Publication

Abstract : Exposure to multiple environmental pollutants necessitates advanced monitoring techniques that can detect and classify multiple analytes. This work presents a novel AI-driven graphene oxide (GO) composite sensing system that addresses the limitations of single-parameter sensors through intelligent, real-time performance capabilities. A customised 9-sensor panel was developed using functionalized GO composites (Au-NP/GO, Ag-NP/GO, Pd-NP/GO, TiO₂/GO, ZnO/GO, and Fe₃O₄/GO) specialized for distinct pollutant classes. The system incorporates a dual hybrid neural network combining convolutional neural networks and bidirectional long short-term memory networks with multi-head attention mechanisms for enhanced feature extraction and classification accuracy. The AI-enhanced system achieved a classification performance of 96.2 % ± 1.4 % across heavy metals, pharmaceuticals, pesticides, and industrial chemicals, with detection limits ranging from 0.58 to 2.34 μg/L. Edge computing implementation enabled real-time processing under 100 ms. Multi-task learning facilitated the simultaneous identification and concentration estimation of pollutants, with mean absolute errors below 3.2 μg/L in complex environmental matrices. Six-month field deployments achieved 97.8 % operational uptime across urban environments. Comparative analysis against EPA reference methods demonstrated superior correlation coefficients (0.887–0.997) and 2.1–3.4× improved detection limits. The system identified 14 pollution episodes with zero false negatives, significantly outperforming conventional monitoring strategies. This intelligent sensing platform represents a paradigm shift toward autonomous environmental monitoring, offering robust solutions for protecting human health and environmental quality.