378735 Design of Robust Renewable Energy Supply Systems By Providing Visibility Based upon Improved Neural Network

Monday, November 17, 2014: 4:55 PM
M304 (Marriott Marquis Atlanta)
Soo bin Lee, POSTECH, Pohang, South Korea, Jun-hyung Ryu, Nuclear and Energy System Engineering, Dongguk University, Gyeongju, South Korea and In-Beum Lee, Chemical Engineering, POSTECH, Pohang, South Korea

Climate change and rising supply costs make us agree alternative energy sources should be introduced in replace of fossil fuels. Various renewable energy sources are thus investigated. It is clear that renewable energies become a practical contributor to energy supply system when we can manipulate their performances. However unpredictable nature of renewable energy outputs make it difficult to incorporate them into the main energy supply system directly.  The wide implementation of renewable energies makes it even more necessary.

Alternatively the estimated output of renewable energy supply should be available before they are realized. When such estimation is available, the visibility of energy output would be greatly improved. In the end the portion of fossil fuels can be reduced.

This paper investigates existing forecasting methodologies such as neural network and the up-to-date linear / nonlinear time series methodologies are with the comparison of their performances. We propose a new optimized neural network based forecasting methodology for the case of photovoltaic power generation. The applicability of the proposed methodology is illustrated in 10KW real case in Korea.

Extended Abstract: File Not Uploaded
See more of this Session: Achieving Sustainable Buildings through Chemical Engineering
See more of this Group/Topical: Environmental Division