Downloaded from https://www.everyspec.com A-aK 706-157 ) of X. (It is presumed that the integal con- 2. The failure rate is a constant, and thus Vele absolutely; if not, a textbook ought to the distribution is very tractale. be consulted.) The most popular distribution for mate- The nth moment, about the mean, of X is ral properties, device parameter, and gener- the s-expected value of (X - I: alized "streses", "potentials", and "currents" is the s-normal distribution. There are two F,{(X - On}"J (x - P)" fx)dx(3-10) reasons for its popuity. where p a B {X 1 1.
The distribution fits many data with- out doing too much violence to an engineer- Let X and Y be random variables ing concept of goodness-of-fit. f(x) pdf{X} 2. The distribution is so tractable, has no g(y) pdf{Y} parameters for the basic distribution, and con- h(xvy) pdf(X,Y} volves into itself. A E -{X) Most distributions can be transformed in- ,- E{y} to something that looks different by a linear then Var{X} = -((x - A)? } tansformation of the variable. Custom, more and (3-11) than anything else, determines what the standard form is. If a linear transformation X Cov{X,Y=aE{(x-p.)(y-P,)I = aU + b is applied to a distribution, the =ff (x - p.
)(y -* p, )h(xy)dxdy mean and variance are transformed as follows: jXJY (3-12) E{X}=aE{U}+ b (3-14a) The linear-correlation coefficient is defined as Var{X}= a 2 Var{U) (3.14b) P Cov (x3 There are usually several ways of writing the [Var{X)Var{YP] (3-13) parameters of a distribution, e.g., a scale parameter can be used in the form Xx or xla (where x is the random variable and a,A are 3-4.2 DISTRIBUTIONS AND THEIR PROP- parameters). The forms in Table 31 are cho- ERTIES sen to be useful to reliability engineers. The most popular distribution for time- to-failure or time-between-failures is the expo- nential. There are two reasons for this popu- REFERENCE larity. 1.
The distribution fits many data with- 1. W. G. Ireson, Ed., Reliability Handbook, out doing too much violence to an engineer- McGraw-Hill Book Comrany, Inc., ing concept of goodness-of-fit. N.Y. 1966. 3-5