摘要
针对目前很多现有模型和算法难以应用在小样本电力市场数据集上的问题,本文提出了一种基于日前披露数据相似性的电力市场出清价格预测方法。为了更加精准地在小样本数据上对电力市场出清价格进行预测,本文通过计算日前披露数据各指标与出清价格之间的相关性系数,分析出清价格与各指标之间的相关性,并设计了一种基于相关系数的相似性计算方法,将待预测数据与历史数据的相似性进行计算和对比,找出相似度最高的历史数据,并将该条历史数据的出清价格作为预测结果。本文在某区域电力市场3个月的试运行数据上进行了验证,经过测试该方法能够避免时间不连续以及价格钉对于电力市场出清价格预测产生的影响,实验结果表明该方法在小样本数据集上达到预期的预测结果。
Aiming at the problem that many existing models and algorithms are difficult to be applied to small sample power market data sets, this paper proposed electricity market clearing price forecasting method based on the similarity of day ahead disclosure data. In order to predict the clearing price in the power market more accurately on small sample data, this paper calculates the correlation coefficient between each index and clearing price, analyzes the correlation between clearing price and each index, and designed a similarity calculation method based on correlation coefficient, calculate and compare the similarity between the data to be predicted and the historical data, find the historical data with the highest similarity, and take the clearing price of the historical data as the prediction result. In this paper, the test is carried out on the three-month trial operation data of some regional power market. After testing, this method can avoid the impact of time discontinuity and price spike on the clearing price prediction of power market. Experimental results show that this method achieved expected prediction results on a small sample data set.
作者
杨乘胜
张世超
朱海东
赵竟
张永涵
张庭玉
YANG Chengsheng;ZHANG Shichao;ZHU Haidong;ZHAO Jing;HUI Mingcheng;ZHANG Tingyu(Nanjing Huadun Electric power Information Security Assessment Co.,Ltd.,Nanjing 210000Jiangsu,China;China Huadian Co.,Ltd.,Beijing 100031,China)
出处
《电力大数据》
2022年第1期59-66,共8页
Power Systems and Big Data
关键词
电力市场
出清价格预测
相关性
披露数据
数据相似性
electricity market
clearing price forecast
spot market
disclosed data
short term forecast
