Crow-AMSAA Confidence Bounds

From ReliaWiki

Jump to: navigation, search


Index

Appendix C  
Crow-AMSAA Confidence Bounds  

Contents


In this appendix, we will present the two methods used in the RGA software to estimate the confidence bounds for the Crow-AMSAA (NHPP) model when applied to developmental testing data. The Fisher matrix approach is based on the Fisher information matrix and is commonly employed in the reliability field. The Crow bounds were developed by Dr. Larry Crow.

Note regarding the Crow Bounds calculations: The equations that involve the use of the chi-squared distribution assume left-tail probability.

Individual (Non-Grouped) Data

This section presents the confidence bounds for the Crow-AMSAA model under developmental testing when the failure times are known. The confidence bounds for when the failure times are not known are presented in the Grouped Data section.

Beta

Fisher Matrix Bounds

The parameter \beta \,\! must be positive, thus \ln \beta \,\! is treated as being normally distributed as well.

\frac{\ln \hat{\beta }-\ln \beta }{\sqrt{Var(\ln \hat{\beta }})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds are given as:

C{{B}_{\beta }}=\hat{\beta }{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{\beta })}/\hat{\beta }}}\,\!

\alpha \,\! in {{z}_{\alpha }}\,\! is different ( \alpha /2\,\!, \alpha \,\! ) according to a 2-sided confidence interval or a 1-sided confidence interval, and variances can be calculated using the Fisher matrix.

\left[ \begin{matrix}
   -\tfrac{{{\partial }^{2}}\Lambda }{\partial {{\lambda }^{2}}} & -\tfrac{{{\partial }^{2}}\Lambda }{\partial \lambda \partial \beta }  \\
   -\tfrac{{{\partial }^{2}}\Lambda }{\partial \lambda \partial \beta } & -\tfrac{{{\partial }^{2}}\Lambda }{\partial {{\beta }^{2}}}  \\
\end{matrix} \right]_{\beta =\hat{\beta },\lambda =\hat{\lambda }}^{-1}=\left[ \begin{matrix}
   Var(\hat{\lambda }) & Cov(\hat{\beta },\hat{\lambda })  \\
   Cov(\hat{\beta },\hat{\lambda }) & Var(\hat{\beta })  \\
\end{matrix} \right]\,\!

\Lambda \,\! is the natural log-likelihood function:

\Lambda =N\ln \lambda +N\ln \beta -\lambda {{T}^{\beta }}+(\beta -1)\underset{i=1}{\overset{N}{\mathop \sum }}\,\ln {{T}_{i}}\,\!

And:

\frac{{{\partial }^{2}}\Lambda }{\partial {{\lambda }^{2}}}=-\frac{N}{{{\lambda }^{2}}}\,\!
\frac{{{\partial }^{2}}\Lambda }{\partial {{\beta }^{2}}}=-\frac{N}{{{\beta }^{2}}}-\lambda {{T}^{\beta }}{{(\ln T)}^{2}}\,\!
\frac{{{\partial }^{2}}\Lambda }{\partial \lambda \partial \beta }=-{{T}^{\beta }}\ln T\,\!

Crow Bounds

Failure Terminated

For the 2-sided (1-\alpha )\,\! 100% confidence interval on \beta \,\!, calculate:

\begin{align}
   {{D}_{L}}= & \frac{N\cdot \chi _{\tfrac{\alpha }{2},2(N-1)}^{2}}{2(N-1)(N-2)} \\ 
  {{D}_{U}}= & \frac{N\cdot \chi _{1-\tfrac{\alpha }{2},2(N-1)}^{2}}{2(N-1)(N-2)}  
\end{align}\,\!

Thus, the confidence bounds on \beta \,\! are:

\begin{align}
   {{\beta }_{L}}= & {{D}_{L}}\cdot \hat{\beta } \\ 
  {{\beta }_{U}}= & {{D}_{U}}\cdot \hat{\beta }  
\end{align}\,\!

Time Terminated

For the 2-sided (1-\alpha )\,\! 100% confidence interval on \beta \,\!, calculate:

\begin{align}
  & {{D}_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2(N-1)} \\ 
 & {{D}_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N}^{2}}{2(N-1)}  
\end{align}\,\!

The confidence bounds on \beta \,\! are:

\begin{align}
   {{\beta }_{L}}= & {{D}_{L}}\cdot \hat{\beta } \\ 
  {{\beta }_{U}}= & {{D}_{U}}\cdot \hat{\beta }  
\end{align}\,\!

Growth Rate

Since the growth rate, \alpha \,\!, is equal to 1-\beta \,\!, the confidence bounds for both the Fisher matrix and Crow methods are:

\alpha_L=1-\beta_U\,\!
\alpha_U=1-\beta_L\,\!

{{\beta }_{L}}\,\! and {{\beta }_{U}}\,\! are obtained using the methods described above in the confidence bounds on Beta.

Lambda

Fisher Matrix Bounds

The parameter \lambda \,\! must be positive; thus, \ln \lambda \,\! is treated as being normally distributed as well. These bounds are based on:

\frac{\ln \hat{\lambda }-\ln \lambda }{\sqrt{Var(\ln \hat{\lambda }})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on \lambda \,\! are given as:

C{{B}_{\lambda }}=\hat{\lambda }{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{\lambda })}/\hat{\lambda }}}\,\!

where:

\hat{\lambda }=\frac{n}{{{T}^{*\hat{\beta }}}}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta.

Crow Bounds

Failure Terminated

For the 2-sided (1-\alpha )\,\! 100% confidence interval, the confidence bounds on \lambda \,\! are:

\begin{align}
   {{\lambda }_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2{{T}^{{\hat{\beta }}}}} \\ 
  {{\lambda }_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N}^{2}}{2{{T}^{{\hat{\beta }}}}}  
\end{align}\,\!

where:

  • N\,\! = total number of failures.
  • T\,\! = termination time.

