Predicting field reliability based on two-dimensional warranty data with learning effects

Authors:
Publisher website
Google Scholar
Abstract
Understanding the field reliability of a sold product is crucial to both managers and engineers for monitoring product quality and improving warranty service design. When practitioners model warranty data, they it often assume that the lifetimes of products manufactured on different days is homogeneously distributed (i.e., product reliability remains the same over time). Based on a two-dimensional warranty data set collected from an automobile manufacturer, we find that the reliability of products improves over time. A log-linear regression model on the failure rate of the products is proposed by considering the usage rate and manufacturing day as covariates. The maximum likelihood approach is used to estimate the parameters. The results show the existence of learning effects in reliability in the early stage of manufacturing. A learning-curve model is then used to predict the reliability of new items produced.
Funding Information
  • National Natural Science Foundation of China (NSFC) (#71472132, #71225006, #71531010, #71325003, and #71532008)

This publication has 23 references indexed in Scilit: