Method and system for predicting attributes of straight-run naphtha

Abstract

The embodiment of the invention provides a straight-run naphtha attribute prediction method and system, and belongs to the technical field of petrochemical engineering. The prediction method comprises the following steps: acquiring the proportion of each type of crude oil added for synthesizing straight-run naphtha to be predicted, the yield of straight-run naphtha corresponding to each type of crude oil and distillation range data; and inputting the proportion of each crude oil, the yield of the straight-run naphtha corresponding to each crude oil and the distillation range data into a pre-trained prediction model, and obtaining a first attribute of the straight-run naphtha to be predicted based on a distillation range cutting model in the prediction model. According to the technical scheme, the obtained proportion of each type of crude oil added for synthesizing the straight-run naphtha to be predicted, the yield of the straight-run naphtha corresponding to each type of crude oil and the distillation range data are input into the flow cutting model in the pre-constructed prediction model, the first attribute of the straight-run naphtha to be predicted is determined, and the straight-run naphtha to be predicted is predicted according to the first attribute of the straight-run naphtha to be predicted. The naphtha attribute analysis period is shortened, and the accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of petrochemical engineering, and in particular to a method and system for predicting properties of straight-run naphtha. Background Art

[0002] Straight-run naphtha is an important oil resource produced by crude oil regular decompression units. It is an important raw material for producing ethylene and propylene by cracking in tubular furnaces and for producing benzene, toluene and xylene by catalytic reforming. In actual production processes, near-infrared spectroscopy is usually used to analyze the properties of straight-run naphtha, which has a long cycle and low accuracy for analyzing naphtha properties. Summary of the invention

[0003] The purpose of the embodiments of the present invention is to provide a method and system for predicting properties of straight-run naphtha, so as to solve the technical problems existing in the above-mentioned prior art.

[0004] In order to achieve the above-mentioned object, an embodiment of the present invention provides a method for predicting the properties of straight-run naphtha, comprising: obtaining the proportion of each crude oil added to synthesize the straight-run naphtha to be predicted, the yield of the straight-run naphtha corresponding to each crude oil, and distillation process data;

[0005] The proportion of each crude oil, the yield of straight-run naphtha corresponding to each crude oil, and the distillation range data are input into a pre-trained prediction model, and the first attribute of the straight-run naphtha to be predicted is obtained based on the distillation range cutting model in the prediction model.

[0006] Optionally, the distillation process data includes an initial distillation point, a final distillation point, and an increased distillation ratio of each crude oil corresponding to the distillation temperature.

[0007] Optionally, the distillation range cutting model is constructed according to the following formula:

[0008]

[0009] In the formula, M represents the number of crude oil types, P i Indicates the increase in distillation ratio corresponding to the distillation temperature, Yield k Indicates the yield of straight-run naphtha corresponding to each crude oil, Ratio k Indicates the proportion of each crude oil.

[0010] Optionally, the distillation ratio of straight-run naphtha from each crude oil at a given distillation temperature is determined according to the following formula:

[0011]

[0012] Wherein, x = [x1, x2, ..., x7] represents the distillation temperature, w0, w1, ..., wn is the coefficient and n is the order.

[0013] Optionally, the order is determined by the minimum of the following formula:

[0014]

[0015] In the formula, SSE represents the error sum of squares, m represents the number of samples, y(x i , W) represents the true value of the distillation ratio, t i represents the fitted value.

[0016] Optionally, the distillation ratio P of the straight-run naphtha corresponding to the distillation temperature of each crude oil distillation is determined according to the following formula: i :

[0017] P i =y(x i+1 , W)-y(x i , W)i=1,2,...,N

[0018] Where P i It indicates the distillation ratio corresponding to the increase in distillation temperature, and i is the total increase from the initial distillation point to the final distillation point.

[0019] Optionally, the prediction method further includes:

[0020] The distillation range cutting model is corrected by a rolling correction method to obtain a corrected distillation range cutting model.

[0021] Optionally, the prediction model further includes a pre-built linear prediction sub-model, and the prediction method further includes:

[0022] Acquire the type of each crude oil added to synthesize the straight-run naphtha to be predicted and the second attribute of the straight-run naphtha corresponding to each crude oil;

[0023] The type of each crude oil, the proportion of each crude oil, the second attribute and yield of the straight-run naphtha corresponding to each crude oil are input into the pre-built linear prediction sub-model to obtain the second attribute of the straight-run naphtha to be predicted.

[0024] Optionally, the first attribute includes distillation process information of the straight-run naphtha to be predicted, and the second attribute includes at least one of a PIONA value, a yield and a density of the straight-run naphtha to be predicted.

