Inferences by Using Convolution Models for Field Failure Warranty Data with Lag Time Problem

• Main Authors: Yen-YuChen, 陳彥伃
• Other Authors: Shuen-Lin Jeng
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/53342089593925509525

碩士 === 國立成功大學 === 統計學系碩博士班 === 98 === A warranty is a guarantee between a manufacturer and a consumer which requires the manufacturer to rectify failures that occurred in non-human factors within certain time. Most companies maintain warranty databases for purposes of warranty expense forecasting and the ODM (Original Design Manufacturers) penalty program. The warranty database may have the sales lag and report lag problems. These problems cause the difficulty of failure time analysis which the real interest is the failure time after the customer starts to use the product. In certain literature, the convolution models were applied to the warranty database with the lag time problem. However, not much research has been done for the problem that the data have two types of lag at the same time. In this study, we propose a new method which are used to estimate the report lag distribution, when the report lag time are not recorded in the warranty databases. We also build a flexible parametric model which describes the failure rate of field data that are collected with sales lag and report lag and predicts the future failure rate with uncertainty limits. Finally, we use the predicted failure rate to estimate the required spare parts, to indicate the possible quality outbreak for early warning and suggest a ODM penalty rule. A new graphical tool, aggregate failure-rate monitor plot, is proposed for the surveillance of product quality.