A Preliminary Study on Collusion Detection In Transactive Energy Systems

With deeper penetration of distributed energy resources, a need for decentralized control and economic operation has initiated the concept of transactive energy systems. As this framework is still under discussions among policymakers, there are some potential issues regarding the mitigation of market power and collusive behavior of market participants. In repeated transactions, a distribution system has favorable conditions to form tacit collusion. This work proposes a collusion detection (screening) tool for distribution system operators. This detection tool uses maximum likelihood estimation and simulated-annealing based Q-learning algorithm. The preliminary results indicate that collusive prices at the transition period do not follow cost price changes and at the stationary period change very slightly with changes in cost price changes. The results show the change with season varies the profits with collusive pricing as well as Nash equilibrium prices, such as profits are higher in winter months compared to the profits of summer months.