Scaling Up Cooperative Game Theory-Based Energy Management Using Prosumer Clustering

Game theory based energy sharing schemes emerged in recent years to incentivize efficient management of the increasing amount of distributed energy resources. Among these, cooperative game theoretic schemes provide detailed financial incentives on the individual prosumer level. The nucleolus, a mechanism to allocate these financial incentives, has been proven to guarantee the prosumers' willingness to participate. However, the computation time of the nucleolus increases exponentially with the number of participants, strictly limiting the size of this scheme. This study proposes to incorporate clustering techniques to estimate the nucleolus at reduced computation times, where a novel marginal contribution profile is used as the clustering features. A stratified random sampling based approach is formulated to evaluate the estimation performance, showing that the proposed method is able to scale up the cooperative energy management scheme from less than 15 players to over 100 players while maintaining high accuracy of the nucleolus estimation.

Automated summary

A novel approach is proposed in this study to estimate nucleolus by incorporating clustering techniques, reducing computation time significantly. A stratified random sampling method is formulated to scale up the cooperative energy management scheme from less than 15 players to over 100 players while maintaining high accuracy of the nucleolus estimation.