Multi-objective optimization based power management strategy for smart grid with distributed energy resources

Abstract

This study presents a power management method for smart grid that incorporate renewable energy systems (RESs) and energy storage systems (ESSs), utilizing multi-objective optimization. Power management not only reduces peak load and electricity cost, but also improves energy supply reliability. In this thesis, firstly, a peer-to-peer (P2P) power trading algorithm is introduced for power management of multiple nanogrids. P2P power trading enables participants to engage in direct power transactions, leading to cost savings for power consumption and the reduction of carbon emissions by promoting the utilization of RESs and ESSs. P2P power trading is executed based on cooperative game theory with the aim of increasing the benefits for both sellers and buyers. Furthermore, in order to maximize the battery's lifespan with an efficient charging/discharging strategy, a battery aging model or degradation cost model is considered within the context of multi-objective optimization. Secondly, we propose a P2P power trading algorithm for power management of nanogrid clusters. The scale of power transactions between nanogrid clusters is relatively larger than that of individual nanogrids. Although ESSs are not utilized, higher capacity of photovoltaic systems are applied to each nanogrid cluster for P2P power trading. By integrating predicted power data based on a deep learning (DL) network into the power trading algorithm, the efficiency of P2P power trading is improved. Thirdly, power management of microgrid integrated with electric vehicle (EV) charging station is presented. While multiple EVs can be charged simultaneously at EV charging station, excessive charging demand may lead to grid instability issues. In the proposed EV charging/discharging scheduling algorithm, the maximum discharging power of parked EVs is determined based on the measured power data. Charging/discharging scheduling of EVs and power management of microgrid are performed in consideration of the maximum discharging power of EVs. The use of DL network incorporates forecasted power data into the EV charging/discharging scheduling algorithm, further enhancing the economic feasibility of EV charging stations.