摘要
为提高光伏功率预测精度以减少光伏接入对电网运行的不利影响,提出一种基于K-means++和LSTM网络的光伏功率预测方法。首先,利用改进的K-means方法对历史数据进行聚类;然后,建立基于LSTM神经网络的预测模型,用聚类后数据集对提出的预测模型进行了训练和测试,为提高模型的预测精度,进行了一系列的仿真和参数选择;最后,将方法和单一LSTM网络和BP神经网络进行了比较。结果表明中,方法具有较好的准确性和通用性。
A photovoltaic power prediction method based on K-means++and LSTM network was proposed to improve photovoltaic power prediction accuracy so as to reduce adverse influence of photovoltaic access on grid operation.Firstly,the improved K-means method was used for the clustering of historical data.Then,a prediction model based on the LSTM neural network was established,and the proposed prediction model was trained and tested by means of clustered data set.A series of simulation and parameter selections were carried out to improve prediction accuracy of the model.Finally,the proposed method was compared with single LSTM network and BP neural network.The results showed that the proposed method was quite accurate and universal.
作者
黄亚峰
何威
吴光琴
李丹
Huang Yafeng;He Wei;Wu Guangqin;Li Dan(Northeast Electric Power University,Jilin city Jilin Province 132012,China)
出处
《电气自动化》
2020年第5期25-27,34,共4页
Electrical Automation
基金
国家自然科学基金委员会-国家电网公司智能电网联合基金项目(U1866601)
吉林省科技发展计划项目(20160101346JC)。
