SYSTEM RELIABILITY PREDICTION BY FUNCTION. VOLUME 1. DEVELOPMENT OF PREDICTION TECHNIQUES.
Technical documentary rept., 27 July-27 Apr 63.
ARINC RESEARCH CORP WASHINGTON D C
Pagination or Media Count:
A technique was developed for predicting the reliability of ground electronic systems prior to the actual design of the systems. Reliabil ity data from 51 systems representing over two million system operating hours were analyzed to develop prediction equations based on genera lized system parameters such as complexity, power consumption, and frequency. The prediction equations are of the form Ln theta bo blX1 b2 X2 ... bnXn, where theta is the predicted system mean life and the Xi are measures of equipment characteristics. The coefficients bi were determined by multiple regression analysis. Approximately 40 possible characteristics were investigated. The technique also allows the estimation of confidence intervals for Ln theta. With the use of new data, the prediction tech nique can be updated at any time by the computa tion of new regression equations. The required program 1401 computer, FORTRAN is presented in an appendix. Te step-by-step procedure is pre sented in Volume II of this report. Author