A smart hierarchical transactive energy system in the presence of renewable energies, and demand-side management

With the constant development of energy systems, multi-energy networks are becoming increasingly popular, which integrates renewable energies and demand-side management. This presents a significant challenge for developing smart energy management frameworks. Furthermore, unlike traditional energy systems, the transaction energy system is dynamic and complex, which is enriched by the interdependence among multiple energies, the uncertainty of renewable energies and the complexity of demand-side management. Therefore, advanced modelling techniques and solution methods are required to be developed to overcome the difficulties. This thesis addresses the intricate challenges of developing a smart hierarchical transactive energy system that seamlessly marries multi-energy sources, renewable energy integration, and sophisticated demand-side management strategies. The research unfolds in three pivotal research topics: 1) The formulation and analysis of a game-theoretic decision-making model for energy retailers’ strategic bidding and offering in both wholesale and local energy markets while considering customers’ switching behaviour; 2) The introduction of a customised multi-energy pricing scheme, which is formulated as a bilevel optimisation model. The proposed model not only maximises the profit of energy retailers but also considers the multi-energy interdependencies and the diverse characteristics of microgrids; 3) The development of an innovative forecasting model named Patchformer, based on Transformer-based architectures and patch embedding method, for the prediction of long-term multienergy loads. This model improves forecasting accuracy, which enables a more reliable and efficient energy system management by predicting energy demands with high precision. This work presents a comprehensive approach to improving the effectiveness of transactive energy systems by merging advanced modelling techniques and machine/ deep learning models. This thesis tackles the current challenges in the field of transaction energy systems while also providing information for future research that aims to unlock the full potential of smart energy management in smart grids.