Department of Economics Seminar Series 



"Spatial dependency and complex energy systems: an integrated agent-based modelling and artificial neural networks approach"




Ali Alderete-Peralta & Nazmiye Balta-Ozkan

(Cranfield University)



Date: April 25, 2022 (Monday)

Time: 14:00


Synchronous Online Seminar

Zoom Platform


Zoom Link

Meeting ID: 962 6861 4522

Passcode: 319178


The ambitions to reduce carbon-emissions to net-zero requires significant changes in the ways we generate, distribute and consume energy. It has been long recognised that consumers are far from rational decision makers. And yet, all the actors need to recognise the externalities their choices create and adopt different behaviours. The interaction of multiple actors and their vested interests highlight how complexity science can inform energy systems modelling and research. With increasing availability of high granular data on energy behaviours and choices, there is scope to integrate elements of human cognition into the modelling approaches, reflecting the complexity of decision-making.

In our research, we adopt the Agent-based modelling (ABM) approach because of its capabilities to characterise the interactions and decision-making of the different actors in the energy system. Moreover, we draw from the Spatial Regression and Artificial Neural Networks (ANN) to create an approach able to account for: (i) spatial regularities, (ii) temporal regularities, (iii) and social dynamics.

We investigate the adoption of electric vehicles and solar panels, producing insights on the location and pace of the adoption process. Results suggest that the model can account for the spatio-temporal and social dynamics that drive the decision-making, and can estimate adoption rates for up to five months with an accuracy of 80%. The framework addresses some of the limitations of ABMs that use a rule or equation-based decision-making, by adopting an aggregated agents' definition and using ANN as decision-making criteria. Therefore, integrated model provides a more realistic characterisation of the decision-making and its evolution over time, moreover, the results can inform network operators and policymakers explicitly about the location and pace of EV and PV adoption.

Finally, the research is a paragon for the modelling of the spatio-temporal adoption patterns of EVs and PVs, which can be exploited as confidence in artificial intelligence models increases and empirical datasets become more and more available. Moreover, this research leads towards more complex approaches of decision-making that recognise the multiple dynamics driving the adoption process of low-carbon technologies.

Last Updated:
18/04/2022 - 20:35