A method for predicting the yield of naphtha cracking furnace products
By selecting key variables and normalized processing, establishing and optimizing the product prediction model of naphtha cracking furnace, the problem of large errors in the existing technology is solved, and the accurate prediction and fault diagnosis of naphtha cracking furnace products is achieved, which improves the Economic benefits of naphtha cracking devices.
Patent Information
- Application Number
- CN202110310557.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-03-23
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2041-03-23
AI Technical Summary
The existing naphtha cracking furnace product yield prediction model has large errors, difficult to solve the model, frequent negative and extremely trace values, and the calibration error of industrial furnace models is large, making it difficult to achieve accurate prediction and fault diagnosis.
By selecting key variables and special normalization treatment methods, a cracking product prediction model based on naphtha cracking simulation evaluation device data is established, model verification and optimization is carried out, industrial cracking furnace data is associated, and prediction model is optimized to reduce errors and improve accuracy sex.
Accurate prediction of the product yield of naphtha cracking furnace is achieved, the calculation accuracy of the model is improved, and as a tool for fault diagnosis and operation optimization, the economic benefits of naphtha cracking device are improved.
Smart Images

Figure BDA0002987983710000081 
Figure BDA0002987983710000091 
Figure BDA0002987983710000101
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of petrochemical engineering, and more specifically to a method for predicting the yield of naphtha cracking furnace products. Background Art
[0002] Low-carbon olefins such as ethylene, propylene, butylene and butadiene are all very basic chemical raw materials. In particular, the production capacity of ethylene is often regarded as a symbol of the level of petrochemical development in a country or region, and is hailed as the cornerstone of the modern chemical industry. At present, about 98% of the world's ethylene production uses tubular furnace steam cracking technology. This process has a high ethylene yield and co-produces propylene, butadiene, benzene, toluene, xylene, C5, C9 and other basic chemical raw materials, playing a pivotal role in petrochemicals. The raw materials used for ethylene steam cracking are mainly light raw materials, such as ethane, C3, C4, light hydrocarbons, naphtha, light diesel, etc. Among them, naphtha is the most important ethylene cracking raw material, accounting for 50% of the world's ethylene production.
[0003] In recent years, my country's ethylene industry has developed rapidly. With the continuous expansion of the production capacity of ethylene cracking units, the supply of ethylene cracking raw materials has shown a trend of shortage, making the development and optimization of cracking raw materials increasingly valued by major ethylene producers. For ethylene steam cracking, raw materials account for about 70% of the ethylene production cost. In order to broaden the source of ethylene cracking raw materials, ethylene cracking raw materials are gradually developing in the direction of heavy quality and diversification. Due to the complexity of the composition of the cracking raw materials themselves and the frequent changes in their sources, coupled with the small batches of each raw material, and even frequent mixed storage, the cracking performance of the raw materials has changed. In particular, naphtha, which is the main raw material for steam cracking, has a complex composition and large differences between different sources, which can easily lead to problems such as low yield, high energy consumption, and poor economic benefits in the operation of the device, which brings considerable difficulties to the operation of the cracking furnace.
[0004] The cracking performance of different components in naphtha varies significantly. In terms of ethylene yield, the ethylene yield of normal alkanes is the highest, cycloalkanes are in the middle, and aromatics are the worst. Through the prediction technology of naphtha steam cracking product yield, the cracking product yield of the cracking furnace under different operating conditions can be predicted according to the changes of different naphtha raw materials in the ethylene plant, and then the more optimized operating conditions for the production of target products (ethylene + propylene, ethylene + propylene + butadiene, etc.) can be found, providing a basis for maximizing the value of naphtha raw materials and cracking furnace operation. In addition, through the high-accuracy cracking product yield prediction technology, compared with the actual working conditions of the cracking furnace, it can be used for fault diagnosis of the cracking furnace and judge the link where the fault occurs, such as abnormal furnace tube temperature, thermocouple indication deviation, raw material changes, etc. The simulation prediction of cracking product yield can also be used to evaluate the cracking furnace, determine whether the current raw material is suitable for the cracking furnace operation and provide an important reference for raw material optimization.
