Statistical analysis method for accelerated life testing with incomplete data and competing failure modes

This paper proposes a statistical-analysis method for accelerated life testing (ALT) with incomplete data and competing failure modes. These failure modes can be divided into sudden and degradation failure modes; sudden failure modes are further divided into those with complete data, undefined failure-mode data, and truncated data. For sudden failure modes, a maximum likelihood estimation (MLE) method based on the expectation maximisation (EM) algorithm is proposed, which mitigates the issue concerning incomplete data – which traditional methods cannot solve – and has higher accuracy than MLE. For degradation failure modes, the Wiener process is used to model the degradation, since it can effectively describe the randomness of the performance degradation process. Further, considering the coupling–competing relationship between failure modes, a statistical-analysis model is proposed to comprehensively evaluate the reliability of a given product. Finally, a simulated and applied case study is provided to verify the accuracy of the proposed method.