Congestion Based Dynamic Pricing for Public EVs Charging

The increasing adoption of electric vehicles (EVs) presents both an opportunity and a challenge for modern power systems. While EVs offer potential as flexible, controllable loads that can support grid stability, their uncoordinated charging patterns risk aggravating local congestion, particularly at the distribution level. This thesis proposes a congestion-based dynamic pricing framework designed to mitigate stress on the power grid while maintaining fairness among EV users. The model integrates predictive analytics with optimization techniques to generate real-time, location-aware price signals for EV charging. A Bayesian Ridge Regression (BRR) model is employed to forecast short-term charging demand using synthetic datasets informed by urban mobility and traffic patterns. The predicted demand is fed into a pricing engine that minimizes a joint objective function accounting for grid congestion and pricing fairness, under operational constraints such as price bounds and user flexibility. The methodology is validated through a case study involving two public charging stations in the Helsinki region, simulating both urban and residential demand conditions. Results show that the proposed model effectively reduces peak load by up to 20% and improves revenue by 12% compared to flat pricing strategies. The pricing mechanism is shown to be adaptive, equitable, and responsive to real-time congestion while remaining within predefined fairness bounds. This work demonstrates that data-driven dynamic pricing can serve as a viable and scalable demand-side management solution, especially for distribution system operators seeking to manage growing EV loads without extensive infrastructure investment. Limitations and regulatory challenges are acknowledged, and future directions for expanding model realism and implementation feasibility are outlined.