Sequential Algorithm Based on Number Theoretic Method for Statistical Tolerance Analysis and Synthesis

Abstract

Tolerancing is one of the most important tasks in product and manufacturing process design. In the literature, both Monte Carlo simulation and numerical optimization method have been widely used in the process of statistical tolerance analysis and synthesis, but the computational effort is huge. This paper presents two techniques, quasi random numbers based on the Number Theoretic Method and sequential algorithm based on the Number Theoretic net, to calculate yield and to perform tolerance allocation. An example demonstrates the optimal tolerance allocation design and is employed to investigate the efficiency and accuracy of this solution. This algorithm can efficiently obtain the global optimum, and the amount of calculation is considerably reduced.