随着电力系统数据采集手段的不断完善,基于数据的分析方法在电力系统运行分析中扮演着日益重要的角色。现有的数据分析方法主要分析数据之间的相关关系。事实上,两个强相关变量间通常呈现出不对称的因果关系。若能揭示电力系统运行变量...随着电力系统数据采集手段的不断完善,基于数据的分析方法在电力系统运行分析中扮演着日益重要的角色。现有的数据分析方法主要分析数据之间的相关关系。事实上,两个强相关变量间通常呈现出不对称的因果关系。若能揭示电力系统运行变量间的因果关系,必将有助于深刻地洞察电力系统运行的内在规律性。近年来,因果推断的研究取得很大进展,使得基于数据的因果分析成为可能。该文从物理机制上揭示电力系统中强相关变量之间因果关系的不对称属性;提出一种逆信息熵因果推理(reciprocal information entropy causal inference,RIECI)方法,所构建的指标不仅可以有效判别相关变量间的因果方向,还能正确反映因果强度。在电力系统算例中的验证表明,RIECI方法能有效揭示电力系统运行数据中的因果关系。对电力系统运行数据中因果关系的分析对于认知电力系统运行机理和正确调控电力系统运行状态有重要意义。展开更多
Causal reasoning plays an important role in situation assessment (SA). Using Bayesian networks to find out the hidden patterns between situation hypothesis and events is the function needed to accomplish in situation ...Causal reasoning plays an important role in situation assessment (SA). Using Bayesian networks to find out the hidden patterns between situation hypothesis and events is the function needed to accomplish in situation as sessment. Based on different link relationship,a Bayesian network model for situation assessment is analyzed in this paper. To overcome the weakness of this model in application for dynamic changed scenario ,this paper presents an approach that uses a dynamic Bayesian network to represent features of the situation hypothesis and events. And the algorithms of propagation of corresponding information through the network are introduced respectively.展开更多
文摘随着电力系统数据采集手段的不断完善,基于数据的分析方法在电力系统运行分析中扮演着日益重要的角色。现有的数据分析方法主要分析数据之间的相关关系。事实上,两个强相关变量间通常呈现出不对称的因果关系。若能揭示电力系统运行变量间的因果关系,必将有助于深刻地洞察电力系统运行的内在规律性。近年来,因果推断的研究取得很大进展,使得基于数据的因果分析成为可能。该文从物理机制上揭示电力系统中强相关变量之间因果关系的不对称属性;提出一种逆信息熵因果推理(reciprocal information entropy causal inference,RIECI)方法,所构建的指标不仅可以有效判别相关变量间的因果方向,还能正确反映因果强度。在电力系统算例中的验证表明,RIECI方法能有效揭示电力系统运行数据中的因果关系。对电力系统运行数据中因果关系的分析对于认知电力系统运行机理和正确调控电力系统运行状态有重要意义。
文摘Causal reasoning plays an important role in situation assessment (SA). Using Bayesian networks to find out the hidden patterns between situation hypothesis and events is the function needed to accomplish in situation as sessment. Based on different link relationship,a Bayesian network model for situation assessment is analyzed in this paper. To overcome the weakness of this model in application for dynamic changed scenario ,this paper presents an approach that uses a dynamic Bayesian network to represent features of the situation hypothesis and events. And the algorithms of propagation of corresponding information through the network are introduced respectively.
