the first near-infrared spectrum; input the first spectral data into the sub-spectrum conversion model corresponding to the target wavenumber point, and output second spectral data corresponding to the reforming feedstock fraction at the target wavenumber point. The second spectral data corresponding to the reforming feedstock fraction at at least one target wavenumber point constitutes the second near-infrared spectrum.

[0190] In some embodiments, the sub-spectrum conversion model is converted by a moving window partial least squares method based on a preset number of windows. The conversion module 1102 is specifically configured to: determine adjacent wavenumber points centered on the target wavenumber point, and the number of wavenumber points of the target wavenumber point and the adjacent wavenumber points is the same as the preset number of windows; extract first spectral data corresponding to the target wavenumber point and the adjacent wavenumber points from the first near-infrared spectrum.

[0191] In some embodiments, the apparatus further includes a preprocessing module 1104, and the preprocessing module 1104 is configured to preprocess the target wavenumber segment of the first near-infrared spectrum. The preprocessing includes at least one of the following: wavelet transform processing, standard normal variate transformation processing.

[0192] In some embodiments, the apparatus further includes a construction module 1105, and the construction module 1105 is configured to: obtain at least one set of crude oil samples, and collect a third near-infrared spectrum corresponding to each set of crude oil samples; perform true boiling point distillation on each set of crude oil samples to cut out reforming feedstock fraction samples corresponding to a preset temperature range; collect a fourth near-infrared spectrum corresponding to each set of reforming feedstock fraction samples; construct a spectrum conversion model according to the third near-infrared spectrum and the fourth near-infrared spectrum.

[0193] In some embodiments, the construction module 1105 is specifically configured to: determine a target wavenumber range, the target wavenumber range including at least one target wavenumber point; for each target wavenumber point in the target wavenumber range, extract third spectral data related to the target wavenumber point from the third near-infrared spectrum, and extract fourth spectral data corresponding to the target wavenumber point from the fourth near-infrared spectrum; construct a sub-spectral conversion model corresponding to the target wavenumber point according to the third spectral data and the fourth spectral data, and the sub-spectral conversion models corresponding to at least one target wavenumber point constitute a spectral conversion model.

[0194] In some embodiments, the construction module 1105 is specifically configured to: determine adjacent wavenumber points centered on the target wavenumber point, the number of wavenumber points of the target wavenumber point and the adjacent wavenumber points being consistent with the number of preset windows; extract third spectral data corresponding to the target wavenumber point and the adjacent wavenumber points from the third near-infrared spectrum; construct a sub-spectral conversion model according to the third spectral data and the fourth spectral data by using a moving window least squares conversion method with the number of preset windows.

[0195] In some embodiments, the construction module 1105 is specifically configured to: obtain first property information corresponding to each group of reformate fraction samples; construct relevant weight coefficients for each wavenumber point corresponding to the reformate fraction samples according to the fourth near-infrared spectrum and the first property information of each group of reformate fraction samples; determine the wavenumber range with relevant weight coefficients greater than a preset value as the target wavenumber range.

[0196] In some embodiments, the construction module 1105 is further configured to: perform preprocessing on the third near-infrared spectrum and the fourth near-infrared spectrum, the preprocessing including at least one of the following: wavelet transform processing, standard normal transform processing.

[0197] In some embodiments, the construction module 1105 is further configured to: obtain first property information corresponding to each group of reformate fraction samples; collect fifth near-infrared spectra corresponding to at least one group of actual reformate fraction samples, and obtain second property information corresponding to each group of actual reformate fraction samples; construct a reformate property detection model according to the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information.

[0198] In some embodiments, the construction module 1105 is specifically configured to: construct a reformate property detection model by using partial least squares method according to the fourth near-infrared spectrum, the fifth near-infrared spectrum, and the corresponding first property information and second property information.

[0199] In some embodiments, the construction module 1105 is further configured to: perform preprocessing on the target wavenumber range of the fourth near-infrared spectrum and the fifth near-infrared spectrum, the preprocessing including at least one of the following: wavelet transform processing, standard normal transform processing.

[0200] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working process and corresponding beneficial effects of the device for directly predicting the properties of reforming feedstock from crude oil described above can refer to the corresponding process in the foregoing method examples, and will not be elaborated here.

[0201] As Figure 12 shown, an embodiment of the present invention provides an electronic device, including a processor 1201, a communication interface 1202, a memory 1203, and a communication bus 1204. Among them, the processor 1201, the communication interface 1202, and the memory 1203 complete mutual communication through the communication bus 1204.

[0202] The memory 1203 is used to store a computer program.

