Root Causes Identification Approach Based on Association Rule Mining for Product Infant Failure

The high infant failure rate is a long-standing problem for complex engineering products, the studies about the infant failure mechanism still remain in its infancy which only depending on burn-in tests or failure rate estimations, these remedies, though, are palliatives. In order to resolve this dilemma, in view of the theory of big data and data mining, an approach based on association rule mining to identify the product infant failure root causes is proposed in this paper. Firstly, the connotation of product infant failure mechanism is summarized in the life-cycle view, and the root causes of product infant failure are expressed based on the domain mapping theory in axiomatic design, Secondly, the root causes of product infant failure are represented in the form of failure relational tree, and the association relationships are clarified also. The failure relation weight is mined based on the information entropy principles of association rule. Finally, the validity of the proposed method is verified by a case study of analyzing a computer board electricity infant failure, and the result shown that the proposed approach is conducive to identify the product infant failure root causes in the big data circumstances.