Crop pathogens pose a significant threat to global agricultural production, resulting in substantial yield and economic losses. Conventional detection methods often exhibit limitations in accuracy, speed, and timely intervention. To address these challenges, this study presents a Pentagon-Shaped Terahertz (THz) Photonic Crystal Fiber (PCF) Biosensor integrated with a Decision-Cascaded 3D-Return-Dilated Secretary-Bird-Aligned Convolutional Transformer Network (DC3D-SBA-CTN). The proposed model employs a multi-stage feature extraction approach using Cascaded 3D-Dilated Convolutional Networks (CD-Net) and a Return-Aligned Decision Transformer (RADT) for accurate pathogen classification. Parameter optimization uses the Secretary-Bird Optimization Algorithm (SBOA), enhancing robustness and reducing false positives.

The biosensor's innovative pentagon-shaped design optimizes light–matter interactions, achieving heightened sensitivity and minimal signal loss. Simulation and experimental evaluations validate the biosensor's exceptional performance, with a detection accuracy of 99.9%, demonstrating resilience against morphological and environmental variations. Additionally, the system’s adaptability ensures its applicability across diverse agricultural settings, providing a reliable solution for real-time pathogen detection. The proposed solution enhances early intervention capabilities, contributing to reduced crop loss and increased agricultural productivity.

These findings establish the Pentagon-Shaped THz PCF Biosensor with DC3D-SBA-CTN as a transformative advancement in smart agricultural technology. The proposed approach contributes to precision farming and supports sustainable agricultural practices by enabling early detection and intervention.