Basu, Asit; Rigdon, Steven (1991, ASQC) Department of Statistics, University of Missouri, Columbia, MO
This paper focuses on the reliability of repairable systems, clarifies various reliability concepts, and defines the power law process as a model for repairable systems. A list of definitions clarifies terminology that is often confusing and misleading.
There are two approaches to collecting data on a repairable system: (1) failure truncation, and (2) time truncation. Statistical inference procedures are slightly different for each.
Unlike nonrepairable systems, repairable systems can not be modeled by a single probability distribution. Since failure rates in repairable systems are not identically distributed or independent, a statistical model for repairable systems must consider these factors. The power law process, which is based on the Poisson process, is a useful technique for modeling the reliability of repairable systems.
The power law process, sometimes called the Weibull process, can be applied to many repairable systems. The biggest advantage of this process is the result of an intensity function. Because its intensity function can change over time, the power law process is flexible, giving it an advantage over the Poisson process.
Statistical methods,Reliability,Repairable systems