[0025] Optionally, the construction of the linear prediction sub-model includes:

[0026] Obtaining historical data of each crude oil in the mixed crude oil, wherein the historical data includes the type of each crude oil, the proportion of each crude oil, and the properties and yield information of the straight-run naphtha obtained by distilling each crude oil;

[0027] Inputting the historical detailed evaluation data of each crude oil into the initial linear prediction sub-model to obtain the prediction attribute information corresponding to the straight-run naphtha distilled from the mixed crude oil;

[0028] The real attribute information corresponding to the straight-run naphtha distilled from the mixed crude oil is obtained, and combined with the predicted attribute information, the initial linear prediction sub-model is corrected by using a rolling correction method to obtain the linear prediction sub-model.

[0029] Optionally, the linear prediction sub-model is constructed according to the following formula:

[0030]

[0031] In the formula, n represents the number of crude oil types, Ratio i Indicates the proportion of each crude oil, X i and Yield i They represent certain properties and yields of straight-run naphtha obtained by distillation of each crude oil, and Y represents the prediction result corresponding to the property.

[0032] Optionally, the rolling correction method includes correction according to the following formula:

[0033] P out =P liner +k·(P LIMS -P liner )

[0034] Where P linear Represents an attribute data output by the linear prediction sub-model or distillation range cutting model, P LIMS Indicates the real data corresponding to the attribute, P out is the output result after correction, and k is the correction coefficient.

[0035] On the other hand, the present invention also provides a prediction system for straight-run naphtha properties, comprising:

[0036] Memory; and

[0037] The processor is configured to execute the prediction method described above.

[0038] Through the above technical scheme, the proportion of each type of crude oil added to the synthetic straight-run naphtha to be predicted, the yield of straight-run naphtha corresponding to each crude oil, and the distillation process data are input into the process cutting model in the pre-constructed prediction model to determine the first attribute of the straight-run naphtha to be predicted. By using the pre-trained and constructed prediction model, the naphtha attribute analysis cycle is shortened and the accuracy is improved.

[0039] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0041] Figure 1 It is a flow chart of an implementation of a method for predicting properties of straight-run naphtha provided by an embodiment of the present invention;

[0042] Figure 2 It is a cutting model flow chart provided by an embodiment of the present invention;

[0043] Figure 3 It is a flowchart of another method for predicting properties of straight-run naphtha provided by an embodiment of the present invention;

[0044] Figure 4 is a linear prediction sub-model flow chart provided by an embodiment of the present invention;

[0045] Figure 5 is a rolling correction flow chart provided by an embodiment of the present invention;

[0046] Figure 6 This is a prediction framework diagram of straight-run naphtha properties provided by an embodiment of the present invention;

[0047] Figure 7 (ad) is a fitting curve corresponding to when the order n is 3, 4, 5, and 6 provided in an embodiment of the present invention;

[0048] Figure 8 The distillation ratio of naphtha increased by one degree Celsius provided by an embodiment of the present invention;

[0049] Fig. 9 is a PIONA value result obtained according to a linear prediction sub-model provided by an embodiment of the present invention;

[0050] Fig.10(ad) is a comparison curve between the predicted value and the actual value of the PIONA value of a 1# atmospheric vacuum top first-class oil provided in an embodiment of the present invention;

[0051] Fig.11 (ae) is a schematic diagram of the comparison results of the actual values ​​and the predicted values ​​of the initial distillation point, 10% distillation temperature, 50% distillation temperature, 90% distillation temperature and final distillation point of a 1# atmospheric vacuum top first-grade oil provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] Ethylene is the main representative product of the petrochemical industry and occupies a dominant position among petrochemical products. Its production capacity has become an important indicator for measuring a country's chemical level internationally.

[0053] In the ethylene production process, ethylene raw materials account for more than 80% of the overall production cost. Different ethylene raw materials directly determine the level of ethylene production costs, and also affect the yield of ethylene and by-products. From the perspective of cracking raw materials used by domestic chemical companies to produce ethylene, most companies mainly use naphtha as raw material, and the proportion accounts for more than 50% of all raw materials. As the main raw material for ethylene cracking, naphtha is divided into light naphtha and heavy naphtha according to the distillation range. It is generally produced by the company itself and has relatively abundant resources. Therefore, it is very necessary for chemical companies to analyze the properties of straight-run naphtha to provide a basis for the molecular optimization of naphtha resources "aromatic if aromatic, olefin if olefin".

[0054] Straight-run naphtha is an important oil resource produced by the crude oil regular decompression unit. It is an important raw material for the production of ethylene and propylene by tubular furnace cracking and benzene, toluene and xylene by catalytic reforming. The properties of straight-run naphtha include density, distillation range, sulfur content, PIONA value, etc. Among them, PIONA value is an important basis for determining the flow direction of naphtha, and it is also an important indicator for evaluating the cracking performance of naphtha and determining the yield of high value-added products of ethylene cracking. In the actual production process, enterprises generally use liquid chromatography to analyze the PIONA value properties of naphtha offline. The offline analysis cycle is long and the timeliness is poor, resulting in a lag in the acquisition of analysis data, which affects the quality and harmony of naphtha. Near-infrared analysis can realize real-time analysis of naphtha properties, but there are problems such as high equipment investment and untimely model maintenance, which will lead to large errors in the analysis results. Therefore, naphtha resource optimization urgently needs low-cost and high-precision analysis or prediction methods.