Time Terminated

For the 2-sided (1-\alpha )\,\! 100% confidence interval, the confidence bounds on \lambda \,\! are:

\begin{align}
   {{\lambda }_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2{{T}^{{\hat{\beta }}}}} \\ 
  {{\lambda }_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N+2}^{2}}{2{{T}^{{\hat{\beta }}}}}  
\end{align}\,\!

where:

  • N\,\! = total number of failures.
  • T\,\! = termination time.

Cumulative Number of Failures

Fisher Matrix Bounds

The cumulative number of failures, N(t)\,\!, must be positive, thus \ln N(t)\,\! is treated as being normally distributed.

\frac{\ln \hat{N}(t)-\ln N(t)}{\sqrt{Var(\ln \hat{N}(t)})}\ \tilde{\ }\ N(0,1)\,\!
N(t)=\hat{N}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{N}(t))}/\hat{N}(t)}}\,\!

where:

\hat{N}(t)=\hat{\lambda }{{t}^{{\hat{\beta }}}}\,\!
\begin{align}
   Var(\hat{N}(t))= & {{\left( \frac{\partial \hat{N}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial \hat{N}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial \hat{N}(t)}{\partial \beta } \right)\left( \frac{\partial \hat{N}(t)}{\partial \lambda } \right)cov(\hat{\beta },\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial \hat{N}(t)}{\partial \beta }= & \hat{\lambda }{{t}^{{\hat{\beta }}}}\ln t \\ 
  \frac{\partial \hat{N}(t)}{\partial \lambda }= & {{t}^{{\hat{\beta }}}}  
\end{align}\,\!

Crow Bounds

The Crow cumulative number of failure confidence bounds are:

\begin{align}
   {N(t)_{L}}= & \frac{t}{{\hat{\beta }}}{IFI}{{(t)}_{L}} \\ 
  {N(t)_{U}}= & \frac{t}{{\hat{\beta }}}{IFI}{{(t)}_{U}}  
\end{align}\,\!

where IFI{{(t)}_{L}}\,\! and IFI{{(t)}_{U}}\,\! are calculated using the process for calculating the confidence bounds on instantaneous failure intensity.

Cumulative Failure Intensity

Fisher Matrix Bounds

The cumulative failure intensity, {{\lambda }_{c}}(t)\,\!, must be positive, thus \ln {{\lambda }_{c}}(t)\,\! is treated as being normally distributed.

\frac{\ln {{{\hat{\lambda }}}_{c}}(t)-\ln {{\lambda }_{c}}(t)}{\sqrt{Var(\ln {{{\hat{\lambda }}}_{c}}(t)})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on the cumulative failure intensity are then estimated from:

CB={{\hat{\lambda }}_{c}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{\lambda }}}_{c}}(t))}/{{{\hat{\lambda }}}_{c}}(t)}}\,\!

where:

{{\hat{\lambda }}_{c}}(t)=\hat{\lambda }{{t}^{\hat{\beta }-1}}\,\!

and:

\begin{align}
   Var({{{\hat{\lambda }}}_{c}}(t))= & {{\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \beta } \right)\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{\lambda }_{c}}(t)}{\partial \beta }= & \hat{\lambda }{{t}^{\hat{\beta }-1}}\ln t \\ 
  \frac{\partial {{\lambda }_{c}}(t)}{\partial \lambda }= & {{t}^{\hat{\beta }-1}}  
\end{align}\,\!

Crow Bounds

The Crow bounds on the cumulative failure intensity (CFI)\,\! are given below. Let:

N=\hat{\lambda }{{t}^{{\hat{\beta }}}}\,\!

Failure Terminated

\begin{align}
   CFI{_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2\cdot t} \\ 
\end{align}\,\!
\begin{align}
  CFI{_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N}^{2}}{2\cdot t}  
\end{align}\,\!

Time Terminated

\begin{align}
   CFI{_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2\cdot t} \\ 
  CFI{_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N+2}^{2}}{2\cdot t}  
\end{align}\,\!

Cumulative MTBF

Fisher Matrix Bounds

The cumulative MTBF, {{m}_{c}}(t)\,\!, must be positive, thus \ln {{m}_{c}}(t)\,\! is treated as being normally distributed as well.

\frac{\ln {{{\hat{m}}}_{c}}(t)-\ln {{m}_{c}}(t)}{\sqrt{Var(\ln {{{\hat{m}}}_{c}}(t)})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on the cumulative MTBF are then estimated from:

CB={{\hat{m}}_{c}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{m}}}_{c}}(t))}/{{{\hat{m}}}_{c}}(t)}}\,\!

where:

{{\hat{m}}_{c}}(t)=\frac{1}{{\hat{\lambda }}}{{t}^{1-\hat{\beta }}}\,\!
\begin{align}
   Var({{{\hat{m}}}_{c}}(t))= & {{\left( \frac{\partial {{m}_{c}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{m}_{c}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{m}_{c}}(t)}{\partial \beta } \right)\left( \frac{\partial {{m}_{c}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\hat{\lambda })\,  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{m}_{c}}(t)}{\partial \beta }= & -\frac{1}{{\hat{\lambda }}}{{t}^{1-\hat{\beta }}}\ln t \\ 
  \frac{\partial {{m}_{c}}(t)}{\partial \lambda }= & -\frac{1}{{{{\hat{\lambda }}}^{2}}}{{t}^{1-\hat{\beta }}}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on the cumulative MTBF (CMTBF)\,\! are given by:

\begin{align}
 & CMTBF_{L}=\frac{1}{CFI_{U}} \\ 
 & CMTBF_{U}=\frac{1}{CFI_{L}}  
\end{align}\,\!

where CFI_L\,\! and CFI_U\,\! are calculated using the process for calculating the confidence bounds on cumulative failure intensity.

Instantaneous MTBF

Fisher Matrix Bounds

The instantaneous MTBF, {{m}_{i}}(t)\,\!, must be positive, thus \ln {{m}_{i}}(t)\,\! is treated as being normally distributed as well.

