Integration of historical data and a small sample of data reliability assessment methods

To extract valuable information from a small sample of fatigue failure data completely and accurately, the author proposes a theory of historical data fusion for a certain kind of products under different stress levels. It is presumed that their fatigue life distribution obeys three-parameter Weibull distribution. The shape parameter is estimated with historical shape parameter under other stress levels under a proper assumption. The relationship between the scale parameter and stress is the main study content. The scale parameter is estimated with data under different stress levels using the Inverse Power Law Model and a critical approximation. At last using the small sample the positional parameter can be computed. The existing historical data fusion methods do not take the differences among three parameters into account. For the reason that different parameters have different relationship with historical data, so it is more proper to use corresponding method to each parameter.