[0055] In the field of naphtha property prediction technology at home and abroad, Reboucas et al. used near-infrared spectroscopy to analyze the properties of naphtha produced by Braskem, a South American petrochemical company, and used the PLS calibration model to predict the PINA value, density and distillation range. The results were close to the real data. The near-infrared spectrum acquisition was used to replace chromatography, distillation and density methods, and the measurement time was shorter. This method has been used in the Braskem laboratory. Ding Yan et al. proposed a method for predicting naphtha properties based on online Raman spectroscopy combined with principal component particle swarm algorithm with the help of intelligent optimization algorithm. Zhou Yongjian et al. analyzed naphtha properties through online near-infrared spectroscopy, established near-infrared PCR and PLS prediction models constructed by different pretreatment methods, compared the average relative error, and analyzed that the smoothing and differential smoothing pretreatment models had better prediction effects. However, the applicant found that although the above scheme solved the problems of relatively long naphtha property analysis time, low accuracy or inability to extrapolate in petrochemical enterprises, there were high equipment investment costs and the model needed to be continuously maintained, otherwise the prediction would be inaccurate.

[0056] With the continuous development of machine learning technology, many scholars have tried to apply machine learning methods to the chemical industry, modeling based on historical data to predict naphtha properties. Cheng Ming et al. established an aromatic potential soft measurement model using RBF neural network technology, using the density and distillation range data of heavy naphtha as network input to predict the aromatic potential content of heavy naphtha, solving the problem of difficult measurement of aromatic potential parameters. The above method can quickly and easily obtain prediction results, solving the problem of high cost of near-infrared method, but it is slightly insufficient in accuracy. The existing technology lacks a soft measurement model for the evaluation of straight-run naphtha properties.

[0057] The applicant found that in the actual production process, the PIONA value analysis cycle of naphtha is relatively long, which is not enough to meet the timeliness of naphtha resource optimization; the use of near-infrared analysis method has problems such as high cost and regular maintenance and update of the model. In view of the shortcomings of existing analysis technology, the present invention proposes a method for predicting the properties of straight-run naphtha, which can quickly and efficiently predict the properties of straight-run naphtha through the types and proportions of crude oil processed at atmospheric and vacuum pressure, and provide auxiliary decision-making for the optimal allocation of naphtha resources.

[0058] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0059] See also Figure 1 As shown, it is a flow chart of an implementation method of a straight-run naphtha property prediction method provided by an embodiment of the present invention, comprising the following execution steps:

[0060] Step 100: Obtain the proportion of each crude oil added to synthesize the straight-run naphtha to be predicted, the yield of the straight-run naphtha corresponding to each crude oil, and distillation process data.

[0061] In some embodiments, the distillation process data includes the initial distillation point, the final distillation point and the distillation ratio of each crude oil corresponding to the distillation temperature.

[0062] Step 101: Input the proportion of each crude oil, the yield of straight-run naphtha corresponding to each crude oil, and distillation range data into a pre-trained prediction model, and obtain the first attribute of the straight-run naphtha to be predicted based on the distillation range cutting model in the prediction model.

[0063] In some embodiments, the first attribute includes distillation process information of the straight-run naphtha to be predicted.

[0064] In some embodiments, the lamination cutting model is constructed according to the following formula:

[0065]

[0066] In the formula, M represents the number of crude oil types, P i Indicates the increase in distillation ratio corresponding to the distillation temperature, Yield k Indicates the yield of straight-run naphtha corresponding to each crude oil, Ratio k Indicates the proportion of each crude oil.

[0067] In some embodiments, the distillation ratio of straight-run naphtha from each crude oil at a given distillation temperature is determined according to the following formula:

[0068]

[0069] Wherein, x = [x1, x2, ..., x7] represents the distillation temperature, w0, w1, ..., w n is the coefficient and n is the order.

[0070] In some implementations, the order is determined according to the minimum of the following formula:

[0071]

[0072] In the formula, SSE represents the error sum of squares, m represents the number of samples, y(x i , W) represents the true value of the distillation ratio, t i represents the fitted value.

[0073] In some embodiments, the distillation ratio P of the straight-run naphtha corresponding to the distillation temperature of each crude oil distillation is determined according to the following formula:i :

[0074] P i =y(x i+1 , W)-y(x i , W)i=1,2,...,N

[0075] Where P i It indicates the distillation ratio corresponding to the increase in distillation temperature, and i is the total increase from the initial distillation point to the final distillation point.

