Research Progress on Intelligent Algorithms for the Detection of Polycyclic Aromatic Hydrocarbons

Abstract

Polycyclic aromatic hydrocarbons(PAHs) are environmentally persistent, bioaccumulative, and pose a potential carcinogenic risk. Their trace, rapid, and accurate detection is of great significance for environmental monitoring and risk assessment. Addressing issues such as peak overlap, structural similarity, matrix interference, and signal fluctuations at low concentrations in the spectral detection of PAHs. This review summarizes research progress on the application of intelligent algorithms in the classification and identification of PAHs, concentration prediction, and the analysis of mixed components. First, the paper reviews recent research progress in the application of machine learning to PAHs detection, including traditional machine learning methods such as principal component analysis(PCA), support vector machines(SVM), random forests(RF), and partial least squares regression(PLSR), as well as examples of their application. Next, the paper explores deep learning methods—including convolutional neural networks(CNN), long-short-term memory networks(LSTM), and attention mechanisms,as well as their performance and practical applications. It analyzes the accuracy, robustness, and field applicability of intelligent PAHs detection and outlines future research directions in deep learning.