Non-dominated sorting genetic algorithm-ii arrangements for optimal electrical distribution networks reconfiguration

Résumé 0

Radiality constraint typically increases genetic algorithms complexity considering that distribution network reconfiguration is - by nature - non-differentiable, mixed integer and highly complex combinatorial. Genetic Algorithms’ potentialities can be seized when coding is enough efficient. Paper presents non-dominated sorting genetic algorithm-II arrangements, implemented in MatLab to solve optimal electrical distribution networks reconfiguration. Initial population is created randomly by using a heuristic approach and genetic operators for generating feasible individuals in every genetic evolution stage which besides are adapted with graph theory help for transforming initial population's infeasible individuals that not satisfied radiality constraint, as well as for avoiding new infeasible individual’s generation and thus, sidestepping the boring mesh check and reducing search space and computational burden. Unlike previous authors that employed an integer-composed chromosome that directly selects a single branch to open from each fundamental loop vector as well as modified genetic operators, the algorithm’s arrangements and a new coding used that ensures chromosome viability for all gene values has been the employed approach. That way, crossover and mutation do not need special genetic operators. Proposal efficacy has been tested with 33-bus and 70-bus test systems and results are promising and better than precedent proposals.