中文摘要：

由于面向地震应急响应系统（EERS）的研判模型有其自身特点,传统决策支持系统中模型管理方法并不能较好地满足其管理需求。提出一种面向地震应急响应的松耦合研判模型管理机制（LC-MM）,包括3个部分：（1）研判资源建模环境,由研判模型描述/控制语言（MDCL）和数据源描述/控制语言（DSDCL）组成;（2）研判资源协同交互框架,该框架分为4层,主要包括应用环境上下文App_Context、模型调用者Model_Invoker和数据通道Data_Channel 3个Agent组件;（3）研判资源协同交互协议,该协议将研判任务、研判模型和数据源的紧耦合关系分解为上述Agent组件之间的协同交互关系。实际应用表明,LC-MM能够有效屏蔽研判资源的异构性,提高EERS的可扩展性、动态适应性以及平台无关性,能够较好地满足EERS中研判模型的管理需求。

英文摘要：

Because analysis models in earthquake emergency response system （EERS） have special char- acteristics, traditional model management methods in decision support system are not suitable well for the man- agement requirement of analysis models. We propose a loose-coupled mechanism, called loose-coupled mecha- nism for model management （LC-MM） , for earthquake emergency response. LC-MM is composed of three parts ： 1 ） an analysis resources modeling environment, which consists of 2 modeling languages, i. e. , model description and control language （MDCL） and data source description and control language （DSDCL） ; 2） an Agent-based framework for analysis resources coordinating and interacting, which is divided into 4 layers and 3 Agent components,i, e. , App_Context, Model_Invoker and Data_Channel; 3） a protocol for Agent-based components coordinating. The protocol changes the tight-coupled relationship between analysis models, data sources and tasks into a loose-coupled relationship through Agent components. Practical application shows that LC-MM can effectively cover up analysis resources＇ heterogeneity and improve the scalability, dynamic flexi- bility and fiat-independency of EERS. The proposed mechanism can match the management requirement of analysis models in EERS.