A Multi‐Dimensional Multi‐Scenario GAN Framework for Assessing EV‐Induced Power Quality Degradation in Distribution Networks

The rapid integration of electric vehicles (EVs) into distribution networks poses significant power quality challenges, particularly in terms of harmonic distortion and voltage deviation. This paper proposes a novel multi‐dimensional, multi‐scenario generative adversarial neural network (MM‐GAN) to model the probabilistic behavior of EVs, simultaneously capturing the correlations between stochastic arrival/departure times and energy demands. This approach presents a key novelty by not only enhancing modeling accuracy but also achieving a drastic reduction in computational burden compared to conventional Monte Carlo method. The MM‐GAN framework generates correlated EV load profiles simultaneously in three scenarios, significantly improving the realism and accuracy of power quality assessments. The model is tested on the IEEE 33‐bus distribution network and validated against Monte Carlo simulations using an enhanced backward‐forward sweep harmonic power flow analysis. Simulation results reveal that EV penetration levels of 14%–45% cause a 12%–132% increase in total harmonic distortion (THD), 5%–10% higher voltage deviations and 28%–70% greater power losses. These results underscore the critical need for active mitigation strategies to address power quality challenges posed by rising EV adoption. The proposed framework provides distribution network operators with an advanced tool for realistic EV impact assessment and proactive power quality management.