Research on intelligent monitoring system of supply chain under multi-source heterogeneous data

With the increasing complexity of supply chain network and the increasing trend of multi-source data, the traditional monitoring method of single data source has been difficult to meet the real-time, accuracy and intelligence requirements. This paper proposes an intelligent supply chain monitoring system architecture based on multi-source heterogeneous data fusion, which integrates five types of data sources such as IoT sensors, RFID, satellite positioning, video surveillance and text information to build a unified deep embedded representation. In the fusion layer design, multi-head attention mechanism is introduced to realize dynamic weight allocation, and multi-task learning strategy is combined to accomplish the two core tasks of supply chain state prediction and risk detection. The experimental part conducts model comparison analysis based on real simulation data. The results show that the model proposed in this paper is superior to the traditional fusion model and the single attention mechanism model in key indicators such as accuracy, recall rate and F1 value, and verifies the effectiveness of the system in terms of fusion ability and prediction accuracy.