One-Hour-Ahead Solar Power Forecasting Using Artificial Neural Networks in Taiwan

Abstract

The development plan of the PV power plant in Taiwan with high penetration makes the grid operator has to prepare the strategies to mitigate its intermittency behaviour. The power grid must be more flexible to receive the fluctuated power from PV. There are some options to enhance system flexibility in dealing with renewable energy integration. One of the economical ways is to conduct a solar power forecasting. On the other hand, the use of machine learning is getting more popular in recent days. It has many applications including in prediction. In this paper, the solar power forecasting was conducted using artificial neural networks to predict the next hour of PV power output. The networks were trained using backpropagation (BPNN) and extreme learning machine (ELM). The historical PV power output, weather condition and time were used as the inputs. The extreme learning machine was used due to its advantages in accuracy and computational time over backpropagation neural networks.