基于ACO-VMD-LSTM的光伏功率超短期预测研究
Study on ACO-VMD-LSTM-based Ultra-short Term Prediction of Photovoltaic Power
郑雨1
作者信息
- 1. 西安工程大学电子信息学院,陕西 西安 710048
- 折叠
摘要
针对传统光伏功率超短期预测算法精度不高的问题,提出一种基于改进变分模态分解的长短期记忆网络的光伏功率预测模型.首先利用Pearson相关系数分析光伏功率影响因素,其次利用基于蚁群算法优化的变分模态分解对光伏功率序列进行分解,并将各模态分量级气象因素作为长短期记忆网络的输入,得到预测功率.仿真结果表明,与BPNN、LSTM模型相比,所提出的预测模型具有较高的预测精度,可为光伏电站功率预测提供参考.
Abstract
Aiming at improving the low accuracy of conventional ultra-short term algorithms for photovoltaic power pre-diction,this paper proposes a prediction model based on modified variational mode decomposition and short-term memory network.First the influencing factors of photovoltaic power are analyzed using Pearson correlation coefficient.Second the photovoltaic power sequence is decomposed using an ant colony algorithm-based variational mode decomposition,and the predicted power is obtained by input meteorological factors at all independent modal component levels into long-term and short-term memory network.The proposed model is shown by simulation results more accurate than BPNN and LSTM models,and may be useful in predicting powers of photovoltaic generations.
关键词
光伏功率/超短期预测/蚁群优化算法/长短期记忆网络Key words
photovoltaic power/ultra-short term prediction/ant colony optimization algorithm/long and short term memory network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据