Short-Term Photovoltaic Prediction Based on Transfer Learning and LSTM

• Published in: 2023 IEEE 7th Conference on Energy Internet and Energy System Integration (EI2) pp. 3438 - 3443
• Main Authors: Zhang, Tianyu, Zheng, Kedi, Chen, Qixin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 15.12.2023
• Subjects: Data models; LSTM; photovoltaic prediction; Photovoltaic systems; Predictive models; Renewable energy sources; System integration; Transfer learning; Weather forecasting
• DOI: 10.1109/EI259745.2023.10513117

With the proposal of carbon peaking and carbon neutrality goals, the proportion of renewable energy generation continues to increase and the number of new photovoltaic power stations is growing rapidly. Accurate photovoltaic power prediction is of great significance for the safe, stable and operation of power system. In order to solve the problem of insufficient historical data of newly constructed PV power plants, this paper proposes a short-term PV power prediction model based on Long Short Term Memory and transfer learning. The model uses meteorological data, numerical weather forecast data and PV power as features. According to the transfer learning theory, the Long Short Term Memory model is pre-trained using the historical data of other previously-built PV power plants, and then a small amount of local operation data is used to fine-adjust the model. Case study based on Hebei open source photovoltaic and meteorological data set proves that the proposed transfer learning model can effectively improve the short-term PV power prediction accuracy, and the MAPE can be improved by up to 15%.