电力现货市场的售电公司短期负荷预测Short-term Load Forecasting Based on Power Spot Market for Power Sales Corporation
唐猛,董晓琦,蒋睿辰
TANG Meng,DONG Xiaoqi,JIANG Ruichen
摘要(Abstract)：
针对国内售电公司即将参与电力现货市场,需要对客户日负荷曲线进行预测的实际需求,提出了一种基于相似日的狼群-支持向量机短期负荷预测方法。该方法在相似日选取时引入近期同类型日的日平均负荷相关系数,充分考虑了负荷的延续性,同时采用模糊聚类方法来识别不同影响因素对不同负荷的影响程度;将近期同类型日的日均负荷加入到预测模型的输入变量中,并采用改进的人工狼群算法来对支持向量机法的预测模型参数进行优化,从而提高预测模型的精确度。实例验证结果表明:该算法准确度更高,能够为售电公司面对即将到来的电力现货市场提供技术支持。
Domestic power sales corporations will participate in power spot market and need to forecast customer's daily load curve. The paper proposes a short-term load forecasting method of artificial wolf pack algorithm-support vector machine(WPA-SVM) based on a similar day. This paper first introduces the recent daily average load correlation coefficient on selection of the similar day, fully considering the continuity of the load, then uses fuzzy clustering method to identify the influence degree of the different factors on different load curves. Moreover, the recent daily average load is added to the input variables of forecasting model, and the WPA-SVM algorithm is used to optimize the forecasting model parameter to improve the precision. As is shown in the example that the algorithm is more accurate and can provide technical support to power sales corporation in the face of the upcoming power spot market.
关键词(KeyWords)：
电力现货市场;售电公司;超短期负荷预测;相似日;WPA;SVM
power spot market;power sales corporation;super short-term load forecasting;similar day;WPA;SVM
基金项目(Foundation):
作者(Author):
唐猛,董晓琦,蒋睿辰
TANG Meng,DONG Xiaoqi,JIANG Ruichen
DOI: 10.19585/j.zjdl.201903016
参考文献(References)：
- [1]黎灿兵,李晓辉,赵瑞,等.电力短期负荷预测相似日选取算法[J].电力系统及其自动化,2008,32(9):69-73.
- [2]黄国勇,邵宗凯.基于日特征量相似日的PSO-SVM短期负荷预测[J].中国电力,2013,46(7):69-73.
- [3]樊唯钦,张伟.基于改进人体舒适指数的微电网超短期负荷预测[J].广东电力,2017,12(4):36-40.
- [4]任艺.考虑用户需求响应的售电公司购售电决策研究[D].北京:华北电力大学,2011.
- [5]SAINI L M,SONI M K.Artificial neural network-based peak load forecasting using conjugate gradient methods[J].IEEE Trans on Power Systems,2002,12(3):907-912.
- [6]王新,孟玲玲.基于EEMD-LSSVM的超短期负荷预测[J].电力系统保护与控制,2015,66(35):10-14.
- [7]马睿.超短期电力负荷预测的多模型极限算法[D].上海:上海交通大学,2011.
- [8]陈乐.基于加权相似度和加权支持向量机的短期电力负荷预测研究[D].广州:华南理工大学,2012.
- [9]曾勇.基于智能电网的实时电价研究[D].重庆:重庆大学,2011.
- [10]BASHIR Z,HAWARY M.Applying wavelets to short-term load forecasting using PSO-Based neural networks[J].IEEE Transactions on Power Systems,2009,23(1):20-27.
- [11]刘晓娟,方建安.基于双修正因子的模糊时间序列日最大负荷预测[J].中国电力,2013,46(10):115-118.
- [12]张贲,史沛然,蒋超.气象因素对京津唐电网夏季负荷特性影响分析[J].电力自动化设备,2013,33(12):140-144.
- [13]王雁凌,吴梦凯.经济新常态下基于偏最小二乘回归的中长期负荷预测模型[J].电力自动化设备,2018,9(3):50-53.
- [14]陈蓉珺,何永秀,陈奋开,等.基于系统动力学和蒙特卡洛模拟的电动汽车日负荷预测[J].中国电力,2018,24(9):34-37.
- [15]张宁,刘天键.考虑影响因素的短期负荷预测核函数ELM方法[J].武汉大学学报(工学版),2018,12(8):44-52.