Validation and selection of simulation model based on the feature differences

To validate alternative models and select the most credible one when the models have many outputs of different kinds, a validation and selection method of simulation model based on feature differences is proposed. The outputs are divided into three kinds: static data, gradual data and fast data, and the measure models of feature differences for each kind data are given. The correlation among the feature differences is eliminated via principal component analysis, and several independent principal components are gained. Furthermore, the outputs of simulation models are divided into K kinds of clusters based on K-means clustering according to the principal components. Which cluster the output of actual system belongs to is judged by Fisher discriminant analysis. So the validation and selection of alternative models are realized. Finally, the method is validated in an application.