[0005] After years of research on the steam cracking reaction process, three methods for predicting the yield of steam cracking products have been formed: empirical model of thermal cracking reaction kinetics, semi-empirical and semi-theoretical model, and mechanism model. Foreign countries started early in the study of cracking product yield, and compiled it into cracking furnace simulation software for cracking furnace operation optimization, and achieved the improvement and optimization of cracking furnace ethylene and propylene yields, such as TECHNIP's SPYRO software.
[0006] With the continuous changes in the market situation of cracking products, in addition to the traditional trienes and triphenyls with high economic added value, cracking products such as hydrogen, C4 fraction, cracking gasoline, and cracking diesel have shown considerable economic value. For naphtha cracking companies, the growth point of their benefits will no longer rely solely on the yield of ethylene and propylene, but also on the comprehensive consideration of cracking products. Therefore, ethylene steam cracking companies need more comprehensive and accurate cracking product yield prediction technology.
[0007] Chinese patents CN103087750B and CN103087758B both disclose a method for constructing a naphtha industrial cracking furnace value maximization model. Both patents establish a naphtha cracking furnace cracking product yield prediction model based on naphtha steam cracking experimental data, and verify the prediction model by collecting the normal pentane yield data of the naphtha industrial cracking furnace. Afterwards, based on the verification model, different naphtha cracking furnace cracking product value maximization models are established, and used to optimize the operation of the naphtha cracking furnace, so that the total value of petrochemical products produced by the industrial naphtha cracking furnace is maximized. Among them, the establishment of a naphtha cracking furnace cracking product yield prediction model based on naphtha steam cracking experimental data is the core of the technology. Its yield prediction model is associated with a variety of naphtha physical properties and multiple cracking process conditions, and realizes the yield prediction of methane, ethylene, ethane, propylene, propane, carbon four components, carbon five components, benzene, toluene, xylene, ethylbenzene, styrene, cracking gasoline, cracking diesel, cracking fuel oil, etc. However, since some naphtha characterization analyses can represent the same or similar naphtha physical properties, the physical property parameters in the model are repeated, and too many parameters increase the difficulty of model solution. Secondly, the model calibration of industrial furnaces using single component data has a large error. In addition, components with low cracking product yields may have negative values, extremely small values, etc. in the actual model calculation, which deviates greatly from the actual value, and the accuracy of the prediction model still needs to be improved. Summary of the invention
[0008] The purpose of the present invention is to solve the above problems and provide a method for predicting the yield of naphtha cracking furnace products. By selecting key variables and a special normalization processing method, the accuracy of the prediction results is improved. The prediction results are associated with the data of the industrial cracking furnace to achieve accurate prediction of the cracking products of the industrial cracking furnace, so as to solve the problem of reduced yield of cracking furnace products caused by changes in naphtha raw materials. It can also be used as a guiding tool for fault diagnosis and troubleshooting of cracking furnaces, realize the optimization of the operation of naphtha cracking units and the adjustment of their product structure, and improve the economic benefits of naphtha cracking units.
[0009] In order to achieve the above object, the present invention provides a method for predicting the yield of naphtha cracking furnace products, comprising the following steps:
[0010] Step (1): Establish a cracking product prediction model: Based on the data of the naphtha cracking simulation evaluation device, the physical properties of naphtha (including: normal alkane mass content, isoalkane mass content, cycloalkane mass content, olefin mass content, aromatic mass content, hydrogen content, average molecular weight) and cracking process conditions (including: furnace tube outlet temperature, dilution ratio, residence time) are used as independent variables, and the cracking product yield is used as the dependent variable. A cracking product prediction model is constructed, which can be expressed as:
[0011] yi=f(T, R, t, Pn, Pi, N, O, A, H, M)
[0012] Where yi represents the yield of different pyrolysis product components
[0013] T-furnace tube outlet temperature, i.e. COT temperature (℃)
[0014] R-dilution ratio, i.e. the mass ratio of dilution steam to feed raw material (kg / kg)
[0015] t-residence time (s)
[0016] Pn- normal alkane content (ω%)
[0017] Pi-isoparaffin content (ω%)
[0018] N-cycloalkane content (ω%)
[0019] O-olefin content (ω%)
[0020] A-aromatic content (ω%)
[0021] H-Hydrogen content (ω%)
[0022] M-average molecular weight
[0023] Step (2): Model calibration and optimization: Based on the data of the naphtha cracking simulation evaluation device, the cracking product yield prediction model obtained in step (1) is used to calculate the cracking product yield. The calculation result is normalized and then compared with the data of the naphtha cracking simulation evaluation device. The prediction model with a large relative error is eliminated to complete the overall calibration of the prediction model; then, the difference in the yield of the same cracking product calculated by different prediction models is compared, and the mathematical model with the smallest difference is selected as the calculation model of the cracking product yield, the cracking product yield model structure is optimized, and the optimal cracking product prediction model is obtained.