\frac{\ln {{{\hat{m}}}_{i}}(t)-\ln {{m}_{i}}(t)}{\sqrt{Var(\ln {{{\hat{m}}}_{i}}(t)})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on the instantaneous MTBF are then estimated from:

CB={{\hat{m}}_{i}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{m}}}_{i}}(t))}/{{{\hat{m}}}_{i}}(t)}}\,\!

where:

{{\hat{m}}_{i}}(t)=\frac{1}{\lambda \beta {{t}^{\beta -1}}}\,\!
\begin{align}
   Var({{{\hat{m}}}_{i}}(t))= & {{\left( \frac{\partial {{m}_{i}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{m}_{i}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{m}_{i}}(t)}{\partial \beta } \right)\left( \frac{\partial {{m}_{i}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\hat{\lambda }).  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{m}_{i}}(t)}{\partial \beta }= & -\frac{1}{\hat{\lambda }{{{\hat{\beta }}}^{2}}}{{t}^{1-\hat{\beta }}}-\frac{1}{\hat{\lambda }\hat{\beta }}{{t}^{1-\hat{\beta }}}\ln t \\ 
  \frac{\partial {{m}_{i}}(t)}{\partial \lambda }= & -\frac{1}{{{{\hat{\lambda }}}^{2}}\hat{\beta }}{{t}^{1-\hat{\beta }}}  
\end{align}\,\!

Crow Bounds

Failure Terminated

For failure terminated data and the 2-sided confidence bounds on instantaneous MTBF (IMTBF)\,\!, consider the following equation:

G(\mu |n)=\mathop{}_{0}^{\infty }\frac{{{e}^{-x}}{{x}^{n-2}}}{(n-2)!}\underset{i=0}{\overset{n-1}{\mathop \sum }}\,\frac{1}{i!}{{\left( \frac{\mu }{x} \right)}^{i}}\exp (-\frac{\mu }{x})\,dx\,\!

Find the values {{p}_{1}}\,\! and {{p}_{2}}\,\! by finding the solution G\left( \left. \frac{{{n}^{2}}}{c} \right|n \right)=\frac{\alpha }{2} and G\left( \left. \frac{{{n}^{2}}}{c} \right|n \right)=1-\frac{\alpha }{2} for the lower and upper bounds, respectively.

If using the biased parameters, \hat{\beta }\,\! and \hat{\lambda }\,\!, then the upper and lower confidence bounds are:

\begin{align}
  {{IMTBF}_{L}}= & IMTBF\cdot {{p}_{1}} \\ 
  {{IMTBF}_{U}}= & IMTBF\cdot {{p}_{2}}  
\end{align}\,\!

where IMTBF=\tfrac{1}{\hat{\lambda }\hat{\beta }{{t}^{\hat{\beta }-1}}}\,\!.

If using the unbiased parameters, \bar{\beta }\,\! and \bar{\lambda }\,\!, then the upper and lower confidence bounds are:

\begin{align}
   {{IMTBF}_{L}}= & IMTBF\cdot \left( \frac{N-2}{N} \right)\cdot {{p}_{1}} \\ 
  {{IMTBF}_{U}}= & IMTBF\cdot \left( \frac{N-2}{N} \right)\cdot {{p}_{2}}  
\end{align}\,\!

where IMTBF=\tfrac{1}{\bar{\lambda }\bar{\beta }{{t}^{\bar{\beta }-1}}}\,\!.

Time Terminated

Consider the following equation where {{I}_{1}}(.)\,\! is the modified Bessel function of order one:

H(x|k)=\underset{j=1}{\overset{k}{\mathop \sum }}\,\frac{{{x}^{2j-1}}}{{{2}^{2j-1}}(j-1)!j!{{I}_{1}}(x)}\,\!

Find the values {{\Pi }_{1}}\,\! and {{\Pi }_{2}}\,\! by finding the solution x\,\! to H(x|k)=\tfrac{\alpha }{2}\,\! and H(x|k)=1-\tfrac{\alpha }{2}\,\! in the cases corresponding to the lower and upper bounds, respectively. Calculate \Pi =\tfrac{4{{n}^{2}}}{{{x}^{2}}}\,\! for each case.

If using the biased parameters, \hat{\beta }\,\! and \hat{\lambda }\,\!, then the upper and lower confidence bounds are:

\begin{align}
   {{IMTBF}_{L}}= & IMTBF\cdot {{\Pi }_{1}} \\ 
  {{IMTBF}_{U}}= & IMTBF\cdot {{\Pi }_{2}}  
\end{align}\,\!

where IMTBF=\tfrac{1}{\hat{\lambda }\hat{\beta }{{t}^{\hat{\beta }-1}}}\,\!.

If using the unbiased parameters, \bar{\beta }\,\! and \bar{\lambda }\,\!, then the upper and lower confidence bounds are:

\begin{align}
   {{IMTBF}_{L}}= & IMTBF\cdot \left( \frac{N-1}{N} \right)\cdot {{\Pi }_{1}} \\ 
  {{IMTBF}_{U}}= & IMTBF\cdot \left( \frac{N-1}{N} \right)\cdot {{\Pi }_{2}}  
\end{align}\,\!

where IMTBF=\tfrac{1}{\bar{\lambda }\bar{\beta }{{t}^{\bar{\beta }-1}}}\,\!.

Instantaneous Failure Intensity

Fisher Matrix Bounds

The instantaneous failure intensity, {{\lambda }_{i}}(t)\,\!, must be positive, thus \ln {{\lambda }_{i}}(t)\,\! is treated as being normally distributed.

\frac{\ln {{{\hat{\lambda }}}_{i}}(t)-\ln {{\lambda }_{i}}(t)}{\sqrt{Var(\ln {{{\hat{\lambda }}}_{i}}(t)})}\text{ }\tilde{\ }\text{ }N(0,1)\,\!

