Statistical considerations in the estimation of electrical contact reliability

Abstract

Summary form only given. The established definition of reliability involves four key elements: probability, satisfactory performance, operating life, and operating conditions. Reliability of existing systems can be estimated using field data when they are available in suitable form and quantity, or using data from (often accelerated) laboratory tests. Test data are emphasized here. Such data take one of three possible forms: attribute data, data on time to failure for a sample of items, or variables data. The three forms of data differ markedly with respect to the underlying statistical assumptions required for validity of their analysis, with respect to the amount and type of information they can reveal, and with respect to the sample sizes required to achieve precise estimates and convincing statistical confidence.<>