An Anomaly Detection Method for Network Freight Documents Based on Improved Multiple BRB

To address the problem of data confusion and regulatory difficulties in network freight transportation, a network freight bill abnormal detection method based on a Belief Rule Base (BRB) and Evidence Reasoning (ER) algorithm, and incorporating Differential Evolution Algorithm (DEA), is proposed. Firstly, to reduce the complexity of the BRB model, relevant features are selected for input into the model, and a BRB abnormal detection recognition model is established. Secondly, ER inference rules are used to reason the belief rules, and DEA is used to optimize and adjust the parameters of the BRB model. Finally, the experimental results are compared with artificial data and other non-BRB methods such as neural networks and support vector machines. The results show that the proposed abnormal detection model can effectively identify abnormal data, and some of the rules have been applied to the Network Freight Transportation Information Exchange System of the Ministry of Transport.