The approximate confidence bounds on the instantaneous failure intensity are then estimated from:

CB={{\hat{\lambda }}_{i}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{\lambda }}}_{i}}(t))}/{{{\hat{\lambda }}}_{i}}(t)}}\,\!

where

{{\lambda }_{i}}(t)=\lambda \beta {{t}^{\beta -1}}\,\!
\begin{align}
   Var({{{\hat{\lambda }}}_{i}}(t))= & {{\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \beta } \right)\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{\lambda }_{i}}(t)}{\partial \beta }= & \hat{\lambda }{{t}^{\hat{\beta }-1}}+\hat{\lambda }\hat{\beta }{{t}^{\hat{\beta }-1}}\ln t \\ 
  \frac{\partial {{\lambda }_{i}}(t)}{\partial \lambda }= & \hat{\beta }{{t}^{\hat{\beta }-1}}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on the instantaneous failure intensity (IFI)\,\! are given by:

\begin{align}
   {IFI_{L}}= & \frac{1}{{IMTBF}_{U}} \\ 
  {IFI_{U}}= & \frac{1}{{IMTBF}_{L}}  
\end{align}\,\!

where IMTB{{F}_{L}}\,\! and IMTB{{F}_{U}}\,\! are calculated using the process presented for the confidence bounds on the instantaneous MTBF.

Time Given Cumulative Failure Intensity

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!
where:
\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial T}{\partial \beta }= & \frac{-{{\left( \tfrac{{{\lambda }_{c}}(T)}{\lambda } \right)}^{1/(\beta -1)}}\ln \left( \tfrac{{{\lambda }_{c}}(T)}{\lambda } \right)}{{{(1-\beta )}^{2}}} \\ 
  \frac{\partial T}{\partial \lambda }= & {{\left( \frac{{{\lambda }_{c}}(T)}{\lambda } \right)}^{1/(\beta -1)}}\frac{1}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on time given cumulative failure intensity (CFI)\,\! are given by:

\hat{t}={{\left( \frac{CFI}{{\hat{\lambda }}} \right)}^{\tfrac{1}{\hat{\beta }-1}}}\,\!

Then estimate the number of failures, N\,\!, such that:

N=\hat{\lambda }{{\hat{t}}^{{\hat{\beta }}}}\,\!

The lower and upper confidence bounds on time are then estimated using:

\begin{align}
   {{t}_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2\cdot CFI} \\ 
  {{t}_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N+2}^{2}}{2\cdot CFI}  
\end{align}\,\!

Time Given Cumulative MTBF

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!

where:

\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\hat{T}={{(\lambda \cdot {{m}_{c}})}^{1/(1-\beta )}}\,\!
\begin{align}
   \frac{\partial T}{\partial \beta }= & \frac{{{(\lambda \cdot \,{{m}_{c}})}^{1/(1-\beta )}}\ln (\lambda \cdot \text{ }{{m}_{c}})}{{{(1-\beta )}^{2}}} \\ 
  \frac{\partial T}{\partial \lambda }= & \frac{{{(\lambda \text{ }\cdot \text{ }{{m}_{c}})}^{1/(1-\beta )}}}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on time given cumulative MTBF (CMTBF)\,\! are estimated using the process for calculating the confidence bounds on time given cumulative failure intensity (CFI)\,\! where CFI=\frac{1}{CMTBF}\,\!.

Time Given Instantaneous MTBF

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!

where:

\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\hat{T}={{(\lambda \beta \cdot MTB{{F}_{i}})}^{1/(1-\beta )}}\,\!
\begin{align}
   \frac{\partial T}{\partial \beta }= & {{\left( \lambda \beta \cdot MTB{{F}_{i}} \right)}^{1/(1-\beta )}}\left[ \frac{1}{{{(1-\beta )}^{2}}}\ln (\lambda \beta \cdot MTB{{F}_{i}})+\frac{1}{\beta (1-\beta )} \right] \\ 
  \frac{\partial T}{\partial \lambda }= & \frac{{{(\lambda \beta \cdot MTB{{F}_{i}})}^{1/(1-\beta )}}}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

Failure Terminated

If the unbiased value \bar{\beta }\,\! is used then:

IMTBF=IMTBF\cdot \frac{N-2}{N}\,\!

where:

  • IMTBF\,\! = instantaneous MTBF.
  • N\,\! = total number of failures.

Calculate the constants p_1\,\! and p_2\,\! using procedures described for the confidence bounds on instantaneous MTBF. The lower and upper confidence bounds on time are then given by:

{{\hat{t}}_{L}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{p}_{1}}} \right)}^{\tfrac{1}{1-\beta }}}
{{\hat{t}}_{U}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{p}_{2}}} \right)}^{\tfrac{1}{1-\beta }}}

Time Terminated

If the unbiased value \bar{\beta }\,\! is used then:

IMTBF=IMTBF\cdot \frac{N-1}{N}\,\!

where:

  • IMTBF\,\! = instantaneous MTBF.
  • N\,\! = total number of failures.

Calculate the constants {{\Pi }_{1}}\,\! and {{\Pi }_{2}}\,\! using procedures described for the confidence bounds on instantaneous MTBF. The lower and upper confidence bounds on time are then given by:

{{\hat{t}}_{L}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{\Pi }_{1}}} \right)}^{\tfrac{1}{1-\beta }}}\,\!
{{\hat{t}}_{U}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{\Pi }_{2}}} \right)}^{\tfrac{1}{1-\beta }}}\,\!

Time Given Instantaneous Failure Intensity

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!

where:

\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\hat{T}={{\left( \frac{{{\lambda }_{i}}(T)}{\lambda \beta } \right)}^{1/(\beta -1)}}\,\!
\begin{align}
   \frac{\partial T}{\partial \beta }= & {{\left( \frac{{{\lambda }_{i}}(T)}{\lambda \beta } \right)}^{1/(\beta -1)}}\left[ -\frac{\ln (\tfrac{{{\lambda }_{i}}(T)}{\lambda \beta })}{{{(\beta -1)}^{2}}}+\frac{1}{\beta (1-\beta )} \right] \\ 
  \frac{\partial T}{\partial \lambda }= & {{\left( \frac{{{\lambda }_{i}}(T)}{\lambda \beta } \right)}^{1/(\beta -1)}}\frac{1}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on time given instantaneous failure intensity (IFI)\,\! are estimated using the process for calculating the confidence bounds on time given instantaneous MTBF where IMTBF=\frac{1}{IFI}\,\!.

