A behaviour digital twin for residential demand response: Modelling intention and motivation to improve the prediction of the likelihood of reaction to behavioural triggers

Abstract

Background: Residential demand response is a resource in the evolving energy infrastructure which thus far has not achieved its full potential. Amongst the reasons for this underutilisation is a lack of understanding, and therefore predictability, in relation to the uncertainty of the behaviour of human actors and its potential impact on energy demand side management. Optimal model predictive control of energy assets requires a digital twin to operate, however, most approaches so far are focused predominantly on technical indicators only and neglect the individuality of people and their behaviour in the operation. To fully integrate human led actions into such a system, the digital twin must therefore also provide social and psychological indicators to facilitate better predictability of reactions to demand response triggers. Methods: In the following a behaviour digital twin model will be presented based on the theory of planned behaviour and the self-determination theory, which provide well-established and validated tools to capture indicators of intention and motivation. The key contribution of this work is to operationalise and combine these models into a software tool, which continuously adapts its parameters to the evolving behaviour of users and provides up-to-date predictions. Results: The resulting model predicts the likelihood of each individual to react to appropriate demand response triggers, which can be used in model predictive control involving human actors to optimally select whom to target and when. Conclusions: The presented behaviour digital twin aims at bridging the gap between research in psychology to evaluate and assess drivers of behaviour and innovations in the space of model predictive control to optimally facilitate asset operation in residential settings.