[0076] In some embodiments, the prediction method further comprises: after the distillation range cutting model is constructed, the distillation range cutting model is corrected by a rolling correction method to obtain a corrected distillation range cutting model.

[0077] Correction using the rolling correction method includes correction according to the following formula:

[0078] P out =P liner +k·(P LIMS -P liner )

[0079] Where P linear Represents an attribute data output by the linear prediction sub-model or distillation range cutting model, P LIMS Indicates the real data corresponding to the attribute, P out is the output result after correction, and k is the correction coefficient.

[0080] In some embodiments, see Figure 2 As shown, it is a flow chart of a cutting model provided by an embodiment of the present invention. The linear prediction sub-model has defects in the prediction of distillation range attributes, which are mainly manifested in the initial distillation point and the final distillation point. Because the initial distillation point and the final distillation point of the mixed oil product are cut according to the initial distillation point and the final distillation point of the naphtha distilled from the various crude oils before mixing, the cutting starts when the minimum initial distillation point temperature of the naphtha distilled from the various crude oils before mixing is reached; the cutting stops when the maximum final distillation point temperature of the naphtha is reached. In other words, the initial distillation point of the naphtha produced by the atmospheric and vacuum distillation unit is the minimum initial distillation point of the naphtha distilled from all the feed crude oils of the unit on that day, and the final distillation point is the maximum final distillation point of the naphtha distilled from all the feed crude oils of the unit on that day. In view of the problems existing in the distillation range prediction, a naphtha distillation range cutting model for each crude oil is built based on the naphtha distillation range data. The distillation range of naphtha distilled from a certain crude oil, including: initial distillation point, 10% distillation temperature, 30% distillation temperature, 50% distillation temperature... and the final distillation point, is calculated by using a polynomial curve fitting model. According to the appropriate Select the order (i.e. n) in the curve fitting model, according to the formula P i =y(x i+1 , W)-y(xi , W)i=1,2,...,N determine the distillation ratio corresponding to the distillation point, and combine the crude oil type M, naphtha yield Yield k And the oil processing ratio Ratio k According to the following formula, the increase in the distillation ratio BP of the atmospheric and vacuum unit naphtha at each distillation point is determined: i :

[0081]

[0082] In some embodiments, it is assumed that the naphtha distillation range of a certain crude oil is as shown in Table 1:

[0083] Table 1 Naphtha distillation range of a certain crude oil

[0084]

[0085] The distillation range data for each straight-run naphtha analysis is from the initial distillation point to the final distillation point, with a total of 7 data. The distillation ratio of the initial distillation point is 0, and the distillation ratio of the final distillation point is 100%. In order to improve the accuracy of the distillation range cutting model, it is necessary to accumulate the increase in the distillation ratio of each naphtha for each degree Celsius increase in temperature to obtain the temperature at which the mixed oil meets a certain distillation ratio. In order to calculate the increase in the distillation ratio for a one-degree Celsius increase, it is necessary to fit the known distillation range data of the naphtha. Common fitting methods include linear regression and polynomial regression. Considering that the distillation range attributes are nonlinear and limited in number, the present invention uses a polynomial curve fitting method. In principle, this method determines each expansion coefficient through least squares, and can simulate nonlinear separable data, which is more flexible than the linear regression method. The polynomial is defined as Where x = [x1, x2, ..., x7], w0, w1, ..., w n is the coefficient of the polynomial, denoted as W, n is the order of the polynomial, y(x, n) is the nonlinear function of x, and also represents the distillation ratio of naphtha distilled from this crude oil at a given distillation temperature.

[0086] When the objective function of the fitting is determined, in order to determine the polynomial order, the fitting effect is evaluated by the sum of squared errors (SSE). The SSE calculation method is shown in the following formula. It is expected that the smaller the SSE value, the better.

[0087]

[0088] Among them, m represents the number of samples, y(x i , W) represents the true value, t i Represents the fitting value. Calculate the SSE after fitting polynomials of different orders, consider the overfitting and underfitting problems, and select the model order corresponding to the appropriate SSE.

[0089] The calculation method for the increase in distillation ratio of naphtha for every 1℃ increase is as follows: i =y(x i+1 , W)-y(x i , W)i=1, 2, ..., N, where N is the total number of increases from the initial distillation point to the final distillation point, P i Indicates the distillation ratio that increases with the distillation temperature.

[0090] For the naphtha distillation range data of each crude oil, the distillation range cutting model is established according to the above method. When predicting the distillation range data of a certain day, the calculation method is shown in the following formula. Assuming that there are M kinds of crude oil, according to the naphtha yield Yield of each crude oil k And the distillation ratio P corresponding to the distillation temperature increase i , and the processing ratio of each crude oil k , calculate the increase in distillation ratio BP of naphtha at each distillation temperature of the atmospheric and vacuum unit i , and then the increased distillation ratio is accumulated to obtain the corresponding relationship between the distillation temperature and the distillation ratio. The predicted distillation point can be obtained according to the corresponding distillation ratio.

