Abstract: Three-parameter Weibull distribution (TPWD) is widely used in reliability analysis for products. It has a very good capability for data fitting and more unambiguous physical meaning for life evaluation to describe the life distribution of products that have the characteristics of progressive failure such as wear, fatigue, corrosion and aging failure. To meet the demand of statistical inference of accelerated life test (ALT) with competing failures when all the failure modes are three-parameter Weibull distribution, the statistical inference method of ALT with three-parameter Weibull distribution competing failures is studied based on the properties of TPWD and basic models of ALT. The maximum likelihood estimation model for parameters estimation is also built. Result of simulation example proves that the statistical inference method and model proposed is correct and has very good effect.

Key words: accelerated life test (ALT);varying-stress;competing failures;three-parameter Weibull distribution (TPWD);maximum likelihood estimation (MLE);reliability