Product Key Reliability Characteristics Identification Method Based on XGBoost in Manufacturing Process

Manufacturing process is the last kilometer of product reliability from design to usage, and it plays a key role in the formation of product reliability. Previous studies mainly focused on fault diagnosis of independent production equipment in manufacturing process, ignoring the key reliability characteristics(KRCs) that affect the realization of design reliability index, which is the prerequisites for improving product reliability by monitoring and controlling the parameters that affect it during the manufacturing stage. Therefore, a key reliability characteristics identification method based on XGBoost in manufacturing process is proposed. First, based on the stream of variation (SoV) theory and RQR chain, from the viewpoint of loss and deviation principle of product reliability transmitted from design index to related parameters in the manufacturing process, the formation and failure mechanism of the product reliability in manufacturing process is expounded. Second, for the reliability improvement objective, the key reliability characteristics is defined based on the theory of key quality characteristics. Third, based on the collection and fusion of big data in manufacturing quality, a key reliability characteristics identification method based on XGBoost(KRCIBX) is established. Finally, an industrial example of subway converter is conducted to verify the proposed approach.