[0024] Step (3): Correlation with industrial cracking furnace data: The optimal cracking product prediction model obtained in step (2) is used to correlate with the actual industrial cracking furnace calibration data to establish a naphtha industrial cracking furnace cracking product prediction model, which can be expressed as:
[0025] zi=φiyi+σi
[0026] Among them, zi represents the yield of different cracking product components of the industrial cracking furnace; φi represents the influence factor of the industrial cracking furnace on different cracking product components; σi is the influence parameter of the industrial cracking furnace's own characteristics on different cracking product components.
[0027] Step (4): Optimization of the prediction model for cracking products of the naphtha industrial cracking furnace: Based on the actual operating conditions of the industrial cracking furnace, the cracking product yield of the prediction model for the naphtha industrial cracking furnace obtained in step (3) is calculated. The calculated result is normalized and then compared with the actual industrial cracking furnace calibration data. The prediction model with a large relative error is eliminated to complete the optimization of the prediction model for cracking products of the naphtha industrial cracking furnace.
[0028] The above-mentioned naphtha cracking furnace product yield prediction method, the cracking product yield includes the yields of components such as hydrogen, methane, ethane, ethylene, acetylene, propylene, propane, propyne, propadiene, butane (isobutane, n-butane), butene (butene-1, isobutylene, trans-butene, cis-butene), 1,3-butadiene, benzene, toluene, ethylbenzene, xylene, styrene, cracked gasoline (excluding benzene, toluene, ethylbenzene, xylene, styrene), cracked diesel and cracked fuel oil.
[0029] Furthermore, the prediction models (cracking product prediction model and naphtha industrial cracking furnace cracking product prediction model) preferably use the least squares principle for regression calculation.
[0030] Furthermore, the normalization in the model verification and optimization in step (2) is to reduce the calculation error and the conditional restrictions adopted for individual products with small predicted yields. The normalization operation process is:
[0031] (1) adding restriction conditions to the individual products with lower yields in the yield prediction data to limit their yields to the same value, which is any value within the range of 0.01 to 0.20ω%;
[0032] (2) Maintaining the yields of products with larger yield values (such as the predicted yields of ethylene, propylene, butadiene, hydrogen, and methane) unchanged;
[0033] (3) Calculate the ratio of the sum of the predicted yields of the products with lower yield values to the sum of the predicted yields of the products with lower yield values, based on the value of the sum of the yields of the products with larger yield values deducted from 100%, as the normalization coefficient;
[0034] (4) Multiply the predicted yield of the product with a lower yield value by the normalization coefficient.
[0035] Furthermore, in the step (2), the individual products with lower yields in the model verification and optimization are components with a predicted yield of less than 0.001ω% or product components with a negative predicted result.
[0036] Furthermore, the principle of eliminating the prediction model with large relative error in the model verification and optimization is: for the product yield with large yield value (data result of naphtha cracking simulation evaluation device), that is, the general yield is higher than 2ω% cracking product, the relative error is >5%; for the product yield with small yield value (data result of naphtha cracking simulation evaluation device), that is, the general yield is not higher than 2ω% cracking product, the relative error is >15%.
[0037] Furthermore, in order to improve the screening efficiency of the prediction model, the principle of eliminating the prediction model with large relative error in the model verification and optimization is: the relative error of ethylene, propylene and butadiene yields is greater than 2.5%, and the relative error of other product yields with larger yield values is greater than 5%.
