Optimal Location and Sizing of Distributed Generators in DC Networks Using a Hybrid Method Based on Parallel PBIL and PSO

Luis Fernando Grisales-Noreña,Alberto-Jesus Perea-Moreno,Quetzalcoatl Hernandez-Escobedo,Carlos Andrés Ramos-Paja,Oscar Danilo Montoya

doi:10.3390/electronics9111808

Abstract

This paper addresses the problem of the locating and sizing of distributed generators (DGs) in direct current (DC) grids and proposes a hybrid methodology based on a parallel version of the Population-Based Incremental Learning (PPBIL) algorithm and the Particle Swarm Optimization (PSO) method. The objective function of the method is based on the reduction of the power loss by using a master-slave structure and the consideration of the set of restrictions associated with DC grids in a distributed generation environment. In such a structure, the master stage (PPBIL) finds the location of the generators and the slave stage (PSO) finds the corresponding sizes. For the purpose of comparison, eight additional hybrid methods were formed by using two additional location methods and two additional sizing methods, and this helped in the evaluation of the effectiveness of the proposed solution. Such an evaluation is illustrated with the electrical test systems composed of 10, 21 and 69 buses and simulated on the software, MATLAB. Finally, the results of the simulation demonstrated that the PPBIL–PSO method obtains the best balance between the reduction of power loss and the processing time.

Highlights

This paper proposes a methodology for the optimal location and sizing of distributed generators (DGs) in direct current (DC) grids, which is performed under peak load conditions in order to determine the optimal process of the integration of the DGs, thereby enabling the reduction of the power loss associated with the transport of energy on the grid in the worst operation scenario
If we observe the results obtained by the solution methods with respect to the objective function, we can appreciate that the methods that employ the Population-Based Incremental Learning (PPBIL) in the master stage achieve the best results in general terms, while the methods that use the parallel version of the Monte-Carlo algorithm (PMC) and the Genetic Algorithm (GA) present similar results
A hybrid method, formed with the PPBIL and the Particle Swarm Optimization (PSO) algorithms was proposed for the optimal integration of DGs in DC grids

Summary

Introduction

DC networks exhibit a simpler process of analysis and operation, compared to AC networks, as the frequency and the reactive power are not present in the mathematical analysis [8,9]. In this sense, the problem of power flow on DC networks has been studied by using different numerical methods, such as Gauss–Jacobi, Gauss–Seidel and Newton–Raphson [10], linear approximations [11], semi-definite and second order cone programming formulations [12,13] as well as convex quadratic models [14,15]

Methods

Results

Conclusion