基于数据挖掘与关联规则的月度统计线损计算

发布时间：2018-01-18 06:23

本文关键词：基于数据挖掘与关联规则的月度统计线损计算　出处：《燕山大学》2016年硕士论文　论文类型：学位论文

【摘要】：目前,供售电量不同期问题在实际线损抄表制度中尤为突出,这使统计线损的计算结果与实际情况极为不符,同时造成统计线损率明显虚增或虚降,导致统计线损与理论线损考核指标无法比对分析。因此,解决供售电量不同期问题对统计线损的计算具有重要意义。本文提出一种基于最小二乘支持向量机法(Least Squares Support Vector Machines,LSSVM)的实际月度售电量计算模型,基于此计算实际统计线损,有效解决了供售电量不同步的问题,使得统计线损与理论线损可以进行对比分析,为供电公司线损考核与管理工作提供有价值的参考。主要内容如下:首先,研究并介绍了数据挖掘理论基本概念及算法,选用适合计算相邻两月不对应区间售电量的LSSVM算法;概述了关联规则挖掘理论及相关算法。其次,利用关联规则相关理论挖掘随机因素发生时长与售电量变化之间的关联关系,并构建经济隶属度函数与变权重系数,同时量化了单位时间内随机因素对售电量变化幅度的影响,对售电量计算结果进行修正,使其更加精确、全面。提出了基于LSSVM与关联规则挖掘的月度统计线损计算方法,同时引入信息融合理论将低压台区用户抄表数据信息融合,在等效抄表例日的基础上,计算得到等效抄表区间,解决低压用户抄表不确定性与复杂性问题。最后,在初步计算结果的基础上,利用经济因素及关联规则挖掘结果对其进行多重修正,得到与供电量同期的月度售电量,从而得到月度实际统计线损。同时提出了基于关联规则推荐的极限统计线损计算方法,通过计算相关系数寻找推荐模型的方法,扩大原有方法的使用范围,提高计算实用性。通过仿真验证并分析所提出方法的可行性与实用性。
[Abstract]:At present, the problem of different periods of electricity supply is particularly prominent in the actual line loss meter reading system, which makes the calculation results of the statistical line loss extremely inconsistent with the actual situation, at the same time, the statistical line loss rate obviously increases or falls. Statistical line loss and theoretical line loss assessment indicators can not be compared and analyzed. It is very important to solve the problem of different time period of electricity supply and sale to calculate the statistical line loss. In this paper, a new method based on least squares support vector machine (LS-SVM) is proposed. Least Squares Support Vector Machines. Based on the actual monthly sales calculation model of LSSVM, the actual statistical line loss is calculated, which effectively solves the problem that the electricity supply is out of sync, and makes the statistical line loss and the theoretical line loss can be compared and analyzed. The main contents are as follows: firstly, the basic concepts and algorithms of data mining theory are studied and introduced. Select the LSSVM algorithm which is suitable to calculate the electricity sales between the adjacent two months without corresponding interval. This paper summarizes the theory of association rules mining and related algorithms. Secondly, mining the association relationship between the length of time of random factors and the change of electricity sales by using the correlation theory of association rules. The economic membership function and variable weight coefficient are constructed, and the influence of random factors on the range of electricity sales in unit time is quantified, and the calculation results of electricity sales are revised to make it more accurate. In this paper, a monthly statistical line loss calculation method based on LSSVM and association rule mining is proposed. At the same time, the information fusion theory is introduced to fuse the data of users' meter reading in low-voltage stations, on the basis of equivalent meter reading examples. The equivalent meter reading interval is obtained to solve the uncertainty and complexity problem of low voltage users meter reading. Finally, based on the preliminary calculation results, the economic factors and association rules mining results are used to modify the data. At the same time, we get the monthly sales of electricity and get the monthly actual line loss. At the same time, we put forward the limit statistical line loss calculation method based on association rules. The method of finding the recommended model by calculating the correlation coefficient expands the scope of use of the original method and improves the practicability of calculation. The feasibility and practicability of the proposed method are verified and analyzed by simulation.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM744;TP311.13

【相似文献】