[0091]

[0092] At this point, the construction of the distillation range cutting model has been completed.

[0093] In some embodiments, see Figure 3 As shown, it is a flowchart of another method for predicting the properties of straight-run naphtha provided by an embodiment of the present invention. The prediction model also includes a pre-built linear prediction sub-model. The prediction method specifically includes the following execution steps:

[0094] S300: Obtain the type of each crude oil added to synthesize the straight-run naphtha to be predicted and the second attribute of the straight-run naphtha corresponding to each crude oil.

[0095] S301: Input the type of each crude oil, the proportion of each crude oil, and the second attribute of the straight-run naphtha corresponding to each crude oil into the pre-built prediction sub-model to obtain the second attribute of the straight-run naphtha to be predicted.

[0096] It should be noted that the second attribute of the straight-run naphtha corresponding to each crude oil is the same as the second attribute of the straight-run naphtha to be predicted, which is at least one of the PIONA value, yield and density attributes.

[0097] In some implementations, before executing step S301, the linear prediction sub-model is constructed, which can be specifically constructed by the following steps:

[0098] S1: Obtaining historical data of each crude oil in the mixed crude oil, wherein the historical data includes the type of each crude oil, the proportion of each crude oil, and the properties and yield information of the straight-run naphtha obtained by distilling each crude oil;

[0099] S2: Inputting the historical detailed evaluation data of each crude oil into the initial linear prediction sub-model to obtain the prediction attribute information corresponding to the straight-run naphtha distilled from the mixed crude oil;

[0100] S3: Acquire the real attribute information corresponding to the straight-run naphtha distilled from the mixed crude oil, and combine it with the predicted attribute information to calibrate the initial linear prediction sub-model using a rolling correction method to obtain the linear prediction sub-model.

[0101] In some embodiments, correction using the rolling correction method includes correction according to the following formula:

[0102] P out =P liner +k·(P LIMS -P liner )

[0103] Where P linear Represents an attribute data output by the linear prediction sub-model or distillation range cutting model, P LIMS Indicates the real data corresponding to the attribute, P out is the output result after correction, and k is the correction coefficient.

[0104] In some implementations, the linear prediction sub-model in step S201 may be constructed according to the following formula:

[0105]

[0106] In the formula, n represents the number of crude oil types, Ratio i Indicates the proportion of each crude oil, X i and Yield i They represent certain properties and yields of straight-run naphtha obtained by distillation of each crude oil, and Y represents the prediction result corresponding to the property.

[0107] In some embodiments, see Figure 4As shown, it is a flow chart of a linear prediction sub-model provided by an embodiment of the present invention, in which the proportion of each crude oil in the mixed crude oil, the number of crude oil types and the raw material properties (normal alkanes, isoalkanes, cycloalkanes, aromatic hydrocarbons, distillation and density) are input into the linear prediction sub-model. The linear prediction sub-model calculates the property value of the naphtha produced by the device in a linear superposition manner based on the mixing proportion of each crude oil and the naphtha properties and yield in the crude oil detailed evaluation database, that is, the linear prediction sub-model outputs the property prediction results corresponding to the naphtha distilled from the mixed crude oil (including normal alkanes, isoalkanes, cycloalkanes, aromatic hydrocarbons, distillation and density).

[0108] In some embodiments, the construction of the distillation range cutting model includes the following execution steps:

[0109] S0: Obtain the distillation ratio of the straight-run naphtha corresponding to each crude oil distillation at a given distillation temperature, and the distillation ratio of the straight-run naphtha corresponding to the distillation temperature increase.

[0110] In some embodiments, the distillation ratio of straight-run naphtha obtained by distilling each crude oil at a given distillation temperature may be determined according to the following formula:

[0111]

[0112] Wherein, x = [x1, x2, ..., x7] represents the distillation temperature, w0, w1, ..., w n is the coefficient and n is the order.

[0113] In some implementations, the order may be determined according to the minimum value of the following formula:

[0114]

[0115] In the formula, SSE represents the error sum of squares, m represents the number of samples, y(x i , W) represents the true value of the distillation ratio, t i represents the fitted value.

[0116] In some embodiments, the distillation ratio P of the straight-run naphtha corresponding to the distillation temperature of each crude oil distillation can be determined according to the following formula: i :

[0117] P i =y(x i+1 , W)-y(x i , W)i=1,2,...,N

[0118] Where P iIt indicates the distillation ratio corresponding to the increase in distillation temperature, and i is the total increase from the initial distillation point to the final distillation point.

