Research on Reliability Assessment Methods Based on Zero-Failure Data

The occurrence of no-failure data is becoming more and more common in the life tests of high-reliability long-life products, which poses a challenge to reliability assessment. To address this problem, an improved Bayes method based on folded normal distribution is proposed to solve the reliability assessment problem under the condition of no-failure data. Based on the engineering experience, the improved Bayes method introduces the folded normal distribution as the prior distribution, which more reasonably reflects the distribution characteristics of the failure probability. The computational analysis based on the truncated test results shows that the improved Bayes method has good feasibility in the process of reliability assessment without failure data, and compared with the existing Bayes method, it can provide more reasonable and accurate reliability assessment results for high-reliability and long-life products. The proposed method provides a new way of thinking for the reliability assessment without failure data, and has good engineering application value.