Two-stage optimization of lens grinding parameters for multi-quality target combining Taguchi method and neural network software

Abstract

In this paper we present an efficient two-stage method combining the merits of the Taguchi method and neural network software to achieve nonlinear fine optimal lens grinding parameters for both the roughness and the curvature deviation robust over a wide range of lens refraction power. Discrete and rough optimal grinding parameters for roughness and for curvature deviation are first obtained respectively using the Taguchi method with an L18 orthogonal array. Then all the experimental data of the 18 experiments are used as input training data for neural network software to obtain a set of compromised nonlinear accurate optimal parameters for the roughness and the curvature deviation. Results of confirmation experiments using these final parameters show that lens surfaces ground with polishers ranging in curvature from −7:00 to +7.00D are robust in desired quality targets.