Numerical Performance Comparison of Distributed Photovoltaic Power Station (DPV) Forecasting Model Based on Two Neural Network Approaches

The power output of distributed photovoltaic power station (DPV) depends on many factors including solar radiation, temperature, wind speed and so on. Because the power output is greatly affected by environmental conditions, it has the characteristics of fluctuation, intermittence, and instability, which brings much challenge for power forecasting [1–2]. In this paper, after studying and analyzing the existing neural network approaches, two photovoltaic power station output prediction methods based on Convolutional Neural Network(CNN) and Long-Short Term Memory(LSTM) are established along with verifying the effectiveness of the algorithm by case studies via the evaluation index.