The Application of Infrared Spectroscopy in Identifying Different Brands of Base Asphalt
Tianjin Port East Technology Co., Ltd. Applied Analysis Department
Due to variations in origin and brand, there are slight differences in the chemical composition of base asphalt. The ability to accurately identify the origin and brand of base asphalt is of great practical significance for both the construction supervision team and the contractor in terms of quality oversight and quality control.
This study focused on base asphalt samples from two brands, Kelian 90 and Mingyuan 90. Spectral scans of the samples were performed using mid-infrared spectroscopy, and a model for distinguishing between base asphalt brands was established using chemometric methods. The results indicate that, following spectral preprocessing, the established models for identifying the origin or brand of base asphalt yielded highly accurate predictions when applied to samples not included in the model training.
1、Experimental Conditions
1.1 Materials and Methods
The asphalt reference samples were obtained from a construction project in Xinjiang and include two brands: Kelian 90 and Mingyuan 90. The samples were stored at a room temperature of 25°C.
1.2 Instruments and Accessories
FTIR-650 High-Specification Fourier Transform Infrared Spectrometer;
Liquid Testing Accessories Single-use ATR accessory (zinc selenide crystal, 45° incidence);
1.3 Test Conditions
Resolution: 4 cm⁻¹
Number of scans: 32
2、Genetic mapping
2.1Overlay of test spectra
2.2Similarity Analysis
A similarity analysis was conducted on the two types of samples. The results are shown in the figure above. The results indicate that there is no significant difference between the base asphalt samples from the two brands; therefore, it is not possible to distinguish between the two brands based solely on the similarity analysis.
2.3Multivariate Statistical Analysis
A multivariate statistical analysis was conducted on two brands of asphalt.
Principal component analysis and partial least squares discriminant analysis were used to analyze asphalt samples from the two brands. As shown in the figure above, the data for the two brands of asphalt are distributed according to their respective characteristics.
2.4Classification and Pattern Recognition
Asphalt Brand Identity
Pattern recognition was performed on the test set data for two brands of asphalt using the Partial Least Squares Discriminant Analysis (PLS-DA) algorithm. The results are as follows:
3、Conclusion
Spectral scans of the samples were performed using mid-infrared spectroscopy to study matrix asphalt samples from the Kelian 90 and Mingyuan 90 brands. A brand identification model for matrix asphalt was established using chemometric methods; The results indicate that, following spectral preprocessing, the established asphalt brand identification model achieved a 100% identification rate for both Keliang 90 and Mingyuan 90 asphalt samples when applied to test samples not used in the model training, yielding highly satisfactory results.