The Evaluation of Storage Life when Disordered Data Appear in Accelerated Performance Degradation Tests

Abstract

Abstract-Motivated by the fact that disordered data may appear in accelerated performance degradation tests when the sample is small, a new storage life evaluation method is proposed to mining information from the existing data. Combining the knowledge that the mean value of the performance parameter will degrade, isotonic regression are used to order the data. Meanwhile, the variance of the parameter is revised by mechanism consistency boundary. Finally, with the ordered data, linear degradation model and accelerated model are used to evaluate the storage life. The prediction method is applied to a battery's data which contains disordered data from accelerated performance degradation tests as example and works well.