基于FCM和ITransformer-TCN的短期风电集群功率预测
Short-term Power Prediction of Wind Power Cluster Based on FCM and ITransformer-TCN
牛甲俊 1张薇 1许达明2
作者信息
- 1. 现代电力系统仿真控制与绿色电能新技术教育部重点实验室(东北电力大学),吉林 吉林 132012
- 2. 国网吉林省电力有限公司白城供电公司,吉林 白城 137000
- 折叠
摘要
准确的风电功率预测对于电力系统安全稳定运行具有至关重要的意义.针对当前在集群预测中存在集群划分不合理以及在短期预测中精度难以得到有效提升的问题,文中提出了一种基于模糊 C 均值(Fuzzy C-means,FCM)和ITransformer-时间卷积网络(Temporal Convolutional Network,TCN)的短期风电集群功率预测方法.首先基于FCM聚类算法划分子集群,再利用ITransformer-TCN模型双重特征提取的优势对各子集群建模预测,最后将文中方法应用于中国吉林省某风电集群,与其他方法对比RMSE平均降低了 10.8%,验证了模型的有效性.
Abstract
Accurate wind power prediction is of great significance for the safe and stable operation of the power system.In view of the problems of unreasonable cluster division in cluster prediction and difficult to effectively improve the accuracy in short-term prediction,this paper proposes a short-term wind power cluster power prediction method based on Fuzzy C-means(FCM)and ITransformer-time convolutional network(Temporal Convolutional Network,TCN).First,divide subclusters based on FCM clustering algorithm,and then use the advantages of ITransformer-TCN model dual feature extraction to model each subcluster.Finally,this method was applied to a wind power cluster in Jilin Province,China,and the RMSE decreased by 10.8%on average compared with other methods,which verified the effectiveness of this paper.
关键词
风电功率预测/FCM/ITransformer-TCN/双重特征提取/短期集群预测Key words
wind power prediction/FCM/ITransformer-TCN/dual feature extraction/short-term cluster forecasting引用本文复制引用
基金项目
国家重点研发计划(2022YFB2403000)
国家电网科技项目(522722230034)
出版年
2024
NETL
NSTL
万方数据