[0203] In an embodiment of the present invention, when the processor 1201 is used to execute the program stored on the memory 1203, it implements the steps of the method for directly predicting the properties of reforming feedstock from crude oil provided by any one of the foregoing method embodiments.

[0204] For the electronic device provided by the embodiment of the present invention, its implementation principle and technical effects are similar to those of the above embodiment, and will not be elaborated here.

[0205] The above-mentioned memory 1203 can be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk, or a ROM. The memory 1203 has a storage space for program codes for executing any method steps in the above methods. For example, the storage space for program codes can include respective program codes for implementing each step in the above methods. These program codes can be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, optical discs (CDs), memory cards, or floppy disks. Such computer program products are usually portable or fixed storage units. The storage unit can have a storage segment or storage space arranged similarly to the memory 1203 in the above electronic device. The program codes can be compressed in an appropriate form. Usually, the storage unit includes a program for executing the method steps according to the embodiments of the present invention, that is, codes that can be read by a processor such as the processor 1201. When these codes are run by the electronic device, they cause the electronic device to execute each step in the method described above.

[0206] An embodiment of the present invention also provides a storage medium. A computer program is stored on the above-mentioned computer-readable storage medium, and when the computer program is executed by a processor, it implements the steps of the method for directly predicting the properties of reforming feedstock from crude oil as described above.

[0207] The computer-readable storage medium may be included in the device / apparatus described in the above embodiments; or it may exist separately without being assembled into the device / apparatus. The above computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of the present invention is implemented.

[0208] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, and may include, for example, but not limited to: portable computer disks, hard disks, random access memories (RAMs), read-only memories (ROMs), erasable programmable read-only memories (EPROMs or flash memories), portable compact disk read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in combination with an instruction execution system, apparatus, or device.

[0209] It should be noted that in this document, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0210] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for directly predicting the properties of reforming feedstock from crude oil, characterized in that: include: Collecting the first near-infrared spectrum corresponding to the crude oil; Inputting the first near-infrared spectrum into a pre-constructed spectral conversion model, and outputting a second near-infrared spectrum corresponding to the reforming raw material fraction cut from the crude oil, wherein the spectral conversion model is constructed by using the near-infrared spectra of the crude oil sample and the corresponding reforming raw material fraction sample using a partial least squares method; inputting the second near infrared spectrum into a pre-constructed reforming feedstock property detection model, and outputting property information of the reforming feedstock fraction, wherein the reforming feedstock property detection model is constructed by using the near infrared spectra and property information of the reforming feedstock fraction sample and the actual reforming feedstock fraction sample, and adopting the partial least square method; The spectrum conversion model includes a sub-spectrum conversion model corresponding to at least one target wavenumber point in the target wavenumber band, the target wavenumber band is a wavenumber band whose weight coefficients related to each wavenumber point are greater than a preset value, the weight coefficients related to each wavenumber point are constructed according to the near-infrared spectrum and property information of each group of reforming raw material fraction samples, and the sub-spectrum conversion model adopts a moving window partial least squares method based on a preset number of windows for conversion; The first near infrared spectrum is input into the pre-built spectrum conversion model, and the second near infrared spectrum corresponding to the reforming raw material fraction cut from the crude oil is output, including: extracting first spectrum data related to a target wave number point from the first near infrared spectrum; The first spectral data is input into the sub-spectral conversion model corresponding to the target wavenumber point, and the second spectral data corresponding to the reforming raw material fraction at the target wavenumber point is output. The second spectral data corresponding to the reforming raw material fraction at at least one target wavenumber point constitutes a second near-infrared spectrum.

2. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 1, characterized in that: Extracting first spectrum data related to a target wave number point from the first near infrared spectrum includes: Determine adjacent wave number points centered on the target wave number point, and the number of wave number points of the target wave number point and the adjacent wave number point is consistent with the number of preset windows; First spectrum data corresponding to the target wave number point and the adjacent wave number points are extracted from the first near infrared spectrum.

3. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 1 or 2, characterized in that: After collecting the first near infrared spectrum corresponding to the crude oil, the following steps are also included: The target wave number band of the first near infrared spectrum is preprocessed, wherein the preprocessing includes at least one of the following: wavelet transformation processing and standard normal transformation processing.

4. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 1 or 2, characterized in that: Before inputting the first NIR spectrum into the pre-built spectral conversion model, it also includes: Obtaining at least one group of crude oil samples, and collecting a third near infrared spectrum corresponding to each group of crude oil samples; Perform actual boiling point distillation on each group of crude oil samples to cut out corresponding reforming raw material fraction samples within a preset temperature range; collecting a fourth near infrared spectrum corresponding to each group of reforming raw material fraction samples; A spectral conversion model is constructed according to the third near-infrared spectrum and the fourth near-infrared spectrum.

5. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 4, characterized in that: A spectrum conversion model is constructed according to the third near infrared spectrum and the fourth near infrared spectrum, including: determining a target wavenumber band, wherein the target wavenumber band includes at least one target wavenumber point; For each target wavenumber point in the target wavenumber band, extract third spectrum data related to the target wavenumber point from the third near-infrared spectrum, and extract fourth spectrum data corresponding to the target wavenumber point from the fourth near-infrared spectrum; A sub-spectrum conversion model corresponding to the target wavenumber point is constructed according to the third spectrum data and the fourth spectrum data, and at least one sub-spectrum conversion model corresponding to the target wavenumber point constitutes a spectrum conversion model.

6. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 5, characterized in that: Extracting third spectrum data related to the target wave number point from the third near infrared spectrum includes: Determine the adjacent wave number points centered on the target wave number point. The number of wave number points of the target wave number point and the adjacent wave number point is consistent with the number of preset windows. extracting third spectrum data corresponding to the target wave number point and the adjacent wave number point from the third near infrared spectrum; Extracting fourth spectrum data related to the target wave number point from the fourth near infrared spectrum includes: Determine the adjacent wave number points centered on the target wave number point. The number of wave number points of the target wave number point and the adjacent wave number point is consistent with the number of preset windows. Extracting fourth spectrum data corresponding to the target wave number point and the adjacent wave number point from the fourth near infrared spectrum; Constructing a sub-spectrum conversion model corresponding to the target wave number point according to the third spectrum data and the fourth spectrum data, including: According to the third spectrum data and the fourth spectrum data, a sub-spectrum conversion model is constructed by adopting a moving window least squares conversion method with a preset number of windows.

7. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 5 or 6, characterized in that: Determine the target wave number band, including: Obtaining first property information corresponding to each group of reforming raw material fraction samples; Constructing the weight coefficients of the wave number points corresponding to the reforming raw material fraction samples according to the fourth near infrared spectrum and the first property information of each group of reforming raw material fraction samples; The wavenumber band whose relevant weight coefficient is greater than the preset value is determined as the target wavenumber band.

8. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 4, characterized in that: Before constructing the spectrum conversion model according to the third near-infrared spectrum and the fourth near-infrared spectrum, it also includes: The target wave number bands of the third near infrared spectrum and the fourth near infrared spectrum are preprocessed, and the preprocessing includes at least one of the following: wavelet transformation processing and standard normal transformation processing.

9. The method of directly predicting the properties of reforming feedstock from crude oil according to claim 4, characterized in that: Prior to inputting the second NIR spectrum into the pre-built reformate feedstock property detection model, it also includes: Obtaining first property information corresponding to each group of reforming raw material fraction samples; Collecting a fifth near infrared spectrum corresponding to at least one group of actual reforming raw material fraction samples, and obtaining second property information corresponding to each group of actual reforming raw material fraction samples; A reforming raw material property detection model is constructed according to the fourth near infrared spectrum, the fifth near infrared spectrum and the corresponding first property information and second property information.

10. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 9, characterized in that: A reforming raw material property detection model is constructed according to the fourth near infrared spectrum, the fifth near infrared spectrum and the corresponding first property information and second property information, including: According to the fourth near infrared spectrum, the fifth near infrared spectrum and the corresponding first property information and second property information, a reforming raw material property detection model is constructed using the partial least squares method.

11. The method for directly predicting the properties of reforming feedstock from crude oil according to claim 9 or 10, characterized in that: Before building the reforming raw material property detection model, it also includes: The target wave number bands of the fourth near infrared spectrum and the fifth near infrared spectrum are preprocessed, and the preprocessing includes at least one of the following: wavelet transformation processing and standard normal transformation processing.

12. A device for applying the method for directly predicting the properties of reforming feedstock from crude oil as claimed in any one of claims 1 to 11, characterized in that: include: A collection module, used for collecting a first near-infrared spectrum corresponding to crude oil; A conversion module, used for inputting the first near infrared spectrum into a pre-built spectrum conversion model, and outputting a second near infrared spectrum corresponding to the reforming raw material fraction cut from the crude oil; The prediction module is used to input the second near infrared spectrum into a pre-built reforming raw material property detection model and output the property information of the reforming raw material fraction.

13. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor is used to implement the steps of the method for directly predicting the properties of reforming feedstock from crude oil as described in any one of claims 1 to 11 when executing the program stored in the memory.

14. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for directly predicting the properties of reforming feedstock from crude oil are implemented as described in any one of claims 1 to 11.

Images