In statistical analyses of life data, the Bayesian approach regards parameters of the underlying life distribution as random variables to cope with the uncertainty included in estimates of these parameters as well as to make use of the prior information. The Bayesian method also enables us to deal with the reliability of a product or a part as a random variable since the reliability is expressed as a function of the parameters. This indicates that the reliability can be guaranteed to be larger than a specified value in terms of probability by adopting the Bayesian method.
Based on this idea, the present study considers a method for analyzing fatigue life data which usually include censored data, assuming a log-normal distribution as the underlying life distribution. The proposed method provides an estimate for the fatigue life whose reliability can be guaranteed to be larger than a specified value with specified probability. Numerical examples are also examined to clarify the characteristics of the proposed method and to show its effectiveness. Through the examination of numerical examples, it is verified that the proposed method provides conservative estimates for both complete data and censored data. It is also shown that the proposed method considerably reflects the specified guarantee probability for complete data, but guarantees the reliability with larger probability than the specified for censored data.