Short-term Photovoltaic Power Forecasting based on Long Short Term Memory Neural Network and Attention Mechanism

Photovoltaic (PV) power generation forecasting is an important topic in field of power system, energy conversion and smart grid construction. The PV power generation has the properties of randomness and volatility is due to the variability of solar irradiance, temperature and other meteorological factors. In order to reduce the volatility, accurate PV power generation forecasting techniques are demanded. A novel hybrid short-term PV power forecasting model based on long short term memory neural network (LSTM) and attention mechanism is proposed in this paper, where LSTM is used to extract useful features from the time series data; and attention mechanism is used to automatically focus on useful information of the extracted features. The experiment is performed with a 20kW PV power plant. In order to evaluate the performance of the proposed model, the traditional forecasting method is compared with the proposed model in different seasons and different forecasting horizons. The experimental results show that the proposed model outperforms all compared methods.