Aggregate fingerprints identification based on its compositions and machine learning algorithm

The type and properties of an aggregate affect the properties of their mixtures with either Portland cements or asphalt binders. How to quickly identify the information on an aggregate, providing a reliable basis for the quality assurance and quality control of aggregates, i.e., guarantee the source of aggregates is vital important. The purpose of this study is to explore a new and rapid detective technology for aggregate fingerprint identification using Fourier Transform Infrared Spectroscopy (FTIR). Machine learning algorithm of statistical analysis software (SPSS) was performed for principal component analysis, cluster analysis and linear discriminant analysis on collected information of the aggregates. The results showed that the aggregates of the same origin can be aggregated well by principal component analysis, cluster analysis and linear discriminant analysis as well. The cross-validation accuracy is very high.