基于规则的建模方法的可解释性及其发展

建模方法的可解释性指其以可理解的方式表达实际系统行为的能力. 随着实践中对可靠性需求的不断提高, 建立出可靠且可解释的模型以增强人对实际系统的认知成为了建模的重要目标. 基于规则的建模方法可更直观地描述系统机理, 并能有效融合定量信息和定性知识实现不确定信息的灵活处理, 具有较强的建模性能. 本文从基于规则的建模方法出发, 围绕知识库、推理机和模型优化梳理了其在可解释性方面的研究, 最后进行了简要的评述和展望.

The Interpretability of Rule-based Modeling Approach and Its Development

The model interpretability refers to the ability to express the real system behavior in an understandable way. With the increasing of reliability requirements in engineering practice, establishing a reliable and interpretable model to enhance human understanding of real systems has become one of the main objectives. Rule-based modeling approach can describe the system mechanism more intuitively. It can not only effectively integrate quantitative information and qualitative knowledge, but can also deal with uncertain information flexibly. This paper combs researches on the interpretability of rule-based modeling approach around the knowledge base, inference engine and model optimization, and finally makes a brief review and prospect.