Promote Data Sharing in Integrated Power-Traffic Networks: A Coalition Game Approach

Abstract

Accurately estimating spatial-temporal electric vehicles' (EVs) charging demands is crucial for the secure and economic operation of power systems. At present, the distribution system operator (DSO) relies on historical data collected at each charging station to estimate future EV charging demand. However, the station-level forecast disregards EVs' spatial correlations within traffic networks (TNs) and might suffer significant forecast error, forcing the DSO to make conservative scheduling at the expense of operation economics. To this end, this paper proposes to leverage cross-sector information (i.e., traffic demand data and network parameters in TNs) to enhance forecast accuracy and avoid over-conservative operations. To facilitate the data sharing among the DSO and TN data holders (i.e., traffic authority and navigation App. companies), we adopt the Coalition Game theory to uncover how these entities could cooperate to benefit each other, and to fairly allocate the extra profits (i.e., the operational cost reduction induced by the improved forecasts) among themselves. The conditional value-at-risk theory is adopted to model the risk-averse behavior of the DSO. In case studies, we reveal the non-negligible impact of TN condition variations on EV charging distributions. Moreover, numerical results show that sharing high-quality traffic data contributes to the reduction in DSO's operating cost by utmost 20.8% as compared to the current practice without data sharing.