Evolving Fuzzy-Model-Based Design of Experiments With Supervised Hierarchical Clustering

This paper presents a new approach to design of experiments (DoE), based on an evolving fuzzy model structure and a supervised hierarchical clustering algorithm. DoE is the field that deals with the problem of how to design the most optimal and economic experimentation. The goal is to identify a highly nonlinear and possibly high-dimensional system, together with the minimal experimental effort required. The theory is well developed for linear and polynomial models; however, they are often not suitable for general use. For this reason, a fuzzy model in the form of Takagi-Sugeno (T-S) is used, because it has the properties of a universal approximator. The method works iteratively by sampling the system in the input domain and evolving the fuzzy model. The method is demonstrated with a simulation, which shows the potential of the proposed approach.