Optimization of distribution networks using evolutionary algorithms
Avilés Arévalo, Juan Pablo
MetadataShow full item record
One of the biggest problems that a distribution network (DN) must face is the constant increase in load demand, which eventually will cause the degradation of its optimal operation. To overcome these challenges the distribution network usually is oversized or reinforced, however, although this is a quick and practical solution, it is not necessarily the most economical and efficient one. For this reason, it is desirable to implement an optimization algorithm to improve the network without increasing investment costs. The optimization of a power distribution network is not an easy task, because this is the most extensive part of the entire electrical system. Due to this extension, along with the high complexity of the topology, and some quality parameters that must be respected, the entire design or improvement of a distribution network can be considered as an extremely hard combinatorial, non-convex, and non-linear optimization problem, difficult to solve by conventional methods. For these reasons, we propose a Two-Stage Multiobjective Evolutionary Approach (TS-MOEAP) capable to design and optimize distribution networks, at primary and secondary levels. Due to the complexity of the optimization problem, the approach is implemented in two stages, that can be summarized as follows: Stage-1. Optimal placement and sizing of generation units, as well as optimal branch routing and conductor sizing. For this purpose, an Improved Particle Swarm Optimization technique (IPSO) combined with a greedy algorithm is introduced. Stage-2. Optimal network reconfiguration. For this, an Improved Nondominated Sorting Genetic Algorithm with a Heuristic Mutation Operator (INSGA-HO) is presented, aiming at minimizing the total power loss and investment cost of the system. Finally, to complement the optimization process, the software DER-CAM will be used to find optimal investment solutions for Distributed Energy Resources (DER). Both algorithms are successfully applied to design and optimize real distribution networks that presented several problems, concluding that the combination of these approaches -network reconfiguration with optimal installation of DERs- can converge toward better configurations than other algorithms.
The following license files are associated with this item: