Due to the low prediction accuracy of the existing early warning methods, a multi-dimensional early warning method of digital e-commerce supply chain risk based on Internet of Things information fusion is studied. Based on the idea of scene information and complex event processing, the event information structure and event processing method based on the fusion of human-object-field information are proposed. Through negotiation between suppliers and retailers, the proportion of wholesale price and profit sharing is determined. Following the principle of selecting early warning indicators, the early warning indicators of financing risk are selected from four aspects: macro and industry environment, online supply chain operation, financing enterprises and core enterprises, and a risk early warning indicator system is constructed. The random forest algorithm is used to calculate the importance of each index and complete the index ranking accordingly, and the key early warning indicators are selected to calculate the potential risk loss of financing business for early warning. The experimental results show that when using this method to identify financing risks, the model can be the most accurate when the number of risk early warning indicators is 40; The accuracy of model prediction using test sample data is 94.59%.