Method for selecting representative molecules of asphalt aromatic fractions based on principal component analysis and hierarchical clustering
Patent Information
- Application Number
- CN202410220419.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-28
- Publication Date
- 2026-10-27
- Estimated Expiration
- 2044-02-28
AI Technical Summary
为需要根据具体工程的沥青试验数据结果进行代表性分子选取的问题提供有效方法
[0039]1)高效快捷地针对分子模拟的不同需求给出沥青分子建模时合适的芳香分代表性分子。
Smart Images
Figure CN118538310B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for selecting representative aromatic molecules when constructing a pitch molecular model in the field of molecular simulation, and particularly to a method for selecting representative aromatic molecules using quantum chemical calculations. Background Technology
[0002] Aromatic components, being the most abundant substance among the four components of asphalt, have a significant impact on asphalt properties. Their content and type affect not only the stability of asphalt but also its interfacial activity. To precisely control asphalt properties, it is necessary to study aromatic molecules in asphalt molecular models at the nanoscale. With the deepening research on asphalt molecular models, more and more aromatic molecule structures have been resolved and selected. However, current research often relies on researchers' subjective judgment to select aromatic molecules, and there is still a lack of systematic argumentation regarding which aromatic molecules should be selected for different types of asphalt. Furthermore, it is impractical and unreasonable to include all aromatic molecules in the model during simulation. Therefore, in practical applications, to reduce the difficulty of constructing asphalt molecular models without losing the accurate description of different asphalt systems, it is necessary to study the selection of representative aromatic molecules for different asphalt systems. Summary of the Invention
[0003] This invention addresses the shortcomings of existing technologies by providing a method for selecting representative aromatic molecules in asphalt based on principal component analysis and hierarchical clustering. In practical applications, this method reduces the difficulty of constructing asphalt molecular models without sacrificing accurate descriptions of different asphalt systems. It studies the selection of representative aromatic molecules for different asphalt systems and provides suitable representative aromatic molecules for asphalt molecular model construction. This provides an effective method for addressing the problem of selecting representative molecules based on asphalt test data from specific engineering projects.
[0004] To achieve the above objectives, the present invention adopts the following technical solution:
[0005] A method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering includes the following steps:
[0006] Step 1: Collect 20 aromatic molecules N1-N20 by consulting literature and databases;
[0007] Step 2: Use King Draw to identify the structural formulas of 20 aromatic molecules and convert them into three-dimensional structures to establish molecular models of the 20 aromatic molecules.
[0008] Step 3: Use Gaussian 09 to perform geometric optimization on the aromatic molecular structure to obtain a stable molecular configuration;
[0009] Step 4: Use Multiwfn software to perform wavefunction analysis on the optimized aromatic molecular structure to obtain molecular morphology and molecular polarity indices, and then use VMD software for visualization.
[0010] Step 5: Select seven molecular morphology indicators, including molecular weight, percentage of sp2 hybridized carbon atoms, van der Waals volume, density, van der Waals surface area, SDP, and MPP. Calculate the scatter matrix and Pearson correlation coefficient of the seven molecular morphology indicators to determine the strength of the linear correlation between them. Through principal component analysis and hierarchical clustering, study and analyze the molecular characteristics of 20 asphalt aromatic compounds from the perspective of molecular morphology, and obtain recommendations for selecting representative molecules.
[0011] Step 6: Select eight molecular polarity indices: ADCH charge, dipole moment, MPI, Pi, polar surface area ratio, average Mayer bond order, HOMO-LUMO band gap, and average normalized MCI. Calculate the scatter matrix and Pearson correlation coefficient of the eight molecular polarity indices to determine the strength of the linear correlation between them. Through principal component analysis and hierarchical clustering, study and analyze the molecular characteristics of 20 asphalt aromatic compounds from the perspective of molecular polarity, and obtain recommendations for selecting representative molecules.
[0012] Step 7: Summarize the characteristics of aromatic molecules in terms of molecular morphology and molecular polarity, as well as the differences between molecules. Based on the microscopic information obtained from the experiment, provide a suggested scheme for selecting representative molecules of asphalt aromatics.
