AI Algorithms for Graph-based Optimization in Energy and Electrified Transport Systems

This thesis explores the application of graph-based optimization techniques to enhance the efficiency and sustainability of electrified bus networks in urban environments. As cities worldwide transition towards electrification to reduce carbon emissions and mitigate the effects of climate change, the need for optimized public transport systems becomes increasingly critical. This research focuses on leveraging graph theory and advanced optimization algorithms to address the complex challenges associated with the deployment and operation of electric bus fleets, including route planning, charging station placement, and energy management. By modelling bus networks as graphs, where nodes represent services, terminal stops or charging stations and edges denote feasibility of serving two services by one bus or the driving range from a terminal stop to a charging station, the study introduces novel methods to optimize key performance indicators such as fleet size, energy consumption, number of charging stations and the total load applied to electrical grid. The proposed optimization framework integrates factors such as traffic patterns, energy constraints, and passenger demand, enabling the design of more resilient and adaptive bus networks. Through extensive simulations and real-world case studies, the thesis demonstrates how graph-based optimization can lead to significant improvements in the efficiency of electric bus operations, reducing overall energy usage and enhancing service levels. In this work five graph-based algorithms are proposed for optimizing fleet size and energy consumption, clustering charging stations for minimum impact on electrical grid, and assessing the feasibility of optimization methods for the existing dispatching plan. Amongst them, two P-class algorithms are for dispatching and charging plans which are proved that can solve the problem and their outputs are certainly optimum. The algorithms are tested for three metropolises: New York City in USA, Manchester in UK and Melbourne in Australia at urban scale. All case studies show successful application of the algorithms for these cities. Furthermore, this research contributes to the development of sustainable urban transport strategies by providing insights into the scalability and adaptability of electrified bus networks in different urban contexts. The findings offer valuable guidance for policymakers, transit authorities, and urban planners in their efforts to create greener, more efficient public transport systems. Ultimately, the thesis underscores the potential of graph-based optimization as a powerful tool for advancing the electrification of public transport, supporting broader environmental and societal goals.