地震灾害应对决策的模型集成规则挖掘

Model integration rule mining in earthquake disaster response decision

面向地震灾害所具有的复杂性、突发性、不确定性等特点,针对信息不完备下地震灾害评估、救助与恢复重建决策支撑系统的综合决策需求,本文提出了基于数据依赖关系的、由模型多维关联度测算和设计结构矩阵模型排序构成的模型集成规则挖掘方法。通过模型之间输入输出关系五个维度,即内容概念、空间范围、时间范围、空间精度、时间粒度的关联度计算,构建模型自动排序的设计结构矩阵,实现多模型的智能选择与集成组配。并以汶川M s8.0地震为例开展不同决策阶段情景下规则挖掘的实证研究,为灾害应对决策过程模型集成提供基础,提高了应对重大自然灾害的辅助决策能力。

Faced with the characteristics of complexity,suddenness,and uncertainty of earthquake disasters,and for the comprehensive decision-making needs of the decision support system for earthquake disaster assessment,rescue and restoration under incomplete information,this paper proposes a model integration rule mining method based on data dependency.This method is composed of the multidimensional correlation measurement of model and the model ranking based on design structural matrix(DSM).Through calculating the five dimensions of the input and output relevance between models,namely the concept of content,spatial scope,temporal scope,spatial precision,and temporal granularity,a DSM for automatic model ranking was constructed,then the intelligent selection and integrationof multiple models was realized.Taking the Wenchuan M s8.0 earthquake as an example,empirical research on rule mining in different decision-making scenarios was carried out,which provides a basis for the model integration of decision-making process in response to disasters and improves the auxiliary decision-making ability to deal with major natural disasters.