Abstract: In response to problems such as information silos, delayed regulation, and resource wastage in traditional greenhouse management, a method for constructing a 3D intelligent greenhouse management platform based on digital twin technology（referred to as intelligent management platform）was proposed．By integrating physical mechanism modelling with machine learning technology, a two-way dynamic mapping digital twin model was constructed to achieve real-time collection of greenhouse environmental data, model updates, and closed-loop control．Intelligent management platform adopted a layered architecture design, integrating multi-source sensing networks, edge computing, 3D visualisation, and intelligent decision-making algorithms to support whole-process closed-loop management from data collection to strategy feedback．Experiments showed that intelligent management platform significantly improved greenhouse management efficiency, with a 70% reduction in root-mean-square error for temperature prediction, 20% decrease in electricity consumption, and 25% decrease in water consumption．Concurrently, of tomato single plant yield increased by 14.6%, while fruit quality（sugar content, vitamin C content）and commercial fruit rate were significantly optimized．Research results provided theoretical support and technical reference for precision and intelligent management in smart agriculture．