Related Topics:

Monte Carlo and SimuMatic

Weibull++ Standard Folio Control Panel

ALTA SimuMatic Example

Weibull++ SimuMatic Example

This example uses Weibull++ SimuMatic to compare different parameter estimation methods for a sample of 10 units following a Weibull distribution where beta = 2 and eta = 100, with all units tested to failure.

Follow the steps outlined below to see the results of an analysis that uses the maximum likelihood (MLE) parameter estimation method.

To compare RRX to the rank regression on Y (RRY) and maximum likelihood estimation (MLE) methods, repeat the steps above, but choose a different option in the Analysis Method drop-down list on the Analysis tab. Then compare the three plots.

The plots for the three methods are shown below:


RRX Method

 


RRY Method

 


MLE Method

The true parameter line (black) shows the time vs. unreliability defined by the parameters of the distribution that was used to generated the simulated data. The median line (yellow) shows the time vs. unreliability defined by the median parameters of all generated data sets. By comparing these two lines, you can estimate how close the estimated parameters would be to the product's assumed parameters (which you provided in the setup). The plots above show that RRX would provide the least deviation (i.e., "bias") from the truth for this sample size and data type.

The lines marking the lower and upper confidence bounds on reliability (pink) allow you to determine how closely grouped together the parameter estimation results are. The plots above show that MLE would provide results with less variation than any of the other analysis methods for this sample size and data type.

Thus given our relatively small sample size and uncensored data, RRX is shown to be best in terms of bias, while MLE is shown to be best in terms of variation (the median line and true parameter line overlap almost perfectly).

 

© 1992-2016. ReliaSoft Corporation. ALL RIGHTS RESERVED.

E-mail Link