Volume 13: Low Carbon Cities and Urban Energy Systems: Part II

Theory-guided LSTM for Day-ahead Forecasting of Photovoltaic Power Generation Xing Luo, Dongxiao Zhang, Xu Zhu

Abstract

This paper considers domain knowledge of photovoltaic (PV) and proposes a theory-guided long-short-term memory (Tg-LSTM) framework to forecast the hourly day-ahead PV power generation (PVPG). It aims to overcome the shortcoming of recent machine learning algorithms that are applied based only on massive data, and are thus easily producing unreasonable forecasts. Real-life PV datasets are adopted to evaluate the feasibility and effectiveness of the models. The results indicate that the proposed Tg-LSTM model possesses stronger forecasting capability than the standard LSTM model. It is more robust against PVPG forecasting, and more suitable for PVPG forecasting with sparse data in practice. The Tg-LSTM model also demonstrates superior performance with higher accuracy of PVPG forecasting compared to conventional machine learning methods.

Keywords Solar energy, Forecasting, Domain knowledge, Theory-guided LSTM