Grouped Data

This section presents the confidence bounds for the Crow-AMSAA model when using Grouped data.

Beta (Grouped)

Fisher Matrix Bounds

The parameter \beta \,\! must be positive, thus \ln \beta \,\! is treated as being normally distributed as well.

\frac{\ln \hat{\beta }-\ln \beta }{\sqrt{Var(\ln \hat{\beta }})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds are given as:

C{{B}_{\beta }}=\hat{\beta }{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{\beta })}/\hat{\beta }}}\,\!
\hat{\beta }\,\! can be obtained by \underset{i=1}{\overset{K}{\mathop{\sum }}}\,{{n}_{i}}\left( \tfrac{T_{i}^{{\hat{\beta }}}\ln {{T}_{i}}-T_{i-1}^{{\hat{\beta }}}\ln \,{{T}_{i-1}}}{T_{i}^{{\hat{\beta }}}-T_{i-1}^{{\hat{\beta }}}}-\ln {{T}_{k}} \right)=0\,\!.

All variance can be calculated using the Fisher matrix:

\left[ \begin{matrix}
   -\tfrac{{{\partial }^{2}}\Lambda }{\partial {{\lambda }^{2}}} & -\tfrac{{{\partial }^{2}}\Lambda }{\partial \lambda \partial \beta }  \\
   -\tfrac{{{\partial }^{2}}\Lambda }{\partial \lambda \partial \beta } & -\tfrac{{{\partial }^{2}}\Lambda }{\partial {{\beta }^{2}}}  \\
\end{matrix} \right]_{\beta =\hat{\beta },\lambda =\hat{\lambda }}^{-1}=\left[ \begin{matrix}
   Var(\hat{\lambda }) & Cov(\hat{\beta },\hat{\lambda })  \\
   Cov(\hat{\beta },\hat{\lambda }) & Var(\hat{\beta })  \\
\end{matrix} \right]\,\!

\Lambda \,\! is the natural log-likelihood function where \ln^{2}T={{\left( \ln T \right)}^{2}}\,\! and:

\Lambda =\underset{i=1}{\overset{k}{\mathop \sum }}\,\left[ {{n}_{i}}\ln (\lambda T_{i}^{\beta }-\lambda T_{i-1}^{\beta })-(\lambda T_{i}^{\beta }-\lambda T_{i-1}^{\beta })-\ln {{n}_{i}}! \right]\,\!
\begin{align}
   \frac{{{\partial }^{2}}\Lambda }{\partial {{\lambda }^{2}}}= & -\frac{n}{{{\lambda }^{2}}} \\ 
  \frac{{{\partial }^{2}}\Lambda }{\partial {{\beta }^{2}}}= & \underset{i=1}{\overset{k}{\mathop \sum }}\,\left[ \begin{matrix}
   {{n}_{i}}\left( \tfrac{(T_{i}^{{\hat{\beta }}}{{\ln }^{2}}{{T}_{i}}-T_{i-1}^{{\hat{\beta }}}{{\ln }^{2}}{{T}_{i-1}})(T_{i}^{{\hat{\beta }}}-T_{i-1}^{{\hat{\beta }}})-{{\left( T_{i}^{{\hat{\beta }}}\ln {{T}_{i}}-T_{i-1}^{{\hat{\beta }}}\ln {{T}_{i-1}} \right)}^{2}}}{{{(T_{i}^{{\hat{\beta }}}-T_{i-1}^{{\hat{\beta }}})}^{2}}} \right)  \\
   -\left( \lambda T_{i}^{{\hat{\beta }}}{{\ln }^{2}}{{T}_{i}}-\lambda T_{i-1}^{{\hat{\beta }}}{{\ln }^{2}}{{T}_{i-1}} \right)  \\
\end{matrix} \right] \\ 
  \frac{{{\partial }^{2}}\Lambda }{\partial \lambda \partial \beta }= & -T_{K}^{\beta }\ln {{T}_{k}}  
\end{align}\,\!
Crow Bounds

The 2-sided confidence bounds on \hat{\beta }\,\! are given by first calculating:

P\left( i \right)=\frac{{{T}_{i}}}{{{T}_{K}}};\text{ }i=1,2,...,K

where:

  • T_i\,\! = interval end time for the {{i}^{th}}\,\! interval.
  • K\,\! = number of intervals.
  • T_K\,\! = end time for the last interval.

Next:

A=\underset{i=1}{\overset{K}{\mathop \sum }}\,\frac{{{[P{{(i)}^{{\hat{\beta }}}}\ln P{{(i)}^{{\hat{\beta }}}}-P{{(i-1)}^{\hat{\beta }}}\ln P{{(i-1)}^{{\hat{\beta }}}}]}^{2}}}{[P{{(i)}^{{\hat{\beta }}}}-P{{(i-1)}^{{\hat{\beta }}}}]}\,\!

And:

c=\frac{1}{\sqrt{A}}

Then:

S=\frac{\left( {{z}_{1-\tfrac{\alpha }{2}}} \right)\cdot c}{\sqrt{N}}

where:

  • {{z}_{1-\tfrac{\alpha }{2}}}\,\! = inverse standard normal.
  • N\,\! = number of failures.

The 2-sided confidence bounds on \beta\,\! are then \hat{\beta }\left( 1\pm S \right)\,\!.

Growth Rate (Grouped)

Since the growth rate, \alpha \,\!, is equal to 1-\beta \,\!, the confidence bounds for both the Fisher matrix and Crow methods are:

\alpha_L=1-\beta_U\,\!
\alpha_U=1-\beta_L\,\!

