Resumen de Effectiveness of system identification for complex systems by the fuzzy adaptative GMDH

Tomonori Nishikawa, Shizue Shimizu, Masufumi Imai

• This paper discusses and verifies effectiveness about the Fuzzy Adaptive GMDH applied the concept of the fuzzy set to the identification process of the adaptive GMDH (Group Method of Data Handling) as the modeling a complex system. The adaptive GMDH [1] is enhancing ability of basic GMDH [2] and is a system identification method by which "Ridge bias parameter k" is introduced to avoid the multicollinearity to the GMDH, and its characteristics is improved [3,4,5]. Furthermore, when we apply this Fuzzy Adaptive GMDH to the time series data, the forecast error of the time series data has been absor-bed to the changing control limit of between upper and lower limits automatically.

  The adaptive GMDH developed as a modeling technique of a complex sys-tem, whose interrelations is nonlinear and more indefinite system to with respect to the factors, which fulfills its function for complex system iden-tification. In addition, it can be said that making the technique the Fuzzy Adap-tive GMDH will effectively function to the identification of a complex system.