The drastic growth in the conventional transportation system raises serious air pollution concerns. Eco-friendly vehicles, in contrast, have been introduced as an alternative to alleviate such environmental issues. To support the Canadian government’s goal of achieving 100% sales of zero-emission vehicles by 2035, there is an increasing need for advancements in charging infrastructure and the performance of Electric Vehicles (EVs). These improvements aim to address range anxiety which is the primary concern of EV consumers who fear running out of electricity during a journey and being unable to find a charging point. However, so far, the main investment focus has been on the installation of Fixed Charging Stations (FCSs) which requires significant budget contributions and proper charging station placements. Therefore, to achieve higher EV popularity, this work aims to elevate user satisfaction and alleviate Range Anxiety by developing an intelligent system to manage EV charging demands, accurately estimating State of Charge (SoC) levels, and offering user-centric suitable service recommendations. Nevertheless, the scarcity of EVs historical data for Artificial Intelligence (AI)-based predictions poses a significant difficulty. To mitigate the aforementioned concern, we present a model based on Deep Transfer Learning (DTL) between domain-variant data sets, to reduce the need for the existence of a vast amount of EV data, including driving characteristics and patterns.

Furthermore, the accurate proposed DTL-based energy consumption estimation mechanism allows us to introduce an optimized Federated Learning (FL)-based recommendation system. Our intelligent system identifies nearby charging sources while preserving user privacy, which is a crucial aspect of the Internet of Things (IoT) framework security. Since the focus of this system is to increase the recommendation performance, we go beyond FCSs by integrating recently developed flexible Mobile Charging Stations (MCSs) to enhance the functionality of our system for EV consumers. This system prioritizes user satisfaction by balancing the desired charging outcomes with minimal inconvenience and expenses among all types of available charging services. Hence, it alleviates range anxiety and enhances the overall user experience within the EV ecosystem. We further investigate the inherently dynamic functioning environment of MCSs, which is influenced by inconstant user preferences, energy demands, and charging service availability, to ensure cost efficiency and fairness. Therefore, in this thesis, we develop an optimization FL-based mechanism that enables decentralized learning with minimal data transfer to maximize the potential profit of MCSs while optimizing their daily operational efficiency.