Cell Mapping Modeling for Complex Systems Based on ISODATA Algorithm

Abstract

This paper presents a new type of modeling framework for complex dynamical systems based on cell mapping. The systems considered are difficult in modeling by physical principle due to complex mechanism. By the method of data driven, the cell mapping model is constructed and the modeling objective is describing statistical dynamic trajectory for this kind of complex system, which overcome the difficulties found in mechanism modeling. Firstly, the clustering of operation modes is achieved by Self-Organizing Data Analysis Techniques Algorithm (ISODATA). Then, a cell division measure is obtained by the clustering result, which is different from the regular partition in other measures. Meanwhile, each class bound is treated as a cell bound, and hence the dynamic trajectory of complex systems is described by the cell mapping. Finally, a simulated experience illustrates the feasibility of the proposed approach.