Relaxation of non-convex problem as an initial solution of meta-heuristics for energy resource management

Energy resources management is an important topic in the context of Smart Grids (SG) and MicroGrids (MG). Virtual Power Players (VPP) emerge as aggregator entities responsible for managing energy resources in SG and MG. In this paper modern meta-heuristics are compared with their hybrid version that uses a deterministic method as initial solution. The meta-heuristics implemented are Differential Search Algorithm (DSA) and Quantum Particle Swarm Optimization (QPSO) to solve the hard combinatorial scheduling problem. The deterministic method employs a relaxation of the non-convex Mixed Integer Non-Linear Programming (MINLP) formulation in order to reduce the computational burden and the number of iterations in metaheuristics. The implemented methods are tested with a 33-bus distribution network, 1800 Electric Vehicles (EV), 15 fixed storage, 66 Distributed Generation (DG), 10 energy suppliers and 1 utility-scale wind generation facility. The results show the excellent performance of the DSA, QPSO and hybrid approaches.