A digital twin is a virtual representation of a physical object or process capable of collecting information from the real environment to represent, validate and simulate the physical twin’s present and future behavior. It is a key enabler of data-driven decision making, complex systems monitoring, product validation and simulation and object lifecycle management. As an emergent technology, its widespread implementation is increasing in several domains such as industrial, automotive, medicine, smart cities, etc. The objective of this systematic literature review is to present a comprehensive view on the DT technology and its implementation challenges and limits in the most relevant domains and applications in engineering and beyond.

Air pollution impact assessment is a major objective for various community councils in large cities, which have lately redirected their attention towards using more low-cost sensing units supported by citizen involvement. However, there is a lack of research studies investigating real-time mobile air-quality measurement through smart sensing units and even more of any data-driven modelling techniques that could be deployed to predict air quality accurately from the generated data-sets. This paper addresses these challenges by: a) proposing a comparative and detailed investigation of various air quality monitoring devices (both fixed and mobile), tested through field measurements and citizen sensing in an eco-neighbourhood from Lorraine, France, and by b) proposing a machine learning approach to evaluate the accuracy and potential of such mobile generated data for air quality prediction. The air quality evaluation consists of three experimenting protocols: a) first, we installed fixed passive tubes for monitoring the nitrogen dioxide concentrations placed in strategic locations highly affected by traffic circulation in an eco-neighbourhood, b) second, we monitored the nitrogen dioxide registered by citizens using smart and mobile pollution units carried at breathing level; results revealed that mobile-captured concentrations were 3 to 5 times higher than the ones registered by passive-static monitoring tubes and c) third, we compared different mobile pollution stations working simultaneously, which revealed noticeable differences in terms of result variability and sensitivity. Finally, we applied a machine learning modelling by using decision trees and neural networks on the mobile-generated data and show that humidity and noise are the most important factors influencing the prediction of nitrogen dioxide concentrations of mobile stations.

Congestion prediction represents a major priority for traffic management centres around the world to ensure timely incident response handling. The increasing amounts of generated traffic data have been used to train machine learning predictors for traffic, however this is a challenging task due to inter-dependencies of traffic flow both in time and space. Recently, deep learning techniques have shown significant prediction improvements over traditional models, however open questions remain around their applicability, accuracy and parameter tuning. This paper proposes an advanced deep learning framework for simultaneously predicting the traffic flow on a large number of monitoring stations along a highly circulated motorway in Sydney, Australia, including exit and entry loop count stations, and over varying training and prediction time horizons. The spatial and temporal features extracted from the 36.34 million data points are used in various deep learning architectures that exploit their spatial structure (convolutional neuronal networks), their temporal dynamics (recurrent neuronal networks), or both through a hybrid spatiotemporal modelling (CNN-LSTM). We show that our deep learning models consistently outperform traditional methods, and we conduct a comparative analysis of the optimal time horizon of historical data required to predict traffic flow at different time points in the future.