Supply chain finance (SCF) plays a key role in easing financing difficulties for small and medium-sized enterprises, but it also comes with risks such as information asymmetry, fraud involving pledged assets, and delays in credit evaluation. In this study, we introduce a dynamic risk management framework driven by IoT and enhanced by the integration of multiple technologies. Built on a four-layer IoT structure, comprising perception, network, processing, and application layers, the framework combines blockchain for secure and trusted data sharing, federated learning for collaborative data processing, and digital twin models for real-time risk simulation. At the perception level, 5th-Generation Mobile Communication Technology (5G)-enabled low-power sensors ensure comprehensive and tamper-proof data collection. The network layer uses blockchain techniques such as sharding and zero-knowledge proofs to safeguard data privacy and institutional trust. In the processing layer, federated learning combined with edge and cloud computing enhances credit evaluation. On the other hand, the application layer employs smart contracts and feedback mechanisms to enable real-time responses and adaptive risk strategies. To put this framework into practice, we propose a phased approach: first building a real-time data ecosystem, then deploying secure risk control systems, optimizing distributed computing, and finally integrating a closed-loop risk control mechanism. This modular, collaborative strategy ensures that technological systems align with actual business needs. Ultimately, the research demonstrates how IoT, blockchain, and AI can work together to create a scalable and practical model for managing risk dynamically in SCF.