Constrained Robust MultiObjective Optimization for Reactive Design in Distribution Systems

A Modified Genetic Algorithm for Optimal Allocation of Capacitor Banks in MV Distribution Networks

“…A new formulation including the robustness of the solution of a constrained multiobjective design of reactive power compensation is presented by Augugliaro et al [3]. The issue of robustness is included due to uncertainty and errors in loads estimation.…”

“…The crossover-like operators are the following: (9) Module crossover (Fig. 10) Two individuals and a set of modules are randomly chosen, between the two individuals the modules of the set are exchanged; in this way, the two new individuals have the same spatial distribution of compensated nodes (with a possible increase or reduction of the compensated nodes, but satisfying the constraint on the maximum number of compensated nodes) 3 ; the total reactive power of the two offspring solution changes, compared to the parents. (10) Position crossover (Fig.…”

“…(10) Position crossover (Fig. 11) Two individuals are randomly chosen as well as an id between the possible 3 The operation executed on the matrices of the parents consists in the identification of a horizontal row relevant to a given number of modules. Each of the two offspring is attained building a matrix starting from one of the parents and putting in a sequence, in the missing parts of the columns, the modules that are in the other row; if the two parents show a number of compensated nodes equal to the maximum number, the two offsprings will therefore have a number of nodes that is equal or smaller; if one or both parents have a number of compensated nodes that is lower than the maximum, the two offsprings have a number of nodes that is smaller or equal to the maximum.…”

NSGA-II with local search for a multi-objective reactive power compensation problem

“…In [36], an Elitist Non-dominated Sorting Genetic Algorithm (NSGA II) is presented, including solutions robustness analysis considering that loads are uncertain in distribution systems and their estimation is often affected by errors. In [31], a NSGA II based approach is also proposed to handle a two objective model: an objective function is related with the return on investment, which combines monetized values for the energy and peak power losses and the cost of capacitor banks, and the other objective function is related with the voltage drop.…”

“…In [31], a NSGA II based approach is also proposed to handle a two objective model: an objective function is related with the return on investment, which combines monetized values for the energy and peak power losses and the cost of capacitor banks, and the other objective function is related with the voltage drop. These approaches are applied to a distribution system with 15 nodes [36], and to a medium-large electrical network with 38 load nodes [31].…”