A Model Predictive Control framework is proposed for optimizing power allocation in Dynamic Inductive Charging systems using edge computing and vehicular communications, prioritizing critical battery states. The framework addresses two key limitations of uncoordinated allocation: suboptimal resource utilization during saturation and unfair distribution of power under scarcity. The open-source framework and simulation tools are released to support further research in this emerging domain.

The incorporation of fairness into distribution network planning and operation aims to achieve social cohesion through fair outcomes, with a cost known as the price of fairness (PoF). Fairness encompasses various notions and metrics, including egalitarian and merit-based criteria, which can have different mathematical complexities. The study reviews these fairness notions and metrics to support consistent and transparent planning and decision-making in distribution network operations.

A cutting-plane method is proposed for efficient graph partitioning with controllable edges in distribution grids to enable real-time reconfiguration while ensuring radial connectivity and resource feasibility. The framework integrates graph-theoretic constraints, resource flow, and network structural properties to enforce an operational hierarchy, achieving median and best-case speedups of 57.5x and over 64x respectively. The approach ensures systematic elimination of infeasible configurations while maintaining distributed controllability.

A regime-adaptive weighted ensemble learning algorithm is proposed for computing-driven dynamic load forecasting in AI data centers, which significantly enhances load forecasting accuracy and adaptivity across various regimes. The method uses a novel feature engineering strategy and dynamically optimizes ensemble weights to facilitate adaptive calibration of inter-submodel contributions. This approach achieves minute-class forecasting errors below 1%, making it suitable for grid-interactive coordination and demand response.

Orbital data centers are being evaluated as solar-powered compute constellations with multiple constraints affecting their feasibility. For a representative 1 MW high-sunlight anchor, the base case yields a photovoltaic area of $5.64 \times 10^3 {\rm m}^2$, radiator area of $2.50 \times 10^3 {\rm m}^2$, and a mass range of 34-59 kg/kW. The combined launch and spacecraft-build cost is significantly lower than the current public Falcon 9 launch price, making space-native preprocessing and communications-integrated edge compute credible early regimes.