ABSTRACT
For solid state drive (SSD) with high reliability and long life, it is nearly impossible to obtain sufficient amount of time-to-failure data within acceptable testing time by testing such products under normal operating environments. Thus, accelerated degradation testing (ADT) are introduced to solve reliability modeling problems based on the products’ degradation information obtained from accelerated tests. The intimate link between performance degradation data and product failures can be obtained according to the degradation threshold failure mechanism. Through the hypothetical degradation process and failure time, we can estimate the failure time with a given threshold value. However, due to diverse users, and uncertainty of what explicit level of degradation will cause a failure, a probabilistic, rather than a deterministic threshold value should be taken into account. On the other hand, complex operation environment would result in inevitable noise, so it is necessary to consider the detecting error in the reliability analysis. This paper propose a reliability assessment method based on fuzzy failure threshold and measurement errors, aiming at improving the assessment precision. We establish the degradation modeling with fuzzy failure threshold and measurement errors, and the maximum likelihood estimation method is adopted to estimate the failure time distribution parameters. Then the reliability model can be used for subsequent forecasting and decisionmaking. A commercial off-the-shelf SSD is shown as an example to illustrate the procedure that how to predict time to failure, of which writing current is used as a precursor parameter and directly monitored. Finally the results show the superior performance of the proposed method over traditional methods.
