Coordinated electric vehicle control in distribution networks towards multi-service provisions: A transformer learning-based risk management strategy

Due to the trend of transport electrification, electric vehicles have garnered significant interest in supporting carbon-aware distribution network operations via coordinated charging and discharging behaviours, owing to their mobility and flexibility in managing complex power network environments. However, given the inherent characteristics of distribution networks exhibiting a high proportion of renewable penetration, potential system security challenges from voltage fluctuations and various uncertainties are highlighted. Therefore, this paper proposes a risk-based carbon and electricity co-optimisation problem for a distribution network integrated with electric vehicles to investigate the benefits of electric vehicles in multi-service provisions, including energy management, carbon service, and voltage regulation. A data-driven risk-management framework is developed to mitigate the impact of uncertainties associated with RESs, load, and price signals. Specifically, an attention-based Transformer model is proposed to predict the uncertainty sets. Then, a two-way risk-based Min–Max optimisation framework is proposed to use prediction values and simulate both risk-aversion and risk-seeking decision processes. Case studies are conducted in the IEEE 15-bus radial distribution network to assess the behaviour of distribution network operators with different risk profiles under various levels of uncertainty. Finally, the scalability is examined by integrating large numbers of electric vehicles within the distribution network.