| 基于支持向量机的电价价格钉预测 |
| |
| 引用本文： | 李正欣,赵林度.基于支持向量机的电价价格钉预测[J].中国电力,2007,40(11):42-45. |
| |
| 作者姓名： | 李正欣 赵林度 |
| |
| 作者单位： | 1. 东南大学,系统工程研究所,江苏,南京,210096;江苏省国信集团,江苏,南京,210005
2. 东南大学,系统工程研究所,江苏,南京,210096 |
| |
| 摘 要： | 电力行业在放松管制、引入竞争机制、提高效率的同时,也带来了电价的剧烈波动问题,成为威胁电力市场稳定安全经济运行的一个重要方面。在确定价格钉判别标准的基础上,分析价格钉的主要分布特点。针对价格钉与正常价格分布的不平衡性,为提高价格钉发生预测的准确性,对标准的支持向量机进行修正:一方面运用少数类样本综合仿制技术(SMOTE)增加价格钉这一少数类样本的数量;另一方面对价格钉与正常价格这2类分别设置不同的惩罚系数。方法应用于澳大利亚昆士兰电力市场中价格钉预测,取得了较好的效果。
|
| 关 键 词： | 价格钉 支持向量机 非平衡数据集 |
| 文章编号： | 1004-9649（2007）11-0042-04 |
| 修稿时间： | 2007-05-25 |
An electricity price spikes forecasting model based on SVM |
| |
| LI Zheng-xin,ZHAO Lin-du.An electricity price spikes forecasting model based on SVM[J].Electric Power,2007,40(11):42-45. |
| |
| Authors: | LI Zheng-xin ZHAO Lin-du |
| |
| Affiliation: | 1.Institute of System Engineering, Southeast University, Nanjing 210096, China; 2. Jiangsu Guoxin Investment Corporation Limited, Nanjing 210005, China |
| |
| Abstract: | The deregulation of power industry brings the competitive environment, and makes the industry more efficient while the price spikes threat the power market at the same time. In this paper, the imbalanced distribution character of price spikes was presented. To improve prediction accuracy of spike occurrence using SVM, the paper combined the SMOTE over-sampling technique with different penalty factor method for spike and normal classes. The model was tested with the Queensland electricity market data with promising results. |
| |
| Keywords: | SMOTE |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
