A NEW STRATEGY FOR INSPECTING ROUNDNESS FEATURES

This article addresses the problem of inspection of roundness error (in contrast to the measurement of roundness error) and the idea of GO-NOGO inspection software. A new approach is presented in which the formalism of linear programming is applied through reexpressing the initial geometrical problem and adding further constraints. The ultimate position is to provide a set of constraints from which the required inspection information can be obtained by examining only the feasibility of the mathematical program, with no need to consider its iteration toward optimality. Feasibility may be explored using an existing, although not widely known, algorithm to guide the GO-NOGO decision. The basis of the algorithm is given along with practical examples of how it works. Tests indicate that on real data it usually involves a computational effort only modestly higher than using an exchange algorithm to find an optimal fit. Thus, it appears quite practical for applications to the relatively small computer of a standard instrument