针对复杂网络拓扑结构分布的不均匀性导致单一指标难以准确反映网络中节点重要程度的问题,将网络的每个节点作为1个方案,将节点重要性评价指标作为描述方案的属性,从节点度指标、介数指标、删除指标、接近中心性指标和子图指标5个方面,...针对复杂网络拓扑结构分布的不均匀性导致单一指标难以准确反映网络中节点重要程度的问题,将网络的每个节点作为1个方案,将节点重要性评价指标作为描述方案的属性,从节点度指标、介数指标、删除指标、接近中心性指标和子图指标5个方面,采用改进的逼近理想解排序法(Technique for Order Preference by Similarity to an Ideal Solution,TOPSIS)对网络节点的重要性进行综合评价,计算每个方案到理想方案的接近度,并按由大到小的顺序进行排序,得出节点重要性综合评价结果。最后,以某装备保障网络实例对该方法的可行性和有效性进行了验证。展开更多
The identification of important nodes in a power grid has considerable benefits for safety. Power networks vary in many aspects, such as scale and structure. An index system can hardly cover all the information in var...The identification of important nodes in a power grid has considerable benefits for safety. Power networks vary in many aspects, such as scale and structure. An index system can hardly cover all the information in various situations. Therefore, the efficiency of traditional methods using an index system is case-dependent and not universal. To solve this problem, an artificial intelligence based method is proposed for evaluating power grid node importance. First, using a network embedding approach, a feature extraction method is designed for power grid nodes, considering their structural and electrical information. Then, for a specific power network, steady-state and node fault transient simulations under various operation modes are performed to establish the sample set. The sample set can reflect the relationship between the node features and the corresponding importance. Finally, a support vector regression model is trained based on the optimized sample set for the later online use of importance evaluation. A case study demonstrates that the proposed method can effectively evaluate node importance for a power grid based on the information learned from the samples. Compared with traditional methods using an index system, the proposed method can avoid some possible bias. In addition, a particular sample set for each specific power network can be established under this artificial intelligence based framework, meeting the demand of universality.展开更多
约束概念格是概念格的特化结构,构造时具有较低的时空复杂度,能从中快速提取比较丰富的信息和知识.为了提取分类规则,在充分分析约束概念格结点外延与数据集等价划分之间关系的前提下,引入了分类支持度和记录支持度的概念,提出了一种面...约束概念格是概念格的特化结构,构造时具有较低的时空复杂度,能从中快速提取比较丰富的信息和知识.为了提取分类规则,在充分分析约束概念格结点外延与数据集等价划分之间关系的前提下,引入了分类支持度和记录支持度的概念,提出了一种面向约束概念格的分类规则提取算法(Classification Rule Acquisition Algorithm based on Constrained Concept Lattice,CRACCL),并采用UCI数据集作为实验集,验证了本算法能够提取更加实用和准确的分类规则.展开更多
文摘针对复杂网络拓扑结构分布的不均匀性导致单一指标难以准确反映网络中节点重要程度的问题,将网络的每个节点作为1个方案,将节点重要性评价指标作为描述方案的属性,从节点度指标、介数指标、删除指标、接近中心性指标和子图指标5个方面,采用改进的逼近理想解排序法(Technique for Order Preference by Similarity to an Ideal Solution,TOPSIS)对网络节点的重要性进行综合评价,计算每个方案到理想方案的接近度,并按由大到小的顺序进行排序,得出节点重要性综合评价结果。最后,以某装备保障网络实例对该方法的可行性和有效性进行了验证。
文摘The identification of important nodes in a power grid has considerable benefits for safety. Power networks vary in many aspects, such as scale and structure. An index system can hardly cover all the information in various situations. Therefore, the efficiency of traditional methods using an index system is case-dependent and not universal. To solve this problem, an artificial intelligence based method is proposed for evaluating power grid node importance. First, using a network embedding approach, a feature extraction method is designed for power grid nodes, considering their structural and electrical information. Then, for a specific power network, steady-state and node fault transient simulations under various operation modes are performed to establish the sample set. The sample set can reflect the relationship between the node features and the corresponding importance. Finally, a support vector regression model is trained based on the optimized sample set for the later online use of importance evaluation. A case study demonstrates that the proposed method can effectively evaluate node importance for a power grid based on the information learned from the samples. Compared with traditional methods using an index system, the proposed method can avoid some possible bias. In addition, a particular sample set for each specific power network can be established under this artificial intelligence based framework, meeting the demand of universality.
文摘约束概念格是概念格的特化结构,构造时具有较低的时空复杂度,能从中快速提取比较丰富的信息和知识.为了提取分类规则,在充分分析约束概念格结点外延与数据集等价划分之间关系的前提下,引入了分类支持度和记录支持度的概念,提出了一种面向约束概念格的分类规则提取算法(Classification Rule Acquisition Algorithm based on Constrained Concept Lattice,CRACCL),并采用UCI数据集作为实验集,验证了本算法能够提取更加实用和准确的分类规则.
