A multi-objective PSO for DC microgrids: Efficient battery management to minimize energy losses and operating costs

Abstract

This article addresses the challenge of battery energy management in a Direct Current (DC) Microgrid (MG) with distributed photovoltaic generators working at their highest power point. This study adopted a multi-objective approach whose main goals is to reduce energy losses and operating costs. To mathematically model this problem, we used Multi-Objective Mixed-Integer Nonlinear Programming (MOMINLP) that incorporated two conflicting objective functions that minimize — at the same time — the energy losses and operating costs of DC MGs. To solve this problem, we followed a master–slave approach. Two multi-objective algorithms were used in the master stage: the Multi-Objective Particle Swarm Optimizer (MOPSO) and the Multi-Objective Lion Ant Optimizer (MOALO). The slave stage applied an hourly load flow analysis employing the method of Successive Approximations (SAs). To validate our methodology, the 33-bus test feeder was adapted by incorporating specific data on generated power and electricity demand from a specific region in Colombia. These data (obtained from the local network operator) are projections of electricity demand for an average day in the region under analysis. The methodology was implemented in the MATLAB environment, and each algorithm was run 100 times to assess standard deviations, maximum reductions, average reductions, and processing times. The two algorithms were compared to identify which one exhibited the best performance when they solved the problem examined here. The results proved that the MOPSO was the most effective approach in terms of minimizing operating costs and significantly reducing energy losses compared to the benchmark scenario.