一种企业能耗预警相关性分析的时空挖掘算法-计算机系统应用.pdfVIP

2012 年 第 2 1 卷 第 7 期 计 算 机 系 统 应 用 一种企业能耗预警相关性分析的时空挖掘算法① 1,2 1 3 郝 平 ，杨剑峰 ，尹 华 1(浙江工业大学 计算机科学与技术学院，杭州 310032) 2(浙江绍兴东越科技有限公司，绍兴 312000) 3(浙江绍兴市科技信息研究院，绍兴 312000) 摘 要：为了挖掘工业企业能源监测点之间的预警关联规则，寻求生产过程中可能存在的能耗相关性，针对工 业企业综合能耗具有的时间特征和空间特征，本文提出了一种新的基于结构化项集的时空挖掘算法，以解决传 统的挖掘算法遇到的诸多瓶颈。该算法对经典的 Apriori 算进法行了改进，引入时间维度和空间维度，可在各个 时间段和不同的空间层中，挖掘出更多的能耗关联知识。该算法通过引入时间维度和空间维度，建立最小能耗 监测单元，构架分时分层的数据挖掘策略，避免传统的挖掘产生过多的候选项集。这种算法应用于国内一家大 型企业工业生产过程的能耗数据分析，取得了较好的实际效果。 关键词：能耗；数据挖掘；Apriori 算法；结构项集 Time and Space Mining Algorithm for Correlation Analysis of Energy Consumption Warning 1,2 1 3 HAO Ping , YANG Jian-Feng , YIN Hua 1(School of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310032, China) 2(DongYue ST Co. Ltd, Shaoxing 312000, China) 3(Science and Technology Information Institute, Shaoxing 312000, China) Abstract ： This paper proposes a structured item-sets time and space algorithm, which is used to find warning association rules between energy consumption monitoring points, according to the time characteristic and spatial characteristics of overall energy consumption in industrial enterprises, and the new algorithm can solve many difficulties that tradition mining algorithm encountered. The new algorithm has been improved on the basis of the classical Apriori algorithm. Introducing the time dimension and space dimension, a new algorithm can dig more associated knowledge during different time periods and in different spatial layers. Through establishing minimum energy monitoring unit and adopting time-sharing and hierarchical data mining strategy, the new algorithm avoids producing excessive candidate sets. The improved algorithm has been applied in energy data analysis of production process of a large industrial enterprise in the domestic and achieve