[0119] S1: Construct the distillation range cutting model based on the distillation ratio of the straight-run naphtha corresponding to each crude oil distilled at a given distillation temperature, the distillation ratio increased by the distillation temperature of the straight-run naphtha corresponding to each crude oil, the straight-run naphtha yield corresponding to each crude oil and the proportion of each crude oil.

[0120] In some embodiments, the distillation range cutting model can be constructed according to the following formula:

[0121]

[0122] In the formula, M represents the number of crude oil types, P i Indicates the increase in distillation ratio corresponding to the distillation temperature, Yield k Indicates the yield of straight-run naphtha corresponding to each crude oil, Ratio k Indicates the proportion of each crude oil.

[0123] In some embodiments, after the distillation range cutting model is constructed, the prediction method further comprises:

[0124] The distillation range cutting model is corrected by a rolling correction method to obtain a corrected distillation range cutting model.

[0125] In some embodiments, correction using the rolling correction method includes correction according to the following formula:

[0126] P out =P liner +k·(P LIMS -P liner )

[0127] Where P linear Represents an attribute data output by the linear prediction sub-model or distillation range cutting model, P LIMS Indicates the real data of the attribute, P out is the output result after correction, and k is the correction coefficient.

[0128] In some embodiments, the parameter Figure 5 As shown, a rolling correction flow chart provided by an embodiment of the present invention is provided. By comparing the PIONA value of naphtha analyzed offline with the predicted value, some properties have large deviations. In combination with the types and proportions of crude oil processed by the atmospheric and vacuum distillation unit, the predicted results are associated with the offline analysis data through rolling correction, the errors of the PIONA value and the distillation range prediction are reduced, and the "rolling" effect is achieved. In the established linear prediction model and distillation range cutting model of naphtha properties, the rolling correction method is used to improve the model prediction accuracy. The correction method is shown in the following formula:

[0129] P out =P liner +k·(P LIMS -P liner )

[0130] Among them, P linear Represents a property data output by the linear prediction model or distillation range cutting model, P LIMS Indicates the latest attribute data in the LIMS system, P out is the output result after correction, and k is the correction coefficient.

[0131] Compare the results predicted by the linear prediction model or the distillation range cutting model with the latest data in the LIMS system to calculate the deviation, multiply it by the correction coefficient, and add the results predicted by the linear prediction model or the cutting model to obtain the corrected output value. The correction coefficient ranges from 0 to 1. Combined with the changes in the types and proportions of crude oil actually put into production by the refining enterprise and the PIONA value and distillation range analyzed by the LIMS system, theoretically, if the changes in the types and proportions of crude oil are smaller when predicting naphtha properties, the correction coefficient is closer to 1; if the difference is large, set the correction coefficient to 0. If no correction is performed, the results obtained using the linear prediction model and the distillation range cutting model can also be manually set by the user according to the actual crude oil processing situation.

[0132] In some embodiments, the parameter Figure 6 As shown, a prediction architecture diagram of straight-run naphtha properties provided by an embodiment of the present invention is provided. The mixing ratio, naphtha properties and yield of each crude oil are input into a pre-constructed linear prediction sub-model, and the distillation range data of the straight-run naphtha obtained after distillation of each crude oil is input into a pre-constructed cutting model to obtain the property prediction results of the straight-run naphtha obtained after distillation of the mixed crude oil. On the established linear prediction sub-model and distillation range cutting model of naphtha properties, the linear prediction sub-model and the cutting model are corrected by using a rolling correction model and a rolling correction method in combination with the original data of the system (i.e., real data) to improve the model prediction accuracy and obtain more accurate naphtha property prediction results.

[0133] In some implementations, in order to further highlight the advantages of the present invention, some technical contents in the specific implementations of the present invention are illustrated below:

[0134] Firstly, the distillation range data of naphtha distilled from crude oil 1 are selected to build the distillation range cutting model of the naphtha. The distillation range data of naphtha distilled from crude oil 1 are shown in Table 2.

[0135] Table 2 Naphtha distillation range of crude oil 1

[0136]

[0137] Using the polynomial fitting method, by selecting the polynomial order, the distillation range and distillation ratio of naphtha obtained by distillation of crude oil 1 are fitted respectively, and fitting curves of different orders are obtained as follows: Figure 7 (ad) shown. Figure 6 (a), 6(b), 6(c), and 6(d) represent the fitting curves corresponding to the order n of 3, 4, 5, and 6, respectively. After the objective function of the fitting is determined, in order to determine the polynomial order, the square sum of error SSE is used to evaluate the fitting effect. The relationship between the obtained polynomial order and SSE is shown in Table 3.

