Intelligent Methods in Design of Experiments (DoE)

摘要

The goal of "design of experiments" (DoE) is to obtain information about a product or a process in order to optimize the significant parameters subsequently affecting one or more response variables. The levels of several assignable causes are modified systematically and simultaneously in order to keep the number of necessary experiments low. Without appropriate knowledge of the characteristics of the experimental design, it is not possible for a worker ("DoE user") responsible for a machine or a process to plan the experiments in a meaningful way. To be able to interpret the results of experiments, complex calculations are necessary. Nowadays, the calculations are usually implemented using software. For better comprehensibility, diagrams are used which provide information to the DoE expert about the connections between the response variables and assignable causes. However, these diagrams are not understandable to the DoE user without appropriate mathematical and statistical knowledge. It is necessary to supply user-friendly tools for all phases of DoE - systems analysis, design, strategy, realization, evaluation, check-up. The DoE user must be given the means to plan, realize, evaluate, and document experiments in such a way that he can use the knowledge obtained beneficially to improve his processes and the planning of further experiments.