基于气象因素的光伏功率短期预测

Short-Term Prediction of Photovoltaic Power Based on Meteorological Factors

光伏发电具有随机性和波动性等特点,在一定程度上影响了与电力系统的匹配。为了解决这个问题,人们提出对光伏功率进行预测,提高精度是光伏预测的关键问题。长短时记忆(LSTM)在时间序列上具有良好的处理效果,本文基于影响光伏发电的气象因素,构建基于LSTM的深度学习模型来预测光伏发电功率。后又通过孤立森林算法对影响光伏发电的气象因素降维,并选取其中主要因素作为特征,预测结果表明,降维后的模型精度对比降维之前的模型精度有所提升。

Photovoltaic power generation has the characteristics of randomness and fluctuation,which affects the matching with power system to some extent.In order to solve this problem,people put forward to predict the photovoltaic power and improve the accuracy is the key problem of photovoltaic prediction.Long-term memory(LSTM)has a good processing effect in time series.Based on the meteorological factors affecting photovoltaic power generation,this paper constructs a deep learning model based on LSTM to predict photovoltaic power generation.Then the dimension of meteorological factors affecting photovoltaic power generation is reduced by isolated forest algorithm,and the main factors are selected as features.The prediction results show that the precision of the model after dimension reduction is higher than that before dimension reduction.