Application and Comparison of Imputation Methods for Missing Degradation Data

A common problem in accelerated degradation testing (ADT) and prognostic and health management (PHM) is the missing of degradation data caused by failure of data transmission or manipulation errors. Facing with such cases, the missing data is usually ignored or even the whole group of data is abandoned. And the loss of valuable information may leads to inaccurate result in the following work. At present, there are various imputation methods have been applied to handling missing data in the field of statistics. These methods estimate the missing values by utilizing the observed data. Unlike most statistical data, degradation data changes over time. But the observed degradation data can still provide valuable information for the estimating. It is therefore reasonable to use these imputation methods to deal with the missing degradation data. The purpose of this paper is to investigate the possibility of using these methods for estimating missing values in degradation data. The missing mechanisms of degradation data are studied at first. Then three of the most widely used imputation methods are researched and used. And comparisons are carried out to show the efficiency of the three methods.