Abstract: The prediction of photovoltaic power generation is influenced by many factors. To improve the accuracy of the prediction, statistical methods are first used to screen and statistically analyze experimental test data. In order to improve the accuracy of photovoltaic power generation prediction, environmental factors with high linear correlation with the power generation of the power station are selected and the SSA-LSTM neural network model are introduced for training and testing. By preprocessing historical data, the model firstly uses Singular Spectrum Analysis (SSA) to extract the main periodic components, and then utlizese Long Short Term Memory Networks (LSTM) for load prediction. The research results indicate that the model can effectively capture the dynamic changes in photovoltaic power generation, which has high accuracy in predicting photovoltaic power generation.

Key words: photovoltaic power generation prediction, SSA-LSTM, environmental factor

摘要： 光伏发电量预测受诸多因素影响,为提高预测的准确性,首先用统计学方法对实验测试数据进行筛选和统计,选择与电站发电量线性相关程度较高的环境因素,引入SSA-LSTM神经网络模型进行训练和测试,以提高光伏发电量预测的准确性。该模型首先利用奇异谱分析(SSA)对历史数据进行预处理,以提取主要的周期性成分,然后通过长短期记忆网络(LSTM)进行负荷预测。研究结果表明,该模型能够有效捕捉光伏发电量的动态变化,在光伏发电量预测方面具有较高的准确性。

关键词: 光伏发电量预测, SSA-LSTM, 环境因素