AI could boost AC/DC hybrid electricity systems and make renewable-heavy power grids more stable, efficient and resilient, according to research in the International Journal of Global Energy Issues, which has considered the future operation of low-carbon high-voltage networks.
The research looked at one of the main engineering challenges that has emerged with the shift towards renewable energy: how to operate electricity grids reliably when large amounts of power come from intermittent sources such as wind and solar.
Modern electricity systems are increasingly evolving into AC/DC hybrid networks, which combine traditional alternating current (AC) infrastructure, such as power stations, with direct current (DC) systems used by technologies such as solar panels, batteries, electric vehicles and power electronics. Hybrid systems can improve efficiency and make renewable integration easier, but they are also much more difficult to control because both electricity supply and demand fluctuate constantly.
The researchers argue that traditional centralised control systems are no longer appropriate for such networks. Conventional grid management relies on a central operator collecting information from across the network and calculating instructions for generators, storage systems, and other equipment. But the growing number of renewable devices and variables now make real-time optimisation far too slow and computationally complex.
The research has looked at how a framework based on multi-agent reinforcement learning (MARL), a form of artificial intelligence (AI) in which software agents learn decision-making behaviour through repeated interaction with their environment might solve this problem. In this approach, different parts of the electricity system, including wind farms, solar installations, and battery storage units, are treated as independent components where rapid, local decisions and the over-arching system coordinates these decisions within the grid as a whole.
Simulations predict a reduction in operating costs of more than 10 percen and an increase in renewable energy use of more than 13 per cent. Efficiency is also improved, with losses reduced by more than 15 per cent compared with traditional centralised optimisation methods.
Wei, B., Yang, C., Liu, K., Tang, W. and Zhang, X. (2026) ‘Optimal scheduling energy for ‘wind-solarload-storage’ AC-DC hybrid distribution network system based on multi-agent algorithm’, Int. J. Global Energy Issues, Vol. 48, No. 8, pp.24–42.
No comments:
Post a Comment