[0038] Furthermore, in the model verification and optimization, the difference in the yield of the same cracking product calculated by different prediction models is compared, and the different prediction models are prediction models after eliminating the prediction model with a large relative error, and the difference is the relative error between the prediction model calculated value of the same cracking product and the corresponding naphtha cracking simulation evaluation device data result.
[0039] Furthermore, the final cracking product yield prediction model in the model verification and optimization can be used to predict the cracking product yield of naphtha, and its calculation result can be used as the final yield prediction result after the above-mentioned normalization.
[0040] Furthermore, the actual industrial cracking furnace calibration data in the associated industrial cracking furnace data is the industrial cracking furnace outlet sampling calibration analysis result data.
[0041] Furthermore, in the associated industrial cracking furnace data, φi is a parameter related to the industrial cracking furnace operating conditions (cracking furnace outlet temperature, material residence time, etc.).
[0042] Furthermore, in the associated industrial cracking furnace data, σi is a parameter related to the industrial cracking furnace operating conditions (cracking furnace outlet temperature, material residence time, etc.).
[0043] Furthermore, the normalization method in the optimization of the prediction model for cracking products of the naphtha industrial cracking furnace is the same as the normalization method in the model verification and optimization.
[0044] Furthermore, the final naphtha industrial cracking furnace cracking product prediction model in the optimization of the naphtha industrial cracking furnace cracking product prediction model can be used to predict the cracking product yield of the naphtha industrial cracking furnace (the industrial cracking furnace from which the calibration data originates), and its calculation result can be used as the final yield prediction result after the above-mentioned normalization.
[0045] Compared with the prior art, the present invention has the following beneficial effects:
[0046] The present invention is based on the data of the naphtha cracking simulation evaluation device, and establishes an accurate cracking product yield prediction model by selecting key variables that have a greater impact on the cracking product yield. A special normalization processing method is used for the calculation results of the prediction model, especially for individual products with negative predicted yields, conditional restrictions are used to reduce the prediction error. Through strict yield relative error restriction conditions, available prediction models are screened, and each cracking product yield calculation model in the prediction model is optimized, thereby improving the accuracy of the prediction model calculation. In addition, in view of the problems of long adjustment cycle of industrial cracking furnace parameters and difficulty in verifying the effectiveness of adjustment, a prediction model for cracking products of naphtha industrial cracking furnace is established by associating the prediction model with actual industrial cracking furnace calibration data, and the model is calibrated for all component yields, thereby ensuring the accuracy of the prediction model and realizing accurate prediction of the yield of industrial cracking furnace products. DETAILED DESCRIPTION
[0047] The present invention is further described below by way of examples, which are only used to explain the technical solutions of the present invention and are not to be construed as limiting the technical solutions of the present invention.
[0048] The present invention provides a method for predicting the yield of naphtha cracking furnace products, comprising the following steps:
[0049] Step (1): Establish a cracking product prediction model: Based on the data of the naphtha cracking simulation evaluation device, the physical properties of naphtha (including: normal alkane mass content, isoalkane mass content, cycloalkane mass content, olefin mass content, aromatic mass content, hydrogen content, average molecular weight) and cracking process conditions (including: furnace tube outlet temperature, dilution ratio, residence time) are used as independent variables, and the cracking product yield is used as the dependent variable. A cracking product prediction model is constructed, which can be expressed as:
[0050] yi=f(T, R, t, Pn, Pi, N, O, A, H, M)
[0051] Where yi represents the yield of different pyrolysis product components
[0052] T-furnace tube outlet temperature, i.e. COT temperature (℃)
[0053] R-dilution ratio, i.e. the mass ratio of dilution steam to feed raw material (kg / kg)
[0054] t-residence time (s)
[0055] Pn- normal alkane content (ω%)
[0056] Pi-isoparaffin content (ω%)
[0057] N-cycloalkane content (ω%)
[0058] O-olefin content (ω%)
[0059] A-aromatic content (ω%)
[0060] H-Hydrogen content (ω%)
[0061] M-average molecular weight
[0062] Step (2): Model calibration and optimization: Based on the data of the naphtha cracking simulation evaluation device, the cracking product yield prediction model obtained in step (1) is used to calculate the cracking product yield. The calculation result is normalized and then compared with the data of the naphtha cracking simulation evaluation device. The prediction model with a large relative error is eliminated to complete the overall calibration of the prediction model; then, the difference in the yield of the same cracking product calculated by different prediction models is compared, and the mathematical model with the smallest difference is selected as the calculation model of the cracking product yield, the cracking product yield model structure is optimized, and the optimal cracking product prediction model is obtained.