[0013] To optimize the above technical solution, the specific measures also include:
[0014] Furthermore, the aromatic molecules collected in step 1 are derived from asphalt molecules classified under SARA components in the literature, as well as molecules considered as aromatics in various crude oil asphalt raw material component determination tests.
[0015] Furthermore, in step 3, the calculation method used for the geometric optimization is density functional theory, the functional is B3LYP, the basis set is 6-31G(d,p), and the convergence criteria used in the geometric optimization process are: maximum force less than 0.00045au, root mean square force less than 0.0003au, and maximum displacement less than and root mean square displacement less than Vibrational frequency analysis was performed on the optimized aromatic molecular structures to ensure that the vibrational frequencies of all aromatic molecular structures were positive.
[0016] Furthermore, the molecular weight is obtained through calculation of the molecular formula, and the sp... 2The proportion of hybrid carbon atoms was obtained through statistical analysis of the molecular structure. The van der Waals volume and van der Waals surface area were calculated using the Marching Tetrahedron algorithm, with the electron density isosurface taken as 0.001au. The density was calculated based on the van der Waals volume and molecular weight.
[0017] SDP and MPP are derived by fitting a plane containing the coordinates of all atoms using the least squares method based on the atomic coordinates of aromatic components. MPP is the root mean square deviation of all atoms from the fitted plane, and SDP is the maximum span perpendicular to the fitted plane.
[0018] Furthermore, in step 6, the molecular polarity index is calculated as follows:
[0019] The dipole moment of the aromatic component was calculated using the fuzzy spatial analysis module in Multiwfn software via the Hirshfeld weighting function; MPI is the mean absolute value of the molecular surface electrostatic potential; Pi is the average deviation of the molecular surface electrostatic potential; the polar surface area ratio refers to the ratio of the area with an absolute value of the molecular surface electrostatic potential greater than 10 kcal / mol to the total molecular surface area; the Mayer bond order between atoms in the aromatic component was calculated using the bond order analysis module in Multiwfn software, and the average Mayer bond order of the entire molecule was obtained; the HOMO-LUMO band gap of the aromatic component was calculated using the molecular orbital analysis module in Multiwfn software; the MCI of the aromatic component was calculated and its value was standardized; the ADCH charge was calculated using the population analysis module in Multiwfn software under the Hirshfeld population method with atomic dipole moment correction.
[0020] Furthermore, in step 5, the specific process of studying and analyzing the molecular characteristics of 20 asphalt aromatic components from a molecular morphology perspective using principal component analysis and hierarchical clustering methods to obtain recommendations for selecting representative molecules is as follows:
[0021] Z-score standardization was performed on the data of seven molecular morphology indicators. Principal component analysis was then used to transform highly correlated indicators into independent indicators without linear relationships. Principal component eigenvectors for each indicator were calculated. Based on the eigenvectors, the scoring functions of the first three principal components were calculated. Using the values of the scoring functions of the first three principal components as Euclidean coordinates, the Euclidean distances of the 20 aromatic compounds were calculated, and a hierarchical cluster diagram of the 20 aromatic compounds was drawn. Based on the hierarchical cluster diagram, suggestions for selecting representative molecules were given.
[0022] The selection recommendations are as follows:
[0023] If the molecular weight of the aromatic component is determined to be less than 180 g / mol, then N3 is selected as the representative molecule of the aromatic component in the modeling of asphalt molecules; if the molecular weight of the aromatic component is determined to be greater than 300 g / mol, then N2 is selected as the representative molecule of the aromatic component; if the SDP value of the aromatic component is determined to be greater than... The aromatic compound N4 was chosen as the representative molecule.
[0024] Furthermore, in step 6, the specific process of studying and analyzing the molecular characteristics of 20 asphalt aromatic components from the perspective of molecular polarity using principal component analysis and hierarchical clustering methods, and obtaining recommendations for selecting representative molecules, is as follows:
[0025] Z-score standardization was performed on the data of eight molecular polarity indices. Principal component analysis was then used to transform highly correlated indices into independent indices without linear relationships. Principal component eigenvectors for each indices were calculated. Based on the eigenvectors, the scoring functions of the first three principal components were calculated. Using the values of the scoring functions of the first three principal components as Euclidean coordinates, the Euclidean distances of the 20 aromatic compounds were calculated, and a hierarchical cluster diagram of the 20 aromatic compounds was drawn. Based on the hierarchical cluster diagram, suggestions for selecting representative molecules were provided.
