Toxicity Detection in Water Based on Polycyclic Aromatic Hydrocarbons Using Machine Learning Algorithms

Polycyclic aromatic hydrocarbons (PAHs) are natural compounds having aromatic ring structures with carbon and hydrogen atoms. PAH can be found in our ecosystem such as air, water, soil, marine and in various food products. Excess presence of PAH is toxic to the environment and leads to carcinogenic effects on the organisms. Further, the presence of PAHs has been observed in aquatic systems which is the result of influent and effluent from wastewater treatment plants, seawater, surface water and ground water. Due to their high affinity and low solubility properties for particulates, PAH concentrations are less in water. However, the toxic nature of the water can be identified based on the concentration limit set by European standards and expert knowledge. Therefore, in this research it is proposed to detect the presence of toxicity in the water based on the PAHs. There are multiple methods to measure the PAH level like Gas chromatography /Mass Spectrometry (GC/MS) and High-Performance Liquid Chromatography (HPLC), but the detection of toxicity is through experimental analysis or expert knowledge. In this consideration, an investigative study has been conducted on the PAHs and the extracted feature sets were formed to evaluate through nonlinear machine learning algorithms such as Artificial Neural Networks. The experimental analysis has been validated through statistical measures and the results show the classification of toxicity in PAHs is promising.