Short term PV power forecasting using empirical mode decomposition based orthogonal extreme learning machine technique

In present power scenario solar power generation plays a significant role, but due to its instability and irregularity nature the grid management problem becomes a prominent and challenging task. To avoid these types of problems here we have investigated an EMD based orthogonal extreme learning machine technique, which can continuously replace the old data with new data. EMD is used to decompose the data before application of various type of ELM, which eliminates the instability of the system. This paper presents the development of a reliable algorithm for short term forecasting of PV generated power of an existing solar power plant situated in Bhubaneswar, odisha, India (the detailed specification is given in table-1). Here the PV power is forecasted EMD based orthogonal ELM algorithm and the results are compared with ELM, EMD-ELM and OS-ELM. The results depict that the proposed machine learning technique performs better than other mentioned algorithms.