Revised GMDH Algorithm Using Quantified Input Variables with the Application to Rolling Model Identification

Abstract

In this paper, a revised GMDH (Group Method of Data Handling) algorithm, which can identify the model by using both qualitative and quantitative input variables, is developed. In the previous GMDH algorithms, the qualitative input variables can not be included in the GMDH model and so the useful qualitative information of the system can not be used in the model identification. In the large scale systems whose structures are very complex, there are many cases in which the qualitative system characters have significant effects on the model structures. The revised GMDH algorithm in this paper quantifies the qualitative input variables for each item by using the quantification theory, and generates the quantified input variables in the first layer. The GMDH model is identified by combining the quantified input variables and the quantitative input variables. The revised GMDH algorithm is applied to the identification problem of rolling model and the combined physical and statistical model obtained by the revised GMDH algorithm is compared with the model obtained by the previous GMDH algorithm.