[0063] Step (3): Correlation with industrial cracking furnace data: The optimal cracking product prediction model obtained in step (2) is used to correlate with the actual industrial cracking furnace calibration data to establish a naphtha industrial cracking furnace cracking product prediction model, which can be expressed as:
[0064] zi=φiyi+σi
[0065] Among them, zi represents the yield of different cracking product components of the industrial cracking furnace; φi represents the influence factor of the industrial cracking furnace on different cracking product components; σi is the influence parameter of the industrial cracking furnace's own characteristics on different cracking product components.
[0066] Step (4): Optimization of the prediction model for cracking products of the naphtha industrial cracking furnace: Based on the actual operating conditions of the industrial cracking furnace, the cracking product yield of the prediction model for the naphtha industrial cracking furnace obtained in step (3) is calculated. The calculated result is normalized and then compared with the actual industrial cracking furnace calibration data. The prediction model with a large relative error is eliminated to complete the optimization of the prediction model for cracking products of the naphtha industrial cracking furnace.
[0067] 1. Establish a prediction model for pyrolysis products
[0068] The PONA values, hydrogen content, average molecular weight and distillation range of different naphtha raw materials are statistically analyzed. The physical properties of the selected naphtha raw materials are: normal alkane content ranges from 8.0 to 63.5 ω, isoalkane content ranges from 10.5 to 75.2 ω, olefin content ranges from 0.1 to 23.6 ω, cycloalkane content ranges from 1.6 to 41.0 ω, aromatic content ranges from 0.9 to 16.8 ω, hydrogen content ranges from 12.8 to 16.3 ω, and molecular weight ranges from 70.1 to 114.1. Process range: 30 to 220 ℃.
[0069] The cracking process conditions are statistically analyzed. The cracking temperature range is 810-870°C, the dilution ratio range is 0.30-0.60kg / kg, and the residence time range is 0.10-0.55s.
[0070] The collected various naphthas were simulated for cracking evaluation, and the cracking product yields under the corresponding cracking conditions were calculated. This embodiment uses a steam cracking simulation evaluation test device to implement steam cracking experiments to obtain experimental data of different steam cracking process conditions for different naphthas. The physical properties of naphtha, cracking process conditions and cracking product yield results are matched one by one. The relevant steam cracking experimental analysis instruments and equipment are as follows:
[0071] (1) Analysis of naphtha group composition (normal alkanes, isoalkanes, olefins, cycloalkanes and aromatics) by Agilent HP-6890 gas chromatograph;
[0072] (2) The composition of the cracked gas, the composition of the liquid cracked products and their mass content were analyzed by HP-7890 gas chromatography from Agilent Technologies, USA.
[0073] For the same naphtha, according to the changing trend of cracking process conditions, the abnormal values of the cracking product yield results are eliminated or corrected. The physical discrimination method or statistical discrimination method is used to judge whether the cracking product yield results are abnormal.
[0074] Based on the data of naphtha cracking simulation evaluation device, a prediction model was constructed with the mass content of normal alkanes, isoalkanes, cycloalkanes, olefins, aromatics, hydrogen content, average molecular weight, furnace outlet temperature, dilution ratio and residence time as independent variables and the yield of cracking products as dependent variable. The specific cracking product yields include the yields of hydrogen, methane, ethane, ethylene, acetylene, propylene, propane, propyne, propadiene, butane (isobutane, n-butane), butene (butene-1, isobutylene, trans-butene, cis-butene), 1,3-butadiene, benzene, toluene, ethylbenzene, xylene, styrene, pyrolysis gasoline (excluding benzene, toluene, ethylbenzene, xylene and styrene), pyrolysis diesel and pyrolysis fuel oil.
