Vehicle Defect Risk Early-Warning Model Based on Multi-Source Information Fusion

The recall management of defective automobiles is an important measure to improve the quality and safety of automobiles. Automobile defect determination is the basis of recall management, which is a complex system of engineering, including basic data acquisition, data analysis, engineering testing, etc. From the perspective of defect risk early-warning, this paper synthesizes the evaluation requirements for batch, safety, design, and manufacturing reasons regarding automobile defect determination; extracts the multi-source quality and safety factors, and constructs the indicator system for automobile defect risk early-warning. Based on the historical data of automobile recall, this paper analyzes the thresholds of quality and safety factors, and establishes logistic regression, classification tree, random forest, and bagging models based on automobile defect risk early-warning and analyzes the examples. The research has shown that the applicability and accuracy of the logistic model can best meet the actual needs of automobile defect risk early-warning.