Introduction to digital twins, models and parameter estimation

The report introduces the concept of digital twins in the built environment. It identifies that data and models that feed on this data are crucial for building a digital twin. The models are classified based on the modelling method as white box, grey box and black box models and based on the modelling problem as forward and inverse models. The report explains each of these models to throw light on how to choose a model based on the data available and the kind of objective to be achieved. The objectives can be performance prediction, parameter estimation, control, optimisation and fault detection and diagnosis. The next part of the report briefly explains the parameter estimation models. The parameter estimation models are essentially grey-box models. They are useful while making retrofit decisions as they help characterise the existing house and also, check the effectiveness of a proposed retrofit option. The report elucidates how the choice of data and the choice of the thermal network configuration influence the estimated parameters. It also further explains why detailed measurements are required to validate parameter estimation models. The last part of the report shortly describes the behavioural models, the challenges in implementing them and the importance of including them to reduce the performance gap. The report includes a few examples of behavioural models from scientific literature highlighting the data used, the modelling approach and the occupant behaviour studied. In conclusion, the report highlights the importance of data to build a digital twin of a residential building. Leveraging data available from smart meters and home automation systems is the first step to getting closer to achieving such a digital twin.