[0075] 2. Model verification and optimization
[0076] According to the data of the naphtha cracking simulation evaluation device, the same naphtha is selected under the same cracking process conditions to calculate the yield of the established prediction model. The calculation results are normalized and compared with the verification data group data to calculate the relative yield of the corresponding cracking products and screen the prediction model. The prediction model retains the conditions that the relative error of the yield of ethylene, propylene, and 1,3-butadiene is less than 2.5%, the relative error of the yield of other products with larger yield values and higher fitting accuracy is less than 5%, and the relative error of the yield of products with smaller yield values and lower fitting accuracy is less than 15%. For products predicted to be negative, it is uniformly limited to 0.10ω%. The calibration of the naphtha cracking simulation evaluation device data used for model verification and the predicted yield is shown in Table 1 below:
[0077] Table 1 Comparison of cracking product prediction results and verification data
[0078]
[0079]
[0080] As can be seen from Table 1, after normalization of models 1, 2, and 3, the relative error between model 1 and the validation data meets the requirements, and the model is retained. After completing the screening of the prediction model for the yield of the pyrolysis product, the prediction model is optimized. For the yield of the same pyrolysis product, the actual calculated value (before normalization) of the screened prediction model is compared with the validation data set data, and the difference is calculated, that is, its relative error is calculated. The mathematical model with the smallest difference is selected as the calculation model for the yield of the pyrolysis product to obtain the final pyrolysis product yield prediction model. The selection and optimization of its product component model are shown in Table 2 below:
[0081] Table 2 Comparison of differences between different prediction models
[0082]
[0083] 3. Related industrial cracking furnace data
[0084] The above optimized prediction model is associated with the actual industrial cracking furnace calibration data to obtain the naphtha industrial cracking furnace cracking product prediction model.
[0085] Table 3 Before and after correlation, calibration data (normalized) and relative error
[0086]
[0087]
[0088] 4. Optimization of the prediction model for cracking products of naphtha industrial cracking furnace
[0089] According to the industrial cracking furnace calibration data, the prediction model obtained by the associated industrial cracking furnace data is optimized. As can be seen from Table 3, the relative error value between the calculation results of the prediction model obtained by the associated industrial cracking furnace data and the industrial cracking furnace calibration data meets the requirements that the general yield is higher than 2ω% of the cracking product, the relative error is ≤5%; the yield is not higher than 2ω% of the cracking product, the relative error is ≤15%, among which the relative error of the yield of ethylene, propylene and butadiene is less than 2%, and the prediction model obtained by the associated industrial cracking furnace data can be retained. If the relative error value is large, refer to the elimination method in Table 1.
[0090] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, technicians familiar with the field may make various corresponding changes and deformations based on the present invention, but these corresponding changes and deformations should all fall within the scope of protection of the claims of the present invention.
Claims
1. A method for predicting the yield of naphtha cracking furnace products, characterized in that: The naphtha cracking furnace product yield prediction method comprises the following steps: Step (1): Establishing a cracking product prediction model: Based on the data of the naphtha cracking simulation evaluation device, the physical properties of naphtha and the cracking process conditions are used as independent variables, and the cracking product yield is used as the dependent variable to construct a cracking product prediction model; Step (2): Model calibration and optimization: Based on the data of the naphtha cracking simulation evaluation device, the cracking product yield prediction model obtained in step (1) is used to calculate the cracking product yield. The calculation result is normalized and then compared with the data of the naphtha cracking simulation evaluation device. The prediction model with a large relative error is eliminated to complete the overall calibration of the prediction model. Then, the difference in the yield of the same cracking product calculated by different prediction models is compared, and the mathematical model with the smallest difference is selected as the calculation model of the yield of the cracking product. The cracking product yield model structure is optimized to obtain the optimal cracking product prediction model. Step (3): Correlation with industrial cracking furnace data: The optimal cracking product prediction model obtained in step (2) is used to correlate with the actual industrial cracking furnace calibration data to establish a naphtha cracking furnace cracking product