{{\beta }_{L}}\,\! and {{\beta }_{U}}\,\! are obtained using the methods described above in the confidence bounds on Beta.

Lambda (Grouped)

Fisher Matrix Bounds

The parameter \lambda \,\! must be positive, thus \ln \lambda \,\! is treated as being normally distributed as well. These bounds are based on:

\frac{\ln \hat{\lambda }-\ln \lambda }{\sqrt{Var(\ln \hat{\lambda }})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on \lambda \,\! are given as:

C{{B}_{\lambda }}=\hat{\lambda }{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{\lambda })}/\hat{\lambda }}}\,\!

where:

\hat{\lambda }=\frac{n}{T_{k}^{{\hat{\beta }}}}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta.

Crow Bounds

Failure Terminated

For failure terminated data, the 2-sided (1-\alpha )\,\! 100% confidence interval, the confidence bounds on \lambda \,\! are:

\begin{align}
   {{\lambda }_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2\cdot T_{k}^{\beta }} \\ 
  {{\lambda }_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N}^{2}}{2\cdot T_{k}^{\beta }}
\end{align}\,\!

where:

  • N\,\! = total number of failures.
  • T_K\,\! = end time of last interval.

Time Terminated

For time terminated data, the 2-sided (1-\alpha )\,\! 100% confidence interval, the confidence bounds on \lambda \,\! are:

\begin{align}
   {{\lambda }_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2\cdot T_{k}^{\beta }} \\ 
  {{\lambda }_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N+2}^{2}}{2\cdot T_{k}^{\beta }}  
\end{align}\,\!

where:

  • N\,\! = total number of failures.
  • T_K\,\! = end time of last interval.

Cumulative Number of Failures (Grouped)

Fisher Matrix Bounds

The cumulative number of failures, N(t)\,\!, must be positive, thus \ln N(t)\,\! is treated as being normally distributed.

\frac{\ln \hat{N}(t)-\ln N(t)}{\sqrt{Var(\ln \hat{N}(t)})}\ \tilde{\ }\ N(0,1)\,\!
N(t)=\hat{N}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{N}(t))}/\hat{N}(t)}}\,\!

where:

\hat{N}(t)=\hat{\lambda }{{t}^{{\hat{\beta }}}}\,\!
\begin{align}
   Var(\hat{N}(t))= & {{\left( \frac{\partial \hat{N}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial \hat{N}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial \hat{N}(t)}{\partial \beta } \right)\left( \frac{\partial \hat{N}(t)}{\partial \lambda } \right)cov(\hat{\beta },\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial \hat{N}(t)}{\partial \beta }= & \hat{\lambda }{{t}^{{\hat{\beta }}}}\ln t \\ 
  \frac{\partial \hat{N}(t)}{\partial \lambda }= & {{t}^{{\hat{\beta }}}}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on the cumulative number of failures are given by:

N{{(t)}_{L}}=\frac{t}{{\hat{\beta }}}IF{{I}_{L}}\,\!
N{{(t)}_{U}}=\frac{t}{{\hat{\beta }}}IF{{I}_{U}}\,\!

where IFI_L\,\! and IFI_U\,\! are calculated based on the procedures for the confidence bounds on the instantaneous failure intensity.

Cumulative Failure Intensity (Grouped)

Fisher Matrix Bounds

The cumulative failure intensity, {{\lambda }_{c}}(t)\,\!, must be positive, thus \ln {{\lambda }_{c}}(t)\,\! is treated as being normally distributed.

\frac{\ln {{{\hat{\lambda }}}_{c}}(t)-\ln {{\lambda }_{c}}(t)}{\sqrt{Var(\ln {{{\hat{\lambda }}}_{c}}(t)})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on the cumulative failure intensity are then estimated from:

CB={{\hat{\lambda }}_{c}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{\lambda }}}_{c}}(t))}/{{{\hat{\lambda }}}_{c}}(t)}}\,\!

where:

{{\hat{\lambda }}_{c}}(t)=\hat{\lambda }{{t}^{\hat{\beta }-1}}\,\!

and:

\begin{align}
   Var({{{\hat{\lambda }}}_{c}}(t))= & {{\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \beta } \right)\left( \frac{\partial {{\lambda }_{c}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{\lambda }_{c}}(t)}{\partial \beta }= & \hat{\lambda }{{t}^{\hat{\beta }-1}}\ln t \\ 
  \frac{\partial {{\lambda }_{c}}(t)}{\partial \lambda }= & {{t}^{\hat{\beta }-1}}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on the cumulative failure intensity (CFI\,\!) are given below. Let:

N=\hat{\lambda }{{t}^{{\hat{\beta }}}}

Then:

\begin{align}
CFI_{L}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2\cdot t} \\ 
CFI_{U}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N+2}^{2}}{2\cdot t}  
\end{align}\,\!

Cumulative MTBF (Grouped)

Fisher Matrix Bounds

The cumulative MTBF, {{m}_{c}}(t)\,\!, must be positive, thus \ln {{m}_{c}}(t)\,\! is treated as being normally distributed as well.

\frac{\ln {{{\hat{m}}}_{c}}(t)-\ln {{m}_{c}}(t)}{\sqrt{Var(\ln {{{\hat{m}}}_{c}}(t)})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on the cumulative MTBF are then estimated from:

CB={{\hat{m}}_{c}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{m}}}_{c}}(t))}/{{{\hat{m}}}_{c}}(t)}}\,\!

where:

{{\hat{m}}_{c}}(t)=\frac{1}{{\hat{\lambda }}}{{t}^{1-\hat{\beta }}}\,\!
\begin{align}
   Var({{{\hat{m}}}_{c}}(t))= & {{\left( \frac{\partial {{m}_{c}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{m}_{c}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{m}_{c}}(t)}{\partial \beta } \right)\left( \frac{\partial {{m}_{c}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\hat{\lambda })\,  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{m}_{c}}(t)}{\partial \beta }= & -\frac{1}{{\hat{\lambda }}}{{t}^{1-\hat{\beta }}}\ln t \\ 
  \frac{\partial {{m}_{c}}(t)}{\partial \lambda }= & -\frac{1}{{{{\hat{\lambda }}}^{2}}}{{t}^{1-\hat{\beta }}}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on cumulative MTBF (CMTBF)\,\! are given by:

CMTB{{F}_{L}}=\frac{1}{CF{{I}_{U}}}\,\!
CMTB{{F}_{U}}=\frac{1}{CF{{I}_{L}}}\,\!

where CFI_{L}\,\! and CFI_{U}\,\! are calculating using the process for calculating the confidence bounds on the cumulative failure intensity.

