Research On Common Asphalt Brand Identification And Rapid Detection Of Technical Properties

Asphalt material is more and more used in highway engineering,its quality directly affects the durability and safety performance of the entire asphalt pavement.Therefore,it is critical to control the quality of asphalt.In some projects,the“three indicators”of asphalt can meet the requirements,but the asphalt quality is still not guaranteed during use,and it is highly possible to use low-quality asphalt instead of high-quality asphalt.Because there are many brands of asphalt at home and abroad,which lead to mixed market in the asphalt market,and other indicators besides the“three indicators”are cumbersome to test,so that the detection frequency in actual work is low,and the phenomenon of substandard filling of asphalt occurs from time to time.Therefore,it is of great significance to carry out the research on asphalt brand recognition and the simple nature of technical properties.This article will use infrared spectroscopy technology to analyze the advantages of fast speed,material composition to determine the characteristics of spectral composition,and to study the application of infrared spectroscopy technology in asphalt brand recognition and simple prediction of asphalt technical indicators.In this paper,based on the analysis of the optical characteristics of the attenuated total reflection Fourier infrared spectroscopy of asphalt,a number of common technical data sets and infrared data sets of asphalt samples are established.The chemometrics method is used to process a series of spectral data,and the optimal spectral pretreatment method is selected and fully demonstrated.The results show that the optimized spectral pretreatment method is”subtracting CO₂+Wavenumber selection+Background correction+Elimination of the outliers+logarithmic scaling unit variance”.In this dissertation,the two methods of pattern recognition,principal component analysis?PCA?and independent clustering cluster?SIMCA?,were used to establish the qualitative analysis model.Thus,the purpose of rapid identification of asphalt brand based on asphalt infrared spectroscopy was realized.And the pattern recognition performance of the established qualitative analysis model is verified with unknown samples.The results show that the clustering effect of the model is ideal.Under the 5%significant level,the inter-class distances between the asphalts are all greater than 1,the recognition rate and rejection rate are all higher than 90%,and the test results of the unknown asphalt samples on the model are all.correct.The paper uses two kinds of multivariate calibration methods,namely partial least squares?PLS?method and artificial neural network?ANN?method to analyze the infrared spectrum of asphalt,and establish the quantification of asphalt technical index value?conventional index and rheological index value?respectively.Predictive model,and finally using the infrared spectrum of unknown samples to verify the prediction ability of the two quantitative analysis models.The results show that for different indicators,the prediction results of the two models for the unknown asphalt technical indicators are close to the true values,but the modeling effect and prediction results of the ANN method are better than the modeling results and prediction results of the PLS method.Thus,based on the established model,a rapid detection of the asphalt technical index value is achieved based on the infrared spectrum of the asphalt.