摘要
对电网调度数据进行数据挖掘是进行深入关联分析和处理的有效手段,采用粗糙集对数据进行建模并采用属性约简算法简化处理难度,是目前可行的一种方法。随着电网调度数据中对象数量的不断膨胀,相关属性组合也呈爆炸式增长,寻找最小属性约简集合已经被证明是个NP难的问题。目前已有的算法受属性约简问题中参数难以定义、优化目标函数不确定等因素的影响,容易在搜索结果时陷入局部最优而无法获得理想的结果。提出一种模拟植物生长寻找最小属性约简的新算法,通过植物的向光性来搜索属性集合的可行解。理论分析和实验表明,该算法不仅复杂度较低,而且能获得更小的属性约简集。
Data mining is an effective way to the deep correlation analysis and treatment of power grid dispatching data. It is a reasonable way to use rough set to model the data and attribute reduction algorithm to simplify the difficulty of data mining. With the rapid increasing of object number in power dispatching data,corresponding attribute combinations are increasing explosively. How to find minimum attribute reduction set has been proved to be a NP-hard problem. Existing algorithms are influenced by factors of hardly defined parameters and uncertain optimal object function in attribute reduction problem so that they are easy to be stick to local optimal solutions and hard to achieve ideal results. A novel plant growth simulation based algorithm to find minimum attribute reduction set is proposed, which uses phototropism of plant to search feasible solutions of attribute set. Theory analysis and experiment results show that the algorithm has lower complexity and can get smaller attribute reduction set.
出处
《广西电力》
2013年第3期4-8,共5页
Guangxi Electric Power
关键词
模拟植物生长算法
电网调度数据
数据挖掘
粗糙集
属性约简
plant growth simulation algorithm
power dispatching data
data mining
rough set
attribute reduction
