Green Neighbourhoods: The Role of Big Data in Low Voltage Networks' Planning

In this chapter, we aim to illustrate the benefits of data collection and analysis to the maintenance and planning of current and future low voltage networks. To start with, we present several recently developed methods based on graph theory and agent-based modelling for analysis and short-and long-term prediction of individual households electric energy demand. We show how maximum weighted perfect matching in bipartite graphs can be used for short-term forecasts, and then review recent research developments of this method that allow applications on very large datasets. Based on known individual profiles, we then review agent-based modelling techniques for uptake of low carbon technologies taking into account sociodemographic characteristics of local neighbourhoods. While these techniques are relatively easily scalable, measuring the uncertainty of their results is more challenging. We present confidence bounds that allow us to measure uncertainty of the uptake based on different scenarios. Finally, two case-studies are reported, describing applications of these techniques to energy modelling on a real low-voltage network in Bracknell, UK. These studies show how applying agent-based modelling to large collected datasets can create added value through more efficient energy usage. Big data analytics of supply and demand can contribute to a better use of renewable sources resulting in more reliable, cheaper energy and cut our carbon emissions at the same time.