Dynamic Optimization System for Dendrobium Officinale Cultivation Based on IoT Edge Computing and Hybrid Digital Twin

This study proposes an innovative Environment Control System (EDT-DHO) that optimizes the cultivation of Dendrobium officinale through the integration of Internet of Things edge computing and hybrid digital twin technology. The system breaks through three major bottlenecks: (1) Dynamic edge scheduling reduces delay from 287 seconds to 1823 seconds (a decrease of 36.5%); (2) Feature decoupling federated learning achieves a prediction accuracy of 92.7% with a privacy risk as low as 32%; (3) The FvCB-3DCNN hybrid model improves control precision to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\pm 0.7^{\mathrm{o}}\mathrm{C}$</tex>. Cross-climate experiments show that the EDT-DHO system increases production by 23.6% and saves energy by 42.6%. Experiments demonstrate that the EDT-DHO provides a reproducible technical framework for the intelligent upgrade of precision agriculture.