prediction model; Step (4): Optimization of naphtha cracking furnace product prediction model: Based on the actual industrial cracking furnace operating conditions, the naphtha cracking furnace cracking product prediction model obtained in step (3) is used to calculate the cracking product yield. The calculation result is normalized and then compared with the actual industrial cracking furnace calibration data. The prediction model with a large relative error is eliminated to complete the optimization of the naphtha cracking furnace product prediction model. In step (1), the physical property parameters include: normal alkane mass content, isoalkane mass content, cycloalkane mass content, olefin mass content, aromatic mass content, hydrogen content and average molecular weight; the cracking process conditions include: furnace outlet temperature, dilution ratio and residence time; the cracking product yield is expressed as: yi=f(T, R, t, Pn, Pi, N, O, A, H, M) Wherein, yi represents the yield of cleavage products; T is the furnace tube outlet temperature °C; R is the dilution ratio kg / kg; t is the residence time s; Pn is the normal alkane content ω%; Pi is the isoparaffin content ω%; N is the cycloalkane content ω%; O is the olefin content ω%; A is the aromatic content ω%; H is the hydrogen content ω%; M is the average molecular weight; In step (3), the prediction model of cracking products of the naphtha industrial cracking furnace is expressed as: zi=φiyi+σi; Among them, zi is the yield of different cracking product components of the cracking furnace; φi is the impact factor of the cracking furnace on different cracking product components; σi is the impact parameter of the cracking furnace's own characteristics on different cracking product components; In step (3), the actual industrial cracking furnace calibration data is the industrial cracking furnace outlet sampling calibration analysis result data; In step (2) or step (4), the normalization operation process is: (1) Add restriction conditions to the individual products with yields less than 0.001ω% or negative values in the yield prediction data, restricting their yields to the same value, which is any value in the range of 0.01~0.20ω%; (2) Maintaining the yield of products with larger yield values unchanged; (3) Calculate the ratio of the sum of the predicted yields of the products with lower yield values to the sum of the predicted yields of the products with lower yield values, based on the value of the sum of the yields of the products with larger yield values deducted from 100%, as the normalization coefficient; (4) Multiply the predicted yield of the product with the lower yield value by the normalization coefficient.
2. The method for predicting the yield of naphtha cracking furnace products according to claim 1, characterized in that: The cracking product yields include the yields of hydrogen, carbon monoxide, carbon dioxide, methane, ethane, ethylene, acetylene, propylene, propane, propyne, propadiene, isobutane, n-butane, butene-1, isobutylene, trans-butene, cis-butene, 1,3-butadiene, benzene, toluene, ethylbenzene, xylene, styrene, cracking gasoline, cracking diesel and cracking fuel oil.
3. The method for predicting the yield of naphtha cracking furnace products according to claim 2, characterized in that: The pyrolysis gasoline does not contain benzene, toluene, ethylbenzene, xylene and styrene.
4. The method for predicting the yield of naphtha cracking furnace products according to claim 1, characterized in that: The product yields with larger numerical yields include the predicted yields of ethylene, propylene, butadiene, hydrogen, and methane.
5. The method for predicting the yield of naphtha cracking furnace products according to claim 1, characterized in that: In step (2), the principle of eliminating the prediction model with a large relative error is: for the product yield with a large yield value, the yield is higher than 2ω% of the cracking product, and the relative error is greater than 5%; for the product yield with a small yield value, the yield is not higher than 2ω% of the cracking product, and the relative error is greater than 15%.
6. The method for predicting the yield of naphtha cracking furnace products according to claim 5, characterized in that: In step (2), the principle for eliminating prediction models with large relative errors is: the relative errors of ethylene, propylene and butadiene yields are greater than 2.5%, and the relative errors of other products with large yield values are greater than 5%.
7. The method for predicting the yield of naphtha cracking furnace products according to claim 1, characterized in that: In step (2), in the comparison of the difference in the yield of the same cracking product calculated by different prediction models, the different prediction models are prediction models after eliminating the prediction model with a large relative error, and the difference is the relative error between the prediction model calculated value of the same cracking product and the corresponding naphtha cracking simulation evaluation device data result.
8. The method for predicting the yield of naphtha cracking furnace products according to claim 1, characterized in that: In step (3), the associated parameters φi and σi are related to the outlet temperature and material residence time of the industrial cracking furnace.
