Sensing pressure is widely needed technique in the industry such as plant, vehicle, aerospace, and etc. Since each pressure sensor has its own ranges of operating conditions, but the manufacturers give quite large design margins. So although the sensors has own lifetimes, they are regarded as hardly failed products. This study developed the accelerated life test system of pressure sensors, and the failure monitoring and predicting module in the system was a differentiated tool developed using a data-driven algorithm. This module were expected to save the test time since specimens which do not fail yet can be regarded as failed ones. As a preliminary commitment, 8 specimens were used in the test bench. Although classical Weibull analysis and accelerated life test method can handle these censored data, prognostic judgments of the censored data can give another analysis result. The signal-based analysis in this study showed that mean and deviation values of pressure sensor signals did not represent the failure prediction metric and so researchers should devised other approaches to substitute the classical reliability analysis.
Pecht, M. G. (2008). Prognostics and Health Management of Electronics. Hoboken, NJ: John Wiley & Sons, Inc