[0026] The selection recommendations are as follows:
[0027] If a nitrogen or sulfur atom substituent is detected in the aromatic fraction during the experiment, N7 or N10 molecules are selected as representative molecules; if oxygen and nitrogen atoms are substituted, N14 molecules are selected as representative molecules; if hydroxyl and nitrogen atoms are substituted, aromatic fraction N15 molecules are selected as representative molecules; if oxygen and sulfur atoms are substituted, aromatic fraction N16 molecules are selected as representative molecules; if two oxygen atoms are substituted, aromatic fraction N19 molecules are selected as representative molecules.
[0028] Furthermore, the scoring function is specifically as follows:
[0029] PC1=0.43A-0.36B+0.49C-0.20D+0.47E+0.32F+0.29G
[0030] PC2=0.40A+0.24B+0.28C+0.42D+0.31E-0.46F-0.47G
[0031] PC3=0.16A+0.50B-0.02C+0.58D-0.06E+0.41F+0.46G
[0032] In the formula, PC1 is the first principal component, PC2 is the second principal component, PC3 is the third principal component, A represents the molecular weight after Z-score normalization, and B represents the sp...2 Hybridized carbon atom percentage, C represents van der Waals volume, D represents density, E represents van der Waals surface area, F represents SDP, and G represents MPP.
[0033] Furthermore, the scoring function is specifically as follows:
[0034] PC1 = 0.12H + 0.49I + 0.49J + 0.43K + 0.43L - 0.34M - 0.12N
[0035] PC2=-0.61H+0.11I+0.09J+0.37K+0.04L+0.40M+0.56N
[0036] PC3=0.66H+0.08I+0.14J-0.10K-0.10L+0.04M+0.72N
[0037] In the formula, PC1 is the first principal component, PC2 is the second principal component, PC3 is the third principal component, H represents the dipole moment after Z-Score normalization, I represents MPI, J represents Pi, K represents the polar surface area ratio, L represents the average Mayer bond order, M represents the HOMO-LUMO band gap, and N represents the average normalized MCI.
[0038] The beneficial effects of this invention are:
[0039] 1) It efficiently and quickly provides representative aromatic molecules for asphalt molecule modeling to meet different needs of molecular simulation.
[0040] 2) By studying the selection of representative aromatic molecules for different asphalt systems, the difficulty of constructing asphalt molecular models can be reduced without losing the accurate description of different asphalt systems.
[0041] 3) In-depth analysis of the molecular morphology and polarity indices of aromatic components was conducted, and their nanoscale properties were investigated. This provides assistance in understanding and describing the properties of complex asphalt systems.
[0042] 4) For major projects with experimental data, the representative molecules can be appropriately adjusted based on the experimental data. In the future, through quantum chemical calculations of more newly discovered aromatic molecules, the accuracy of this method in selecting representative molecules can be further improved.
[0043] 5) By following the main process steps of this method, detailed nanoscale studies can be conducted on other components of asphalt and other materials with similar properties in simulation, and a method for selecting representative molecules can be given, which has good scalability. Attached Figure Description
[0044] Figure 1The 20 asphalt aromatic fraction molecular structural formulas selected in this invention;
[0045] Figure 2 is a schematic diagram of the molecular morphology and molecular polarity indices of aromatic molecules calculated in this invention;
[0046] Figure 3 shows the scatter matrix and Pearson correlation coefficient of the aromatic molecular morphology index and molecular polarity index calculated in this invention.
[0047] Figure 4 shows the principal component scores of the aromatic molecular morphology and molecular polarity indices calculated in this invention.
