GP-Optimized E-Nose: A Low-Cost Solution for Apple Pesticide Detection

This study presents a cost-effective, user-friendly electronic nose (e-nose) system for detecting Malathion pesticide residues on apples. Genetic Programming (GP) was utilized for system optimization. Adapting a previous methodology with different fruits and pesticides, we analyzed 100 apple samples (sprayed and control) using an 11-sensor e-nose. GP was employed to classify data, derive sensor output equations, and select optimal sensor combinations, by aiming to reduce hardware costs and computational load. The optimized system achieved a significant level of classification success. It demonstrated approximately 92\% validation accuracy with only three sensors and achieved 100\% validation accuracy when six sensors were used. These findings offer a viable, accurate, and economical alternative for pesticide detection. Future work is planned to expand the database and develop a home-use device.