实现影响因素多源异构融合的短期负荷预测支持向量机算法

发布时间：2018-01-01 08:17

本文关键词：实现影响因素多源异构融合的短期负荷预测支持向量机算法　出处：《电力系统自动化》2016年15期 　论文类型：期刊论文

【摘要】：针对智能电网大数据环境下,导致电力系统负荷波动的诸多因素存在多源异构性的问题,利用多核函数来对其多源异构特性进行差异化处理和融合,能够描述影响因素的内在分布特性并应对其变化,提高负荷预测精度。选取历史负荷、气温、气压、相对湿度、降雨量、风向、风速、节假日及电价9个属性作为多源异构影响因素,利用样本特征分布法、单变量法及核矩阵秩空间差异法来选择多核函数的构成,采用双层多核学习算法,建立了并行化多核支持向量机(SVM)负荷预测算法流程,并在Hadoop集群上进行了仿真验证。仿真结果表明,多核SVM比单核SVM预测平均相对误差小,双层多核学习、基于lp范数的多核SVM模型预测精度最高。因此,多核SVM能有效处理负荷预测中的多源异构数据,经并行化处理后,能提高负荷预测的速度与精度。
[Abstract]:In the smart grid environment of big data, many factors that lead to power system load fluctuations are heterogeneous problems, using multiple kernel functions to different processing and fusion of multi-source heterogeneous characteristics, which can describe the influence of internal factors and distribution characteristics of the change, improve the accuracy of load forecasting. Select the history load, temperature, air pressure, relative humidity, rainfall, wind direction, wind speed, holidays and price 9 properties as heterogeneous factors, using the sample distribution method, a single variable difference method and kernel matrix rank space method to select multiple kernel function, using the double multi kernel learning algorithm, a parallel multi kernel support vector machine (SVM) load forecasting algorithm, and simulated in the Hadoop cluster. The simulation results show that the multi core SVM than single SVM average relative prediction error is small, the double layer multiple kernel learning, based on LP model The number of multi-core SVM models has the highest prediction accuracy. Therefore, multi-core SVM can effectively handle the multi-source heterogeneous data in load forecasting. After parallelization, it can improve the speed and accuracy of load forecasting.

【作者单位】：电力传输与功率变换控制教育部重点实验室(上海交通大学);国网山东省电力公司菏泽供电公司;
【基金】：国家自然科学基金资助项目(51407116) 国家科技支撑计划资助项目(2015BAA01B02)~~
【分类号】：TM715
【正文快照】： 2.国网山东省电力公司菏泽供电公司,山东省菏泽市274000)上网日期:2016-07-12。0引言随着智能电网研究的不断深入[1],高级管理体系逐步建立,智能用电设备逐渐推广[2],在这一过程中积累了大量的基础用电数据[3],同时分布式电源、储能、电动汽车等多种新型用电形式出现,各类先进

【相似文献】