基于贝叶斯网络的电网事故预警模型-《计算机技术与发展》

文章信息/Info

Title:: Early Warning Model of Power Grid Accident Based on Bayesian Network

Author(s):: LIU Yun-fei; LU Kui; School of Computer Science &Engineering,Anhui University of Science &Technology,Huainan 232001,China

Keywords:: electric power production; accident warning; greedy algorithm; mutual information; Bayesian network

摘要:: 近年来电力生产与作业过程中的安全问题频频发生,给人民生活带来不便,严重阻碍了电力行业的进一步发展。文中基于贝叶斯网络寻求一种电网事故预警模型。贝叶斯网络使用有向无环图来表示随机变量之间的条件独立关系。每个贝叶斯网络定义随机变量的联合实例化的联合概率分布。并且贝叶斯网络被认为是概率知识表示和推理领域中最有用的模型之一。为了构建模型,寻求一种新颖的贪婪算法。首先,统计电网事故致因数据属性的各个概率;其次,计算各个属性之间的互信息;最后,利用贪婪算法,每次选取最大的互信息属性,实现了贝叶斯网络的构造,进而完成电网事故预警模型的构建。使用 Adult 数据集进行模拟实验,结果表明构造的贝叶斯网络可以带来较好的分类效果。

Abstract:: In recent years,the frequent occurrence of safety in electric power production and operation has brought inconvenience to people’s life and seriously hindered the further development of the electric power industry. We seek a grid accident warning model based on Bayesian network. Bayesian networks use directed acyclic graphs to represent conditional independent relationships between random variables. Each Bayesian network defines a joint probability distribution of joint instantiations of random variables,which is considered to be one of the most useful models in the field of probabilistic knowledge representation and reasoning. To build the model,we seek a novel greedy algorithm. First,the probability of the data attribute of the cause of the grid accident is calculated. Secondly,the mutual information between the attributes is computed. Finally,the greedy algorithm is utilized to select the largest mutual information attribute each time for construction of the Bayesian network,and then an early warning model for power grid accidents is constructed. We use the Adult dataset for simulation which shows that the Bayesian network constructed can bring better classification effect.