[0048] Figure 5 is a systematic cluster diagram of the molecular morphology and molecular polarity indices of aromatic fractions calculated in this invention;
[0049] Figure 6 This invention provides representative molecular diagrams of aromatic components suitable for constructing asphalt molecular models to meet different needs in molecular simulation. Detailed Implementation
[0050] The invention will now be described in further detail with reference to the accompanying drawings.
[0051] A method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering includes the following steps:
[0052] Step 1: Consult literature and databases to collect 20 aromatic molecules (N1-N20); the collected aromatic molecules are asphalt molecules classified according to SARA components in the literature, as well as molecules regarded as aromatics in various crude oil asphalt raw material component determination tests.
[0053] Step 2: Use KingDraw software to identify aromatic molecules from literature and databases, obtain molecular structural formulas, check and correct the molecules to ensure consistency with the source molecules, manually draw difficult-to-identify molecules, and convert them into 3D models. 20 asphalt molecule structural formulas are shown below. Figure 1 As shown;
[0054] Step 3: The molecular structure was geometrically optimized using Gaussian 09 to obtain a stable molecular configuration. Density functional theory was selected as the calculation method, with B3LYP as the functional and 6-31G(d,p) as the basis set. The convergence criteria used in the optimization process were: maximum force less than 0.00045au, root mean square force less than 0.0003au, and maximum displacement less than [missing value]. and root mean square displacement less than Vibration frequency analysis was performed on the optimized structure to ensure that the vibration frequency of all structures was positive.
[0055] Step 4: Use Multiwfn software to perform wavefunction analysis on the optimized aromatic molecular structure to obtain molecular morphology and molecular polarity indices, and then use VMD software for visualization.
[0056] Step 5: Select molecular weight, sp. 2 The molecular characteristics of 20 asphalt aromatic compounds were studied and analyzed from the perspective of molecular morphology using seven indicators: the proportion of hybrid carbon atoms, van der Waals volume, density, van der Waals surface area, SDP and MPP (taking aromatic compound N1 as an example, its molecular morphology characteristics are shown in Figure 2(a)).
[0057] The molecular weight of aromatic components is obtained through calculation of the molecular formula. 2 The proportion of hybrid carbon atoms was obtained statistically from the molecular structure. Van der Waals volume and surface area were calculated using the Marching Tetrahedron algorithm, with the electron density isosurface taken as 0.001 au. Density was derived from van der Waals volume and molecular weight. The molecular planarity parameter (MPP) and span of deviation from the plane (SDP) were determined by fitting a plane containing all atomic coordinates using the least squares method based on the atomic coordinates of the aromatic components. MPP is the root mean square deviation of all atoms from the fitted plane, and SDP is the maximum span perpendicular to the fitted plane.
[0058] To understand the linear correlation among the seven variables, exploratory data analysis was conducted on the seven molecular morphology indicators. Scatter matrices and Pearson correlation coefficient plots were generated to determine the strength of the linear correlation among the molecular polarity indicators, as shown in Figure 3(a). To avoid multicollinearity, the data of the seven molecular morphology indicators were Z-score standardized, and principal component analysis was used to convert highly correlated indicators into mutually independent indicators without linear relationships. The principal component analysis results of the aromatic molecular morphology indicators after Z-score standardization are shown in Table 1.
[0059] Table 1. Principal component analysis results of aromatic molecular morphology indices
[0060]
[0061]
[0062] Based on the eigenvectors of the principal components, the scoring functions of the first three principal components are calculated, as shown in Equation (1).
[0063]
[0064] Where A-G represent the Z-Score normalized molecular weight, sp, and sqrt(s), respectively. 2 Hybridized carbon atom ratio, van der Waals volume, density, van der Waals surface area, SDP, and MPP. The scores of principal components PC1, PC2, and PC3 (the scores of the first three principal components of each aromatic fraction are shown in Figure 4(a)) were used as Euclidean space coordinates to calculate the Euclidean distances of the 20 aromatic fractions, and the corresponding hierarchical cluster diagrams were drawn (as shown in Figure 5(a)). Through the study of molecular morphology indicators of asphalt aromatic fractions, molecules with significant differences were successfully identified, and corresponding suggestions for selecting representative molecules were given. If the molecular weight of an aromatic fraction is experimentally determined to be less than 180 g / mol, then N3 is preferentially selected as the representative molecule of the aromatic fraction in the modeling of asphalt molecules; if the molecular weight of an aromatic fraction is determined to be greater than 300 g / mol, then aromatic fraction N2 is preferentially selected; if the SDP value of an aromatic fraction is greater than... The aromatic component N4 is preferred.
