AI-Based Control for Dynamic Plug-and-Play Battery Energy Storage Systems: The Way Forward for Vehicle-to-Grid

Pursuing sustainable and efficient energy solutions, a team of researchers from Teesside University, Mudhafar Al-Saadi and Professor Michael Short, has proposed an innovative multiagent AI-based control system for plug-and-play batteries in DC microgrids. The methods and experimental studies, published in the journal Batteries and several IEEE papers and articles, outlines a promising approach to enhance the management of power-storage flow in microgrids, particularly addressing the challenges posed by dynamic vehicle-to-grid (V2G) charging applications.

Energy is fundamental to the creation and sustenance of life. The primary motivation behind this research is the urgent need for sustainable, low-emission energy systems for applications in industry, society and transportation. These systems aim to tackle climate change and fossil-fuel scarcity through decarbonization, digitalization, and decentralization of electrical power systems. Traditional power-flow management techniques often fall short in handling the dynamic and decentralized nature of modern power distribution networks. The innovative approach proposed by Al-Saadi and Professor Short offers a solution by leveraging multiagent reinforcement learning (MARL), an emerging AI-based technique for improving automated decision-making by learning complex input-output relationships, to improve the efficiency and reliability of power storage in DC microgrids.

Mudhafar Al-Saadi explained, “The influence of DC infrastructure on the control of power-storage flow in microgrids has gained significant attention. Our research aims to address the potential loss of charge-discharge synchronization and the subsequent impact on control stabilization.” 

The researchers highlight that efficient power-flow management is crucial for integrating renewable energy sources and ensuring the sustainability of decentralized power networks. One of the key challenges in this domain is the accurate synchronization of the charge and discharge cycles of batteries, which is often compromised under real environmental conditions, most prominently due to sustained high load variations coupled with battery heterogeneity and degradation, uncertain power system topology due to dynamic plug-in/plug-out insertions and removals of battery elements, infrastructure influences, and environmental (temperature) influences. The proposed multiagent-based control system compensates for these variations in real-time, ensuring a balanced and stable power flow.

In their experiments, the researchers demonstrated significant improvements in various performance metrics. The proposed system achieved reduced convergence times, enhanced output-voltage balance, reduced power consumption, and improved power-flow balance. These results underscore the effectiveness of the proposed control system in real-world scenarios, where dynamic load conditions and varying infrastructure influences are common.

“The real-time compensation for DC infrastructure influences with plug-and-play insertions/removals of dissimilar batteries in a larger, aggregated storage system is a key factor in the success of our control approach,” stated Professor Short. “Our system adapts to unknown and time-varying DC infrastructure influences and battery types without the need for initial estimates of key parameters, ensuring the reliability and sustainability of the microgrid.”

“EV batteries and chargers present key storage assets to ensure grid balance and peak shaving in Vehicle-to-Grid (V2G) and Demand Response applications and can aid decarbonisation efforts. Our approach can improve V2G performance and prevent unnecessary degradation of EV battery capacity and lifetime.”

The research team employed a combination of simulation and hardware-in-the-loop studies to validate their proposed system. They used realistic conditions, including day-long continuous variations in load demand and dynamic switching of heterogeneous battery connections, to test the robustness and effectiveness of their control system.

In conclusion, the multiagent-based control system proposed by Al-Saadi and Professor Short represents a significant advancement in the management of power-storage flow in dynamic DC microgrids. By addressing the challenges posed by dynamic load conditions and infrastructure variations, this innovative approach offers a practical and efficient solution for modern power distribution networks. The findings from this study pave the way for further research and development in the field of decentralized energy systems, contributing to the global efforts towards sustainable and low-emission energy solutions.