Sequential algorithm based on number theoretic method for statistical tolerance analysis and synthesis

Tolerancing is one of the most important tasks in product and manufacturing process design. In the literature, both Monte Carlo simulation and numeric al optimization method have been widely used in the process of statistical tolerance analysis and synthesis, but the computational effort is huge. Thi s paper presents two techniques, quasi random numbers based on the Number T heoretic Method and sequential algorithm based on the Number Theoretic net, to calculate yield and to perform tolerance allocation. An example demonst rates the optimal tolerance allocation design and is employed to investigat e the efficiency and accuracy of this solution. This algorithm can efficien tly obtain the global optimum, and the amount of calculation is considerabl y reduced.