MATHEMATICAL MODELS FOR DRIFT FAILURE ANALYSIS,
POLYTECHNIC INST OF BROOKLYN N Y DEPT OF ELECTRICAL ENGINEERING
Pagination or Media Count:
Several mathematical techniques which are available for the analysis of drift failures are discussed. The conventional techniques of drift failure analysis are stacking-up-the-tolerance, worst case analysis, and expansion of the performance function in a Taylor series which is truncated after the linear terms. These techniques are simple to apply but are rather inaccurate in many cases. The very general approach involving the change of random variables theorem is discussed however, the resulting integral expressions are only amenable to solution and interpretation in special cases. The specific moment relationships are computed for Gaussian and Rectangular distribution models. A discussion of how one can estimate parameter moments from manufacturers specifications is included. The probabilistic expressions including the quadratic terms are derived. The partial derivative coefficients in the Taylor series are related to sensitivity functions used in system theory and network theory. A probabilistic interpretation is given to slacking up the tolerances and worst case analysis.