The geometric approximation to the cusum run length distribution (Q4730663)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 4115802
Language Label Description Also known as
default for all languages
No label defined
    English
    The geometric approximation to the cusum run length distribution
    scientific article; zbMATH DE number 4115802

      Statements

      The geometric approximation to the cusum run length distribution (English)
      0 references
      0 references
      0 references
      1989
      0 references
      Brownian motion
      0 references
      random walk
      0 references
      renewal theory
      0 references
      Cusum control scheme
      0 references
      approximations
      0 references
      To monitor industrial processes for continued satisfactory performance the Cusum control scheme is often used. The output of such a process is modelled by a random variable \(x_ t\) at time t with probability \(g_{\theta}(x_ t)\). The producer wishes to keep \(\theta \leq \theta_ 0\), but will not take corrective action until it is likely that \(\theta >\theta_ 1\) for some \(\theta_ 1>\theta_ 0.\) A random variable \(y_ t=y(x_ t,\theta_ 0,\theta_ 1)\) is defined such that \(y_ t\) tends to be negative when \(\theta <\theta_ 0\) and positive when \(\theta >\theta_ 1\). A Cusum against an increase in \(\theta\) is of the form \[ w_ t=\max (0,w_{t-1}+y_ t)\quad (t=1,2,...) \] with \(w_ 0=k\), for \(0\leq k<h<\infty\). \(T(k,h)=\min (t:w_ t\geq h)\) is the run length of the process, its distribution converges to the geometric form \(c\lambda^ T\). Two new approximations for c and three for \(\lambda\) are derived and compared numerically with existing ones.
      0 references

      Identifiers