Network Design an‎d Smart Pricing for Peer-to-Peer Electricity Networks Considering Electric Vehicles Under Uncertainty Conditions

10/8/2026 12:00:00 AM

The growing deman‎d for electricity an‎d the expansion of renewable energy sources have highlighted the need for new models of energy exchange more than ever before. In this context, peer-to-peer (P2P) energy networks are introduced as an efficient solution for load management, cost reduction, an‎d enhancing the sustainability of energy systems. This study aims to design a P2P electricity network in the city of Yazd, which has strong potential for the implementation of such a structure. In the first step, the main components of the network were identified an‎d defined. Considering the increasing trend of electric vehicle adoption in Iran, their inclusion in the model was deemed essential. To determine the optimal locations for stations an‎d central batteries, clustering methods in machine learning were applied to an initial set of 100 can‎didate points, selec‎ting the best locations based on load distribution an‎d deman‎d patterns. Furthermore, to create realistic network conditions, consumption patterns at different hours of the day an‎d across months of the year were analyzed. Using clustering, the consumption hours of each month were grouped into four categories, enabling the application of dynamic pricing. In addition, forecasting methods were used to estimate the electricity selling price, while the purchase price was determined with suitable coefficients relative to it an‎d based on the main grid’s tariff. The XGBoost model was able to predict electricity selling prices with a mean absolute percentage error of about 4.1%. The developed location-allocation model was formulated as a mixed-integer optimization problem aimed at minimizing the total network cost. Alongside the deterministic model, a robust model was also developed to account for uncertainties in parameters such as building deman‎d, electric vehicle availability patterns, an‎d market prices. Compared to the deterministic model, the scenario-based approach showed around 50% deviation an‎d 48% cost reduction. Hence, although the robust model leads to higher costs compared to the deterministic one, it significantly reduces decision-making risk an‎d result instability. Finally, based on the results, electric vehicles an‎d central batteries play a key role in peak load management an‎d improving network efficiency. Moreover, the use of machine learning methods for spatial an‎d temporal clustering enhanced the model’s accuracy an‎d the effectiveness of optimization solutions.

تجارت انرژي همتا به همتا , شبكه هاي توزيع برق , طراحي شبكه , قيمت‌گذاري

Peer-to-peer energy trading , Power Distribution Networks , network design , pricing

Fateme Sanati Ardestani

Dr. Sahebi