Smart Microgrid QoS and Network Reliability Performance Improvement using Reinforcement Learning

Abstract

A Smart Microgrid consists of physical and communication layered networks. It provides communication services to each connected component and resource through multi-agent system. This paper proposes a reinforcement learning based methodology, Q-reinforcement Learning based Multi-agent based Bellmanford Routing (QRL-MABR), using multiple agents communicating over the microgrid network. It strengthens the decision-making core of the microgrid by improving Quality of service and network reliability of the smart microgrid. The performance analysis of the algorithm is tested over small-scale IEEE microgrid models i.e. IEEE 9 and IEEE 14. The work is tested and compared with four routing oriented decision-making algorithms, Open shortest path first (OSPF), Optimized link state routing (OLSR), Routing information protocol (RIP) and Multi-agent based Bellmanford routing (MABR). The results validate the productivity and learning capabilities of the proposed QRL-MABR algorithm.