The demands on parameter technologies in Sinking EDM are very high. In each processing step, a certain surface roughness has to be achieved in very precise manner. At the same time removal rate (productivity) and relative wear (costs) have to achieve advantageous values. The generation of these parameter technologies is time-consuming and expensive, since many influencing factors affect the target figures. The procedural methods in practice are often unsystematic and ineffective. The technological possibilities of machine and process configuration are very often not used. This publication deals with methods designed to effectively optimize the target figures. At the same time, effort and time spent for generating parameter technologies are reduced. Continuous parameter technologies are used for the generation of discrete parameter technologies. Two different types of continuous mappings are compared: Nonlinear regression functions and artificial neural networks (ANN). Model accuracy and the handling properties of these variants are evaluated in this context. A variety of methods for assessing the adequacy of mappings are discussed. A new, problem-oriented experimental design method for EDM-processes is also described, focusing on two aspects: a reduction of the experimental effort and the achievement of a sufficient accuracy of the continuous mappings at the same time. In order to achieve these characteristics, different scale methods and transformation methods are discussed, as well. As a result, a problem-adapted CAP-software solution is presented. Nonlinear regression functions as well as ANNs for the continuous parameter technologies can be generated, visualized and evaluated within this software. In a next step, software functions are implemented, allowing the technologist to derive optimal pulse parameter series for processing. The application of the presented methods is first of all useful for manufacturers of EDM die sinking machines.

This content is only available via PDF.
You do not currently have access to this content.