Combined forecasting system for short-term bus load forecasting based on clustering and neural networks

• Published in: IET generation, transmission & distribution transmission & distribution, 2020-09, Vol.14 (18), p.3652-3664
• Main Authors: Panapakidis, Ioannis P, Skiadopoulos, Nikolaos, Christoforidis, Georgios C
• Format: Article
• Language: English
• Subjects: aggregated system loads; bus load predictions; bus loads; combined forecasting system; distributed power generation; feedforward neural nets; feed‐forward neural network; forecasting models; load forecasting; main grid; micrographs; microgrids literature; neural nets; neural networks; power engineering computing; power systems design; Research Article; research momentum; robust forecasting system; short‐term bus load forecasting; short‐term horizon; smart grids; smart power grids; transmission system
• ISSN: 1751-8687; 1751-8695; 1751-8695
• DOI: 10.1049/iet-gtd.2019.1057

Micro-grids as ‘micro-graphs’ of the power systems involve the management of small loads, either isolated or connected to the main grid. Load forecasting is a tool of fundamental importance in power systems design and operation. During the last years, many types of research have focused on aggregated system loads. However, few studies deal with small loads and especially with bus loads of the transmission system. While smart grids and micro-grids literature are gathering research momentum, there is an emergent need for more investigation on forecasting models for buses. In this study, the aim of this work is to propose a novel robust forecasting system for bus load predictions on a short-term horizon. The model refers to the hybridisation of clustering and feed-forward neural network (FFNN). Experimental results and analysis indicate the robustness of the model; the combination of clustering and FFNN provides better forecasts compared with the single application of the FFNN.