[0138] Table 3 Relationship between order and SSE

[0139]

[0140] pass Figure 7 Comprehensively evaluated with Table 3, in order to prevent the curve from overfitting and underfitting problems, the polynomial order n is determined to be 5. The corresponding relationship between the distillation temperature and the distillation ratio after polynomial fitting is shown in the following formula, where f(x) represents the distillation ratio of naphtha distilled from crude oil 1 at a given distillation temperature x.

[0141] f(x)=8.541×10 -9 x 5 -5.245×10 -6 x 4 +0.001169x 3 -0.1137x 2 +5.278x-99.53

[0142] According to the distillation range fitting formula of naphtha distilled from crude oil 1, the distillation volume of the naphtha from the initial distillation point to the final distillation point is calculated, and then the increase in the distillation ratio of the naphtha for every increase of one degree Celsius is calculated. The result is as follows: Figure 8 As shown, it can be seen that for every one degree Celsius increase in the naphtha temperature, the increased distillation ratio shows a distribution trend of first increasing and then decreasing.

[0143] According to the types and proportions of crude oil processed by the 1# atmospheric and vacuum unit in the daily production report of the refinery during a certain period, the linear prediction sub-model is substituted and the PIONA value results are obtained as follows: Fig. 9 By comparing the actual data of PIONA in the LIMS system, the predicted content of normal alkanes is too high, while the content of cycloalkanes and aromatics is too low, and the difference with the actual value is large.

[0144] Compare the results predicted by the linear prediction model or the distillation range cutting model with the corresponding latest analysis data in the LIMS system, calculate the deviation, multiply it by the correction coefficient, and add the results predicted by the linear prediction model or the cutting model to obtain the corrected output value. Considering that the types and proportions of the actual feed crude oil are relatively complex and it is difficult to determine the relationship with the correction coefficient, the present invention simplifies the setting of the correction coefficient when verifying the model, and divides it into the following three cases: the type and proportion of crude oil have not changed or have changed slightly, and the correction coefficient is 1; the type and proportion of crude oil have changed significantly, and the correction coefficient is 0.8; the types of crude oil are different, and the correction coefficient is 0. By comparing the model prediction results with the actual data, it is found that the model meets the accuracy requirements, and the prediction results are analyzed below.

[0145] The present invention takes the 1# atmospheric and vacuum unit as an example, selects the PIONA and distillation range measured data of the 1# atmospheric and vacuum top first-class oil from July 1 to October 18 of a certain year in the LIMS system, and compares them with the output results of the prediction model after correction to verify the prediction effect of the model. The comparison curve of the predicted value and the actual value of the PIONA value of the 1# atmospheric and vacuum top first-class oil is shown in the figure. Fig.10 (ad) shown. Fig.10 (a), 10(b), 10(c), and 10(d) respectively show the comparison results of the actual values ​​and predicted values ​​of normal alkanes, isoalkanes, cycloalkanes, and aromatics of 1# atmospheric and vacuum top first-class oil. It can be seen from the figure that the trend of the results after correction is roughly consistent with the measured data, with only a few slightly larger differences. The comparison curve of the predicted value and the actual value of the distillation range of 1# atmospheric and vacuum top first-class oil is shown in Figure 1. Fig.11 (ae) shown.

[0146] Fig.11 (a), 11(b), 11(c), 11(d), 11(e) respectively represent the comparison results of the actual values ​​and predicted values ​​of the initial distillation point, 10% distillation temperature, 50% distillation temperature, 90% distillation temperature and final distillation point of 1# atmospheric vacuum top oil. It can be seen from the figure that the change trend of the predicted results of the distillation range is basically consistent with the real data. In order to verify the accuracy of the model, the model is evaluated by three evaluation indicators: mean relative error (MRE), mean absolute error (MAE) and mean square error (MSE), and the calculation methods are shown in the following formulas.

[0147]

[0148]

[0149]

[0150] Where N is the number of samples evaluated, and Yf(x) represents the difference between the true value and the fitted value, i.e., the residual. By calculating the MRE, MAE, and MSE between the predicted value and the actual value of the PIONA value and the distillation range, the results are shown in Table 4:

[0151] Table 4 Evaluation indicators of prediction models

[0152]

[0153] As can be seen from Table 4, except for cycloalkanes, the MREs of the remaining attributes are all within 5%, and the MAE and MSE are also relatively small. Combined with the requirements for attributes in the subsequent naphtha configuration optimization modeling, the PIONA value is a key indicator affecting the destination of naphtha, and the normal alkane and isoalkane indicators are more important, so the relative error of the predicted normal alkane and isoalkane must be less than 5%, and the aromatic and cycloalkanes content is slightly lower, and the relative error must be less than 10%, so the model meets the accuracy requirements. In addition, the model is more accurate in distillation prediction, and the prediction results of the model are related to the frequency of naphtha attribute analysis in the LIMS system. The calibration is based on the data at the latest time point in the LIMS system, and the PIONA analysis cycle of the atmospheric and vacuum unit naphtha in the LIMS system is longer, and the frequency of distillation analysis is high.

