Picture of model of urban architecture

Open Access research that is exploring the innovative potential of sustainable design solutions in architecture and urban planning...

Strathprints makes available scholarly Open Access content by researchers in the Department of Architecture based within the Faculty of Engineering.

Research activity at Architecture explores a wide variety of significant research areas within architecture and the built environment. Among these is the better exploitation of innovative construction technologies and ICT to optimise 'total building performance', as well as reduce waste and environmental impact. Sustainable architectural and urban design is an important component of this. To this end, the Cluster for Research in Design and Sustainability (CRiDS) focuses its research energies towards developing resilient responses to the social, environmental and economic challenges associated with urbanism and cities, in both the developed and developing world.

Explore all the Open Access research of the Department of Architecture. Or explore all of Strathclyde's Open Access research...

Grey-box modeling for photo-voltaic power systems using dynamic neural-networks

Al-Messabi, Naji and Goh, Cindy and Li, Yun (2017) Grey-box modeling for photo-voltaic power systems using dynamic neural-networks. In: 2017 Ninth Annual IEEE Green Technologies Conference (GreenTech). IEEE, Piscataway, NJ., pp. 267-270. ISBN 9781509045358

[img]
Preview
Text (Al-Messabi-etal-GreenTech-2017-Grey-box-modeling-for-photo-voltaic-power-systems-using-dynamic-neural-networks)
Al_Messabi_etal_GreenTech_2017_Grey_box_modeling_for_photo_voltaic_power_systems_using_dynamic_neural_networks.pdf
Accepted Author Manuscript

Download (484kB) | Preview

Abstract

There exists various ways of modeling and forecasting photo-voltaic (PV) systems. These methods can be categorized, in board-way, under either definite equations models (white or clear-box) or heuristic data-driven artificial intelligence models (black-box). The two directions of modeling pose a number of drawbacks. To benefit from both worlds, this paper proposes a novel method where clear-box model is extended to a grey-box model by modeling uncertainities using focused time-delay neural network models. The grey-box or semi-definite model was shown to exhibit enhanced forecasting capabilities.