Smart microgrids (SMGs) have emerged as a key solution to enhance energy management and sustainability within decentralized energy systems. This paper presents SmartGrid AI, a platform integrating deep reinforcement learning (DRL) and neural networks to optimize energy consumption, predict demand, and facilitate peer-to-peer (P2P) energy trading. The platform dynamically adapts to real-time energy demand and supply fluctuations, achieving a 23% reduction in energy costs, a 40% decrease in grid dependency, and an 85% renewable energy utilization rate. Furthermore, AI-driven P2P trading mechanisms demonstrate that 18% of electricity consumption is handled through efficient decentralized exchanges. The integration of vehicle-to-home (V2H) technology allows electric vehicle (EV) batteries to store surplus renewable energy and supply 15% of household energy demand during peak hours. Real-time data from Saudi Arabia validated the system’s performance, highlighting its scalability and adaptability to diverse energy market conditions. The quantitative results suggest that SmartGrid AI is a revolutionary method of sustainable and cost-effective energy management in SMGs.

A novel topology of a modular uninterruptible power supply (UPS) system, based on switched capacitor circuit and wideband gap (WBG) devices, is proposed. This work deals with the development of a multilevel, multiport, single-phase, transformerless on-line UPS with universal input/output voltage rating (120 V~ 240 V) for main utility and (200 V ~ 400 V) for the battery packs. The proposed converter mainly relies on capacitors and switching devices to accomplish the task of energy transfer and storage. Compared to conventional topologies which have multiple power conversion stages, this topology has only one-stage power conversion from AC to DC to AC. The AC output voltage contains five levels and whose magnitude comes with a boost functionality. The circuit also does not require the use of inductors or transformers to operate, enabling this topology to be small and lightweight. Moreover, size, cost and efficiency are improved due to the reduced number of switching devices, passive components and a lack of a need for isolation transformers in the multilevel operation. A complete set of simulation and experimental results based on a 2 kVA lab prototype of the proposed UPS is presented to demonstrate its performance under different operating conditions. A peak efficiency of 98.3% has been achieved.