21 April 2025 in Energy Reports
9 January 2025 in IET Generation, Transmission & Distribution
The integration of renewable energy sources and the increasing demand for reliable power have posed significant challenges in the design and operation of distribution networks under uncertain conditions. The inherent variability in renewable energy generation and fluctuating consumer load demand requires advanced strategies for Distributed Energy Resources (DERs) allocation and sizing to enhance grid resilience and operational efficiency. This article introduces an innovative framework for optimizing distribution network design under these uncertainties. The approach integrates deep learning‐assisted Distributionally Robust Optimization (DRO) with Generative Adversarial Networks (GANs) to dynamically model and manage the inherent variability in renewable sources and demand fluctuations. Employing a combination of nonlinear optimization techniques and advanced statistical methods, the framework robustly optimizes network configurations to minimize losses and improve voltage stability. The model's efficacy is rigorously tested on the IEEE 33‐bus system, achieving a 15% reduction in power distribution losses and a 20% improvement in voltage stability compared to traditional models. Utilizing open‐source computational tools, the method not only boosts operational reliability and efficiency but also adapts effectively to the increasing integration of volatile renewable energy sources. These results underscore the framework's potential as a scalable and robust solution for modern power network design challenges.
22 July 2025 in Sustainable Energy Technologies and Assessments
31 March 2026 in IEEE Transactions on Industry Applications
This paper develops an integrated optimization framework for biodegradable pop-up data centers intended for temporary or extreme environments where conventional computing infrastructure cannot operate. While prior studies address sustainable materials, energy harvesting, or transient electronics individually, existing literature lacks a unified model that simultaneously captures workload dynamics, renewable energy variability, and degradation-driven end-of-life behavior. To close this gap, the proposed architecture combines biodegradable substrates, ambient solar–thermal energy harvesters, and environmentally triggered dissolution mechanisms into a coordinated system that operates autonomously throughout its lifecycle. The framework formulates transient data center operation as a multi-objective optimization problem linking computational performance, harvested energy availability, and degradation-aware lifecycle control. A Non-dominated Sorting Genetic Algorithm III (NSGA-III) schedules workloads, allocates renewable energy, and regulates dissolution timing under fluctuating environmental and material conditions. This enables the system to maintain balanced operation without external power supply or manual supervision. Simulation results confirm the feasibility of self-powered biodegradable AI and blockchain micro-clusters, demonstrating improved energy utilization under constrained conditions, stable performance across variable workloads, and predictable, residue free end-of-life dissolution. These findings establish a pathway toward zero-waste, energy-autonomous transient computing systems for disaster response, military missions, and other short lived deployments.
15 October 2025 in Energy Reports
1 November 2026 in IEEE Transactions on Consumer Electronics
Data centers, the cornerstone of the digital economy, depend on Uninterruptible Power Supply (UPS) systems, where battery health is critical for operational continuity. However, the practical application of data-driven anomaly detection is severely hampered by two interconnected challenges: distribution shifts between different battery batches degrade model generalization, while data silos arising from commercial and privacy constraints prevent cross-institutional data aggregation. To overcome these barriers, we systematically formulate this industrial challenge as a Federated Unsupervised Domain Adaptation (FUDA) task, which enables collaborative model training without sharing raw data. The core contribution of this paper is the proposal of a novel scenario-driven adaptation strategy, which posits that no single algorithm is universally optimal. We demonstrate that for homogeneous scenarios with small domain gaps, a lightweight discrepancy-based method (CORAL) paired with a powerful feature extractor (Transformer) is most effective. Conversely, for heterogeneous scenarios with large domain gaps, a stable, decoupled adversarial approach (ADDA) combined with a capacity-constrained feature extractor (DeeperMLP) is superior as it ensures convergence and promotes generalization. Experiments on a real-world data center dataset validate our FUDA approach (F1-score of 0.872) not only significantly outperforms local non-adapted models but also achieves performance highly comparable to, and in some cases exceeding, that of privacy-violating centralized models (0.908).