Design and Implementation of Soil Contaminant Level Detector by Optical Fluorescent Method in Earthworm Coelomic Liquid

This paper presents a portable and cost-effective soil contamination detector using an optical fluorescence-based approach with Lumbricus rubellus coelomic fluid as a natural biosensor. The AS7262 spectral sensor captures fluorescence responses under UV excitation at 395-400 nm, where PBS solvent enhances signal stability for up to 60 minutes before fluorescence quenching occurs. Detergent pollutants at concentrations of 0%, 10%, and 30% were tested, showing that higher coelomic fluid concentrations decrease fluorescence intensity as measured by the MFA Colorcompass spectrometer. Principal Component Analysis (PCA) and an Artificial Neural Network (ANN) were applied to classify contamination levels, achieving 70% accuracy. The novelty of this work lies in the integration of riboflavin-based natural fluorescence sensing, machine learning classification, and IoTbased real-time visualization into a portable system for in-situ soil monitoring, providing an accessible alternative to laboratory-based methods.