多维关联规则的分布式能源系统数据挖掘方法研究

Research on data mining method of distributed energy system based on multidimensional association rules

为了提升分布式能源系统数据挖掘精度与效率,保证分布式能源系统安全稳定运行,提出基于多维关联规则的分布式能源系统数据挖掘方法。离散化处理分布式能源系统中的所有数据,将数据转换为满足挖掘需求的数字化形式,明确数据支持度和隶属度值,建立与之对应的相关表,统计操作、运行和执行维度组合后总的支持度和隶属度值,通过分析每个维度的子维度得到正常和异常状态下两种数据。在Apriori算法基础上构建FP-树,不断生成频繁项集,直至不再产生新的项集时停止挖掘,采用剪枝处理去除部分冗余结点,保留下来的即为最终的数据挖掘结果。实验结果表明,所提方法能够精准挖掘出异常数据类型,且挖掘效率高,实际应用效果更好。

In order to improve the accuracy and efficiency of distributed energy system data mining and ensure the safe and stable operation of distributed energy system,a distributed energy system data mining method based on multi-dimensional association rules was proposed.Discretize all the data in the distributed energy system,convert the data into a digital form that meets the mining requirements,clarify the data support and membership values,establish corresponding related tables,count the total support and membership values after the combination of operation,operation and execution dimensions,and obtain two kinds of data under normal and abnormal states by analyzing the sub dimensions of each dimension.The FP-Tree is constructed on the basis of Apriori algorithm,and frequent itemsets are continuously generated until no new itemsets are generated.The mining is stopped.Some redundant nodes are removed by pruning,and the retained data is the final data mining result.Experimental results show that the proposed method can accurately mine abnormal data types,has high mining efficiency and better practical application effect.