基于多时间尺度数据的分布式光伏发电预测模型

Distributed Photovoltaic Power Generation Prediction Model Based on Multi Time Scale Data

摘要: 随着风电等新能源占比的不断提升，电力系统的气候依赖性日趋凸显。基于此，文章提出了一种基于多时间尺度数据的分布式光伏发电预测方法。构建多通道卷积神经网络，利用多通道CNN从多种时间尺度上提取影响因素数据所隐含的深层特征；借助BiLSTM神经网络解决序列信息长期依赖缺失问题；通过仿真实验，对比CNN-BiLSTM预测模型与基于多时间尺度数据的CNN-BiLSTM预测模型，验证新模型具有更高的预测精度。

Abstract: With the continuous increase in the proportion of new energy such as wind power, the climate dependence of the power system is becoming increasingly prominent. Based on this, the article proposes a distributed photovoltaic power generation prediction method based on multi time scale data. The method builds a multi-channel convolutional neural network and uses multi-channel CNN to extract deep features hidden in influencing factor data from multiple time scales. Utilizing BiLSTM neural networks, the method also overcomes the long-term dependency issue in sequential data. Finally, the article compares the CNN BiLSTM prediction model with the CNN BiLSTM prediction model based on multi time scale data through simulation experiments and verifies that the new model has higher prediction accuracy.