Monte Carlo Technique Suitable for Obtaining Complex Space System Reliability Confidence Limits from Component Test Data with Three Unknown Parameters

摘要

A Double Monte Carlo method of obtaining confidence limits for complex systems based on component failure data assuming a three parameter Weibull distribution was developed. Three new parameter estimation routines were developed and compared with the Harter-Moore three parameter maximum likelihood routine for use with the Monte Carlo method. The sensitivity of the method to system reliability, sample size, and number of points in the component reliability distributions was assessed. An approximate method of calculating and correcting for parameter estimation bias was developed and illustrated. The Double Monte Carlo method appears to be effective at system reliabilities from 74% to 96% with component failure sample sizes as well as five with the Linear Least Squares parameter estimation routine developed. (Author)