[0065] Step 6: Select eight indicators, including ADCH charge, dipole moment, MPI, Pi, polar surface area ratio, average Mayer bond order, HOMO-LUMO band gap and average normalized MCI, to study and analyze molecular polarity from the perspective of molecular polarity (taking aromatic N1 as an example, its molecular polarity characteristics are shown in Figure 2(b)).
[0066] The dipole moments of aromatic components were calculated using the fuzzy spatial analysis module in Multiwfn software, employing the Hirshfeld weighting function. The Molecular Polarity Index (MPI) is the mean absolute value of the molecular surface electrostatic potential; the Internal Charge Separation Index (Pi) is the average deviation of the molecular surface electrostatic potential; and the polar surface area ratio refers to the ratio of the area with an absolute electrostatic potential greater than 10 kcal / mol to the total molecular surface area. The Mayer bond order between atoms in the aromatic component was calculated using the bond order analysis module in Multiwfn software, and the average Mayer bond order of the entire molecule was obtained. The HOMO-LUMO band gap of the aromatic component was calculated using the molecular orbital analysis module in Multiwfn software. The Multi-Center Index (MCI) of the aromatic component was calculated and its value standardized. The atomic charge under the Atomic Dipole Moment Corrected Hirshfeld Population (ADCH) method was calculated using the population analysis module in Multiwfn software.
[0067] To understand the linear correlation among the eight variables, exploratory data analysis was conducted on the eight molecular polarity indices. Scatter matrices and Pearson correlation coefficient plots were generated to determine the strength of the linear correlation among the molecular polarity indices, as shown in Figure 3(b). To avoid multicollinearity, the data of the eight molecular polarity indices were Z-score standardized, and principal component analysis was used to convert highly correlated indices into mutually independent indices without linear relationships. The principal component analysis results of the aromatic polarity morphology indices after Z-score standardization are shown in Table 2.
[0068] Table 2. Principal component analysis results of aromatic molecular polarity indices.
[0069]
[0070] Based on the eigenvectors of the principal components, the scoring functions of the first three principal components are calculated, as shown in Equation (2).
[0071]
[0072] Where HN represents the dipole moment, MPI, Pi, polar surface area ratio, average Mayer bond order, HOMO-LUMO band gap, and average normalized MCI after Z-Score normalization, respectively. Using the scores of principal components PC1, PC2, and PC3 (the scores of the first three principal components of each aromatic fraction are shown in Figure 4(b)) as Euclidean space coordinates, the Euclidean distances of the 20 aromatic fractions were calculated, and the corresponding hierarchical cluster diagrams were drawn (as shown in Figure 5(b)). Through the study of the molecular polarity index of asphalt aromatic fractions, molecules with significant differences were successfully identified, and corresponding suggestions for selecting representative molecules were given. If a nitrogen or sulfur atom substituent is found in the aromatic fraction in the experiment, N7 or N10 molecules are preferred as representative molecules; if oxygen and nitrogen atoms are substituted, N14 is preferred as a representative molecule; if hydroxyl and nitrogen atoms are substituted, aromatic fraction N15 is preferred; if oxygen and sulfur atoms are substituted, aromatic fraction N16 is preferred; if two oxygen atoms are substituted, aromatic fraction N19 is preferred.
[0073] Step 7: Summarize the characteristics of aromatic molecules in terms of molecular morphology and molecular polarity, as well as the differences between molecules. Based on the microscopic information obtained from experiments, provide a suggested selection scheme for representative aromatic molecules (e.g., ...). Figure 6 (As shown).
[0074] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should be considered within the scope of protection of the present invention.
