A Comparison of the Liberman-Ross and Mann-Grubbs Methods

Abstract

This paper compares the Lieberman-Ross (LR) method, a statistically exact procedure for computing system reliability bounds, with the Mann-Grubbs (MG) procedure, an approximately optimum method for computing such bounds. For systems with exponentially distributed failure times, it is shown that the MG procedure, using the same data as the LR procedure, will always compute a lower bound than the LR bound. Simulation methods are used to infer that only when failure data are ordered so that a significant portion of the data is not incorporated by the LR procedure, will the LR procedure have a reasonable expectation of producing a superior bound to the MG bound. This is interpreted as dictating a data order that discards failure data on the least reliable component samples.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1976
Accession Number
ADA032655

Entities

People

  • Arthur L. Schoenstadt

Organizations

  • Naval Postgraduate School

Tags

DTIC Thesaurus Topics

  • California
  • Classification
  • Complex Systems
  • Computations
  • Confidence Limits
  • Iterations
  • Mathematical Analysis
  • Mathematics
  • Military Research
  • Normal Distribution
  • Probability
  • Reliability
  • Security
  • Sequences
  • Simulations
  • Software Testing

Readers

  • Linear Algebra
  • Regression Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms
  • AI & ML - Neural Networks