The multimodel approach offers a very satisfactory results in modelling, diagnose and control of complex systems. In the modelling case, this approach passes by three steps: the determination of the model's library, the validities computation and the establishment of the final model. In this context, this paper focuses on the elaboration of a comparative study between three recent methods of validities computation. Thus, it highlight the method that offers the best performances in term of precision. To achieve this goal, we apply, these three methods on two simulation examples in order to compare their performances. 882 the the models'base [14]. Besides, these methods are limited and cannot be useful with presence of complex and/or ill-defined system. To overcome these problems, Ben Messaoud and all. Have proposed recently, three new methods of computation validities. In fact, the first method is based on a local criterion optimization LCO [15]. The second method uses a hierarchical structuring HS to compute validities [16]. In the third method, validities based on optimal computation OC, are obtained by optimizing a constrained least squares problem [6]. In the present paper, we focus our study to compare the last three methods and to determine the one leads to the best model that can describe perferctly the real process. In order to compare performances assumed by the three proposed methods of validities computation, we consider two numerical simulation examples. In the second part of the paper, we present the general structure of the multimodal approach. The third section discusses the three considered calculation validities methods. Two examples of numerical simulation are given in section four. We finish this work by a conclusion.
GENERAL STRUCTURE OF MULTIMODEL APPROACHThe general structure of the multimodel approach is presented in Figure 1. It is formed by three units: the models library unit, the decision unit and the output unit [12,17,18].
Models library unitThis unit can contain different types of models (either input-output models or state space models). These models are distinguished with respect to four features: Partition strategy, submodel structural identification, transition between models and method of realization [19][20][21].