Parameter optimization for random generation of non-ideal surfaces using dual-tree complex wavelet transform.

Tolerancing is important task at design stage to ensure products quality. It is predicted that tolerancing method will change in accordance with the development of computer technology in recent years. Tolerance analysis using skin model shapes, which is parts model with form defects, will be one of the new method. If skin model shapes are generated randomly with both tool and cutter marks caused in machining processes, both designing and manufacturing parameters will be determined rationally based on results of the new tolerance analysis. For this purpose, an appropriate method to generate skin model shapes are required. In previous study, we have proposed a generation method using dual-tree complex wavelet transform to generate skin model shape randomly. The method has three parameters to be tuned for any surfaces, which has characteristics such as both tool mark and cutter mark. To generate more realistic skin model shapes, the parameters should be optimized corresponding to any surfaces. In this paper, the optimization method is developed using genetic algorithms. [ABSTRACT FROM AUTHOR]