69 / 2024-08-14 19:33:08

Research on Power Prediction of Distributed Photovoltaic Station Based on Space Remote Sensing Data AITC 2024+空天之星

Remote Sensing Data,Distributed Photovoltaic Power Prediction,Features Fusion,LSTM Neural Network,ARIMA

With the rapid development of distributed photovoltaic (PV) power generation, the volatility and randomness of PV power generation bring challenges to dispatching of power grid.In this paper, the relationship between satellite cloud image and distributed PV power generation is studied, so as to predict the distributed PV power. Firstly, the PV power prediction data collection platform was built; Then, the cloud image features are extracted, the texture features and the whole features of the cloud image are extracted by gray co-occurrence matrix and convolutional neural network, and the feature dimension reduction and fusion processing are carried out by principal component analysis. Autoregressive Integrated Moving Average (ARIMA) was used to predict the future time cloud image fusion features. Long Short-Term Memory(LSTM) neural network is used to predict distributed photovoltaic power. Finally, power generation is predicted by power-time diagram. The results show that the method proposed in this paper can predict distributed PV power from satellite cloud image data, and its root-mean-square error is 5.7034w. Photovoltaic power can be predicted with high precision.