Energy drive and management of smart grids with high penetration of renewable sources of wind unit and solar panel

This paper proposes a novel reinforcement learning based energy drive and management in smart grids incorporating the uncertain behavior of the electric vehicles and renewable energy sources. To this end, a novel stochastic framework based on evolving point approximation is devised to provide the most optimal power dispatch and minimize the total operation costs. In order to predict the output power of the renewable energy sources of wind unit and solar panel, the proposed approach uses Q-learning technique. This method enhances the prediction of conventional models such as neural networks. Due to the high complexity and nonlinearity of the final optimization framework, a new optimization approach based on dragonfly is devised. Moreover, a novel three-phase correction is introduced to help improving the quality of the final solutions and escape from the optima. The effect of charging/discharging of electric vehicles on the optimal energy management of the smart grid is assessed in two different scenarios. The performance of the proposed model is examined on an IEEE smart grid system.

Automated summary

This paper presents a novel reinforcement learning based approach for energy drive and management in smart grids, utilizing Q-learning technique and dragonfly optimization algorithm for power dispatch optimization and cost minimization.