Claims
1. A method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering, characterized in that, Includes the following steps: Step 1: Collect 20 aromatic molecules N1-N20 by consulting literature and databases; Step 2: Use King Draw to identify the structural formulas of 20 aromatic molecules and convert them into three-dimensional structures to establish molecular models of the 20 aromatic molecules. Step 3: Use Gaussian09 to perform geometric optimization on the aromatic molecular structure to obtain a stable molecular configuration; Step 4: Use Multiwfn software to perform wavefunction analysis on the optimized aromatic molecular structure to obtain molecular morphology and molecular polarity indices, and then use VMD software for visualization. Step 5: Select molecular weight, sp. 2 Seven molecular morphology indicators, including the proportion of hybrid carbon atoms, van der Waals volume, density, van der Waals surface area, SDP, and MPP, were used to calculate the scatter matrix and Pearson correlation coefficient of the seven indicators and determine the strength of the linear correlation between them. Principal component analysis and hierarchical clustering were used to study and analyze the molecular characteristics of 20 asphalt aromatic compounds from the perspective of molecular morphology, and suggestions for selecting representative molecules were obtained. Step 6: Select eight molecular polarity indices: ADCH charge, dipole moment, MPI, Pi, polar surface area ratio, average Mayer bond order, HOMO-LUMO band gap, and average normalized MCI. Calculate the scatter matrix and Pearson correlation coefficient of the eight molecular polarity indices to determine the strength of the linear correlation between them. Through principal component analysis and hierarchical clustering, study and analyze the molecular characteristics of 20 asphalt aromatic compounds from the perspective of molecular polarity, and obtain recommendations for selecting representative molecules. Step 7: Summarize the characteristics of aromatic molecules in terms of molecular morphology and molecular polarity, as well as the differences between molecules. Based on the microscopic information obtained from the experiment, provide a suggested scheme for selecting representative molecules of asphalt aromatics.
2. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 1, characterized in that, The aromatic molecules collected in step 1 are asphalt molecules classified according to SARA components in the literature, as well as molecules considered as aromatics in various crude oil asphalt raw material component determination tests.
3. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 1, characterized in that, In step 3, the calculation method used for geometric optimization is density functional theory, with B3LYP as the functional and 6-31G(d,p) as the basis set. The convergence criteria used in the geometric optimization process are: maximum force less than 0.00045au, root mean square force less than 0.0003au, and maximum displacement less than... and root mean square displacement less than Vibrational frequency analysis was performed on the optimized aromatic molecular structures to ensure that the vibrational frequencies of all aromatic molecular structures were positive.
4. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 1, characterized in that, In step 5, the molecular weight is obtained by calculating the molecular formula, and the sp... 2 The proportion of hybrid carbon atoms was obtained through statistical analysis of the molecular structure. The van der Waals volume and van der Waals surface area were calculated using the Marching-Tetrahedron algorithm, with the electron density isosurface taken as 0.001 au. The density was calculated based on the van der Waals volume and molecular weight. SDP and MPP are derived by fitting a plane containing the coordinates of all atoms using the least squares method based on the atomic coordinates of aromatic components. MPP is the root mean square deviation of all atoms from the fitted plane, and SDP is the maximum span perpendicular to the fitted plane.
5. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 1, characterized in that, In step 6, the molecular polarity index is calculated as follows: The dipole moment of the aromatic component was calculated using the fuzzy spatial analysis module in Multiwfn software via the Hirshfeld weighting function; MPI is the mean absolute value of the molecular surface electrostatic potential; Pi is the average deviation of the molecular surface electrostatic potential; the polar surface area ratio refers to the ratio of the area with an absolute value of the molecular surface electrostatic potential greater than 10 kcal / mol to the total molecular surface area; the Mayer bond order between atoms in the aromatic component was calculated using the bond order analysis module in Multiwfn software, and the average Mayer bond order of the entire molecule was obtained; the HOMO-LUMO band gap of the aromatic component was calculated using the molecular orbital analysis module in Multiwfn software; the MCI of the aromatic component was calculated and its value was standardized; the ADCH charge was calculated using the population analysis module in Multiwfn software under the Hirshfeld population method with atomic dipole moment correction.
6. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 1, characterized in that, In step 5, the specific process of studying and analyzing the molecular characteristics of 20 asphalt aromatic compounds from a molecular morphology perspective using principal component analysis and hierarchical clustering methods, and obtaining recommendations for selecting representative molecules, is as follows: Z-score standardization was performed on the data of seven molecular morphology indicators. Principal component analysis was then used to transform highly correlated indicators into independent indicators without linear relationships. Principal component eigenvectors for each indicator were calculated. Based on the eigenvectors, the scoring functions of the first three principal components were calculated. Using the values of the scoring functions of the first three principal components as Euclidean coordinates, the Euclidean distances of the 20 aromatic compounds were calculated, and a hierarchical cluster diagram of the 20 aromatic compounds was drawn. Based on the hierarchical cluster diagram, suggestions for selecting representative molecules were given. The selection recommendations are as follows: If the molecular weight of the aromatic component is determined to be less than 180 g / mol, then N3 is selected as the representative molecule of the aromatic component in the modeling of asphalt molecules; if the molecular weight of the aromatic component is determined to be greater than 300 g / mol, then N2 is selected as the representative molecule of the aromatic component; if the SDP value of the aromatic component is determined to be greater than... The aromatic compound N4 was chosen as the representative molecule.
7. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 1, characterized in that, In step 6, the specific process of studying and analyzing the molecular characteristics of 20 asphalt aromatic components from the perspective of molecular polarity using principal component analysis and hierarchical clustering methods, and obtaining recommendations for selecting representative molecules, is as follows: Z-score standardization was performed on the data of eight molecular polarity indices. Principal component analysis was then used to transform highly correlated indices into independent indices without linear relationships. Principal component eigenvectors for each indices were calculated. Based on the eigenvectors, the scoring functions of the first three principal components were calculated. Using the values of the scoring functions of the first three principal components as Euclidean coordinates, the Euclidean distances of the 20 aromatic compounds were calculated, and a hierarchical cluster diagram of the 20 aromatic compounds was drawn. Based on the hierarchical cluster diagram, suggestions for selecting representative molecules were provided. The selection recommendations are as follows: If a nitrogen or sulfur atom substituent is detected in the aromatic fraction during the experiment, N7 or N10 molecules are selected as representative molecules; if oxygen and nitrogen atoms are substituted, N14 molecules are selected as representative molecules; if hydroxyl and nitrogen atoms are substituted, aromatic fraction N15 molecules are selected as representative molecules; if oxygen and sulfur atoms are substituted, aromatic fraction N16 molecules are selected as representative molecules; if two oxygen atoms are substituted, aromatic fraction N19 molecules are selected as representative molecules.
8. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 6, characterized in that, The scoring function is specifically as follows: PC1=0.43A-0.36B+0.49C-0.20D+0.47E+0.32F+0.29G PC2=0.40A+0.24B+0.28C+0.42D+0.31E-0.46F-0.47G PC3=0.16A+0.50B-0.02C+0.58D-0.06E+0.41F+0.46G In the formula, PC1 is the first principal component, PC2 is the second principal component, PC3 is the third principal component, A represents the molecular weight after Z-score normalization, and B represents the sp... 2 Hybridized carbon atom percentage, C represents van der Waals volume, D represents density, E represents van der Waals surface area, F represents SDP, and G represents MPP.
9. The method for selecting representative molecules of asphalt aromatic components based on principal component analysis and hierarchical clustering as described in claim 7, characterized in that, The scoring function is specifically as follows: PC1=0.12H+0.49I+0.49J+0.43K+0.43L-0.34M-0.12N PC2=-0.61H+0.11I+0.09J+0.37K+0.04L+0.40M+0.56N PC3=0.66H+0.08I+0.14J-0.10K-0.10L+0.04M+0.72N In the formula, PC1 is the first principal component, PC2 is the second principal component, PC3 is the third principal component, H represents the dipole moment after Z-Score normalization, I represents MPI, J represents Pi, K represents the polar surface area ratio, L represents the average Mayer bond order, M represents the HOMO-LUMO band gap, and N represents the average normalized MCI.