关键词: 误差特征分析, 模型验证, 信息熵, 灰色关联分析, 仿真

Abstract: Aiming at the problems existed in traditional model validation methods, a methodology of grey synthesis evaluation of model credibility based on error feature analysis was proposed. Error feature indicators were analyzed and multi-indicator error feature matrix was established. Furthermore, grey synthesis evaluation method based on entropy weight and grey relational analysis was discussed in detail. Finally, the proposed method was applied to model validation. Application result showed that this method could reduce risks faced by traditional error analysis effectively, which were caused by using only on indicator to assess model credibility.

Key words: error feature analysis, model validation, information entropy, grey relational analysis, simulation