Energy optimization in the energy internet

Abstract

This thesis aims to bridge the gap between energy routing and P2P energy trading by developing a system that integrates both the economic and physical dimensions of energy exchange. The objectives are to optimally match prosumers, maximize producer profits, minimize consumer costs, and ensure efficient, reliable, and collision-free energy transfers. To achieve these objectives, several models and algorithms are proposed. First, the trading problem is formulated as a fractional knapsack problem and solved using a greedy algorithm combined with Dijkstra’s shortest-path algorithm. Second, simulated annealing is applied to producer subset determination, demonstrating superior convergence and ability to escape local optima compared to other heuristic optimization approaches proposed in the literature. Third, power loss is incorporated into path optimization through a modified greedy search algorithm that outperforms traditional shortest-path methods. Fourth, a quantum genetic algorithm is employed to pair prosumers, accounting for both costs and physical losses, thereby significantly reducing computation time while improving efficiency. A dynamic scheduling mechanism is then introduced to mitigate congestion, prevent collisions, and enhance fairness and reliability. Finally, an adaptive multi-commodity flow (MCF) framework with Mirror Descent learning is developed to simultaneously address all three routing challenges in a unified optimization approach validated on a real-world dataset of 300 Australian households over 2,648 trading hours. The proposed framework demonstrates significant improvements in cost-effectiveness, transmission efficiency, and system reliability. Results show up to 39.34% cost reduction depending on infrastructure capacity, transmission losses maintained below 1.2%, 55.9% grid independence, and sub-millisecond optimization enabling real-time market operation. The adaptive approach eliminates manual parameter tuning through online learning, automatically identifying binding constraints and adjusting routing priorities across diverse operating conditions. By unifying economic and physical aspects of energy transfers at scale, this thesis contributes a novel foundation for sustainable, decentralized, and intelligent energy markets.