mqr.spc.util.
solve_arl#
- mqr.spc.util.solve_arl(h4, p, lmda, N=20)#
Find the in-control ARL of an MEWMA chart.
Follows the modified Gauss-Legendre method in [1] to evaluate:
\[L(\alpha) = 1 + \int_0^h L(u)/\lambda^2 f(u/\lambda^2 | \eta(\alpha)) du\]where
\[\begin{split}\begin{gather} h = h_4 \lambda / (2 - \lambda),\text{ and} \\ \eta(\alpha) = \alpha ((1 - \lambda) / \lambda)^2. \end{gather}\end{split}\]The algorithm makes a change of variables (\(\alpha \rightarrow \alpha^2\)), which improves its performance.
- Parameters:
- h4float
Upper control limit on MEWMA chart.
- pint
Dimension of monitored vector.
- lmdafloat
Decay factor for EWMA statistic.
- Nint, optional
Number of quadrature sample points (Gauss-Legendre).
- Returns:
- float
References
[1]Knoth, S. (2017). ARL numerics for MEWMA charts. Journal of Quality Technology, 49(1), 78-89.