[0154] On the other hand, an embodiment of the present invention further provides a prediction system for straight-run naphtha properties, comprising:

[0155] Memory; and

[0156] A processor is configured to execute the prediction method described in any of the above embodiments.

[0157] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0158] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0159] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0160] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0161] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0162] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0163] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0164] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0165] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for predicting properties of straight-run naphtha, characterized in that: include: Obtaining the proportion of each crude oil added to synthesize the predicted straight-run naphtha, the yield of the straight-run naphtha corresponding to each crude oil, and distillation process data; The proportion of each crude oil, the yield of straight-run naphtha corresponding to each crude oil, and the distillation range data are input into a pre-trained prediction model, and the first attribute of the straight-run naphtha to be predicted is obtained based on the distillation range cutting model in the prediction model.

2. The prediction method according to claim 1, characterized in that: The distillation process data include the initial distillation point, the final distillation point and the distillation ratio increased by each crude oil corresponding to the distillation temperature.

3. The prediction method according to claim 1, characterized in that: The distillation range cutting model is constructed according to the following formula: In the formula, M represents the number of crude oil types, P i Indicates the increase in distillation ratio corresponding to the distillation temperature, Yield k Indicates the yield of straight-run naphtha corresponding to each crude oil, Ratio k Indicates the proportion of each crude oil.

4. The prediction method according to claim 3, characterized in that: The distillation ratio of straight-run naphtha from each crude oil at a given distillation temperature is determined according to the following formula: Wherein, x = [x1, x2, ..., x7] represents the distillation temperature, w0, w1, ..., w n is the coefficient and n is the order.

5. The prediction method according to claim 4, characterized in that: The order is determined by the minimum of the following formula: In the formula, SSE represents the error sum of squares, m represents the number of samples, y(x i , W) represents the true value of the distillation ratio, t i represents the fitted value.

6. The prediction method according to claim 4, characterized in that: According to the following formula, determine the distillation ratio P of the corresponding straight-run naphtha distilled from each crude oil corresponding to the distillation temperature increase i : P i =y(x i+1 ,W)-y(x i ,W) i=1,2,...,N Where P i It indicates the distillation ratio corresponding to the increase in distillation temperature, and i is the total increase from the initial distillation point to the final distillation point.

7. The prediction method according to claim 1, characterized in that: The prediction method further comprises: The distillation range cutting model is corrected by a rolling correction method to obtain a corrected distillation range cutting model.

8. The prediction method according to claim 1, characterized in that: The prediction model also includes a pre-built linear prediction sub-model, and the prediction method also includes: Acquire the type of each crude oil added to synthesize the straight-run naphtha to be predicted and the second attribute of the straight-run naphtha corresponding to each crude oil; The type of each crude oil, the proportion of each crude oil, and the second attribute of the straight-run naphtha corresponding to each crude oil are input into the pre-built linear prediction sub-model to obtain the second attribute of the straight-run naphtha to be predicted.

9. The prediction method according to claim 8, characterized in that: The first attribute includes distillation process information of the straight-run naphtha to be predicted, and the second attribute includes at least one of a PIONA value, a yield, and a density of the straight-run naphtha to be predicted.

10. The prediction method according to claim 8, characterized in that: The construction of the linear prediction sub-model includes: Obtaining historical data of each crude oil in the mixed crude oil, wherein the historical data includes the type of each crude oil, the proportion of each crude oil, and the properties and yield information of the straight-run naphtha obtained by distilling each crude oil; Inputting the historical detailed evaluation data of each crude oil into the initial linear prediction sub-model to obtain the prediction attribute information corresponding to the straight-run naphtha distilled from the mixed crude oil; The real attribute information corresponding to the straight-run naphtha distilled from the mixed crude oil is obtained, and combined with the predicted attribute information, the initial linear prediction sub-model is corrected by using a rolling correction method to obtain the linear prediction sub-model.

11. The prediction method according to claim 8, characterized in that: The linear prediction sub-model is constructed according to the following formula: In the formula, n represents the number of crude oil types, Ratio i Indicates the proportion of each crude oil, X i and Yield i They represent certain properties and yields of straight-run naphtha obtained by distillation of each crude oil, and Y represents the prediction result corresponding to the property.

12. The prediction method according to claim 7 or 10, characterized in that: Correction using the rolling correction method includes correction according to the following formula: P out =P liner +k·(P LIMS -P liner ) Where P linear Represents an attribute data output by the linear prediction sub-model or distillation range cutting model, P LIMS Indicates the real data corresponding to the attribute, P out is the output result after correction, and k is the correction coefficient.

13. A prediction system for straight-run naphtha properties, characterized in that: include: Memory; as well as A processor, configured to execute the prediction method according to any one of claims 1 to 12.

Images