Fuzzy Cognitive Map Optimisation via Genetic Algorithms for Non-Destructive Maturity Prediction of Hass Avocados Using Bioimpedance Spectroscopy

Key Points

• The aim is to develop a non-destructive method for predicting the maturity of Hass avocados using bioimpedance spectroscopy and fuzzy cognitive maps optimized by genetic algorithms.
• Longitudinal measurements were taken from 150 Hass avocados at 50 kHz and 100 kHz frequencies over four visits.
• Inputs such as soil nitrogen, phosphorus, potassium, and humidity were monitored, while employing a leave-one-fruit-out protocol for training.
• The framework accuracy was assessed by comparing with five baseline models using mean absolute error as the metric.
• FCM-GA achieved MAE = 0.053, significantly outperforming baseline methods (p < 0.01).
• Impedance measurements indicated 92% harvest identification effectiveness when plateau conditions were met.
• Soil nitrogen and humidity were identified as significant factors influencing impedance.

Abstract