Instantaneous MTBF (Grouped)

Fisher Matrix Bounds

The instantaneous MTBF, {{m}_{i}}(t)\,\!, must be positive, thus \ln {{m}_{i}}(t)\,\! is approximately treated as being normally distributed as well.

\frac{\ln {{{\hat{m}}}_{i}}(t)-\ln {{m}_{i}}(t)}{\sqrt{Var(\ln {{{\hat{m}}}_{i}}(t)})}\ \tilde{\ }\ N(0,1)\,\!

The approximate confidence bounds on the instantaneous MTBF are then estimated from:

CB={{\hat{m}}_{i}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{m}}}_{i}}(t))}/{{{\hat{m}}}_{i}}(t)}}\,\!

where:

{{\hat{m}}_{i}}(t)=\frac{1}{\lambda \beta {{t}^{\beta -1}}}\,\!
\begin{align}
   Var({{{\hat{m}}}_{i}}(t))= & {{\left( \frac{\partial {{m}_{i}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{m}_{i}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{m}_{i}}(t)}{\partial \beta } \right)\left( \frac{\partial {{m}_{i}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{m}_{i}}(t)}{\partial \beta }= & -\frac{1}{\hat{\lambda }{{{\hat{\beta }}}^{2}}}{{t}^{1-\hat{\beta }}}-\frac{1}{\hat{\lambda }\hat{\beta }}{{t}^{1-\hat{\beta }}}\ln t \\ 
  \frac{\partial {{m}_{i}}(t)}{\partial \lambda }= & -\frac{1}{{{{\hat{\lambda }}}^{2}}\hat{\beta }}{{t}^{1-\hat{\beta }}}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on instantaneous MTBF (IMTBF)\,\! are given by first calculating:

P\left( i \right)=\frac{{{T}_{i}}}{{{T}_{K}}};\text{ }i=1,2,...,K

where:

  • T_i\,\! = interval end time for the {{i}^{th}}\,\! interval.
  • K\,\! = number of intervals.
  • T_K\,\! = end time for the last interval.

Calculate:

A=\underset{i=1}{\overset{K}{\mathop \sum }}\,\frac{{{\left[ P{{(i)}^{{\hat{\beta }}}}\ln P{{(i)}^{{\hat{\beta }}}}-P{{(i-1)}^{\hat{\beta }}}\ln P{{(i-1)}^{{\hat{\beta }}}} \right]}^{2}}}{\left[ P{{(i)}^{{\hat{\beta }}}}-P{{(i-1)}^{{\hat{\beta }}}} \right]}\,\!

Next:

D=\sqrt{\frac{1}{A}+1}

And:

W=\frac{\left( {{z}_{1-\tfrac{\alpha }{2}}} \right)\cdot D}{\sqrt{N}}

where:

  • {{z}_{1-\tfrac{\alpha }{2}}}\,\! = inverse standard normal.
  • N\,\! = number of failures.

The 2-sided confidence bounds on instantaneous MTBF are then IMTBF\left( 1\pm W \right)\,\!.

Instantaneous Failure Intensity (Grouped)

Fisher Matrix Bounds

The instantaneous failure intensity, {{\lambda }_{i}}(t)\,\!, must be positive, thus \ln {{\lambda }_{i}}(t)\,\! is treated as being normally distributed.

\frac{\ln {{{\hat{\lambda }}}_{i}}(t)-\ln {{\lambda }_{i}}(t)}{\sqrt{Var(\ln {{{\hat{\lambda }}}_{i}}(t)})}\tilde{\ }N(0,1)\,\!

The approximate confidence bounds on the instantaneous failure intensity are then estimated from:

CB={{\hat{\lambda }}_{i}}(t){{e}^{\pm {{z}_{\alpha }}\sqrt{Var({{{\hat{\lambda }}}_{i}}(t))}/{{{\hat{\lambda }}}_{i}}(t)}}\,\!

where {{\lambda }_{i}}(t)=\lambda \beta {{t}^{\beta -1}}\,\! and:

\begin{align}
   Var({{{\hat{\lambda }}}_{i}}(t))= & {{\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \beta } \right)\left( \frac{\partial {{\lambda }_{i}}(t)}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial {{\lambda }_{i}}(t)}{\partial \beta }= & \hat{\lambda }{{t}^{\hat{\beta }-1}}+\hat{\lambda }\hat{\beta }{{t}^{\hat{\beta }-1}}\ln t \\ 
  \frac{\partial {{\lambda }_{i}}(t)}{\partial \lambda }= & \hat{\beta }{{t}^{\hat{\beta }-1}}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on the instantaneous failure intensity (IFI)\,\! are given by:

\begin{align}
  IF{{I}_{U}}= & \frac{1}{IMTB{{F}_{L}}} \\
  IF{{I}_{L}}= & \frac{1}{IMTB{{F}_{U}}}
\end{align}\,\!

where IMTB{{F}_{L}}\,\!and IMTB{{F}_{U}}\,\! are calculated using the process for calculating the confidence bounds on the instantaneous MTBF.

Time Given Cumulative Failure Intensity (Grouped)

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!

where:

\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\begin{align}
   \frac{\partial T}{\partial \beta }= & \frac{-{{\left( \tfrac{{{\lambda }_{c}}(T)}{\lambda } \right)}^{1/(\beta -1)}}\ln \left( \tfrac{{{\lambda }_{c}}(T)}{\lambda } \right)}{{{(1-\beta )}^{2}}} \\ 
  \frac{\partial T}{\partial \lambda }= & {{\left( \frac{{{\lambda }_{c}}(T)}{\lambda } \right)}^{1/(\beta -1)}}\frac{1}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on time given cumulative failure intensity (CFI)\,\! are presented below. Let:

\hat{t}={{\left( \frac{CFI}{{\hat{\lambda }}} \right)}^{\tfrac{1}{\hat{\beta }-1}}}\,\!

Then estimate the number of failures:

N=\hat{\lambda }{{\hat{T}}^{{\hat{\beta }}}}\,\!

The confidence bounds on time given the cumulative failure intensity are then given by:

\begin{align}
   {{t}_{L}}= & \frac{\chi _{\tfrac{\alpha }{2},2N}^{2}}{2\cdot {CFI}} \\ 
  {{t}_{U}}= & \frac{\chi _{1-\tfrac{\alpha }{2},2N+2}^{2}}{2\cdot {CFI}}  
\end{align}\,\!

Time Given Cumulative MTBF (Grouped)

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!

where:

\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\hat{T}={{(\lambda \cdot {{m}_{c}})}^{1/(1-\beta )}}\,\!
\begin{align}
   \frac{\partial T}{\partial \beta }= & \frac{{{(\lambda \cdot \,{{m}_{c}})}^{1/(1-\beta )}}\ln (\lambda \cdot \text{ }{{m}_{c}})}{{{(1-\beta )}^{2}}} \\ 
  \frac{\partial T}{\partial \lambda }= & \frac{{{(\lambda \cdot {{m}_{c}})}^{1/(1-\beta )}}}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on time given cumulative MTBF (CMTBF)\,\! are estimated using the process for calculating the confidence bounds on time given cumulative failure intensity (CFI)\,\! where CFI=\frac{1}{CMTBF}\,\!.

Time Given Instantaneous MTBF (Grouped)

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!

where:

\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\hat{T}={{(\lambda \beta \cdot {{m}_{i}}(T))}^{1/(1-\beta )}}\,\!
\begin{align}
   \frac{\partial T}{\partial \beta }= & {{\left( \lambda \beta \cdot \text{ }{{m}_{i}}(T) \right)}^{1/(1-\beta )}}\left[ \frac{1}{{{(1-\beta )}^{2}}}\ln (\lambda \beta \cdot {{m}_{i}}(T))+\frac{1}{\beta (1-\beta )} \right] \\ 
  \frac{\partial T}{\partial \lambda }= & \frac{{{(\lambda \beta \cdot \text{ }{{m}_{i}}(T))}^{1/(1-\beta )}}}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

Failure Terminated

Calculate the constants p_1\,\! and p_2\,\! using procedures described for the confidence bounds on instantaneous MTBF. The lower and upper confidence bounds on time are then given by:

{{\hat{t}}_{L}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{p}_{1}}} \right)}^{\tfrac{1}{1-\beta }}}
{{\hat{t}}_{U}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{p}_{2}}} \right)}^{\tfrac{1}{1-\beta }}}

Time Terminated

Calculate the constants {{\Pi }_{1}}\,\! and {{\Pi }_{2}}\,\! using procedures described for the confidence bounds on instantaneous MTBF. The lower and upper confidence bounds on time are then given by:

{{\hat{t}}_{L}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{\Pi }_{1}}} \right)}^{\tfrac{1}{1-\beta }}}\,\!
{{\hat{t}}_{U}}={{\left( \frac{\lambda \beta \cdot IMTBF}{{{\Pi }_{2}}} \right)}^{\tfrac{1}{1-\beta }}}\,\!

Time Given Instantaneous Failure Intensity (Grouped)

Fisher Matrix Bounds

The time, T\,\!, must be positive, thus \ln T\,\! is treated as being normally distributed.

\frac{\ln \hat{T}-\ln T}{\sqrt{Var(\ln \hat{T}})}\ \tilde{\ }\ N(0,1)\,\!

Confidence bounds on the time are given by:

CB=\hat{T}{{e}^{\pm {{z}_{\alpha }}\sqrt{Var(\hat{T})}/\hat{T}}}\,\!

where:

\begin{align}
   Var(\hat{T})= & {{\left( \frac{\partial T}{\partial \beta } \right)}^{2}}Var(\hat{\beta })+{{\left( \frac{\partial T}{\partial \lambda } \right)}^{2}}Var(\hat{\lambda }) \\ 
   & +2\left( \frac{\partial T}{\partial \beta } \right)\left( \frac{\partial T}{\partial \lambda } \right)cov(\hat{\beta },\,\,\,\hat{\lambda })  
\end{align}\,\!

The variance calculation is the same as given above in the confidence bounds on Beta. And:

\hat{T}={{\left( \frac{{{\lambda }_{i}}(T)}{\lambda \beta } \right)}^{1/(\beta -1)}}\,\!
\begin{align}
   \frac{\partial T}{\partial \beta }= & {{\left( \frac{{{\lambda }_{i}}(T)}{\lambda \beta } \right)}^{1/(\beta -1)}}\left[ -\frac{\ln (\tfrac{{{\lambda }_{i}}(T)}{\lambda \beta })}{{{(\beta -1)}^{2}}}+\frac{1}{\beta (1-\beta )} \right] \\ 
  \frac{\partial T}{\partial \lambda }= & {{\left( \frac{{{\lambda }_{i}}(T)}{\lambda \beta } \right)}^{1/(\beta -1)}}\frac{1}{\lambda (1-\beta )}  
\end{align}\,\!

Crow Bounds

The 2-sided confidence bounds on time given instantaneous failure intensity (IFI)\,\! are estimated using the process for calculating the confidence bounds on time given instantaneous MTBF where IMTBF=\frac{1}{IFI}\,\!.

Personal tools
ReliaWiki.org
Create a book