AG Kommunikationstheorie


Design of a minimum-cost reliable communication network for power distribution grid


This thesis investigates the design of heuristic algorithm approach for designing a reliable communication network. Related network should provide a sound base for drafting a communication model in the smart grids. Due to its complexity and computational time requirements of NP-hard problems, a genetic algorithm (GA) is considered as a viable option. In the first part of our work, the ILP formulation is tested via integer linear programming software tool, which is used as a benchmark for the analysis of the problem based on designed heuristic algorithm. Since the considered problem is a constrained problem, the formulation is redesigned by the interior penalty method. On the basis of redefined formulation, a customized GA is designed where the binary encoding is used to represent the optimization variables. The half of the generated cus- tomized solutions is evaluated by the whole fitness function, while the other half is created by Kruskal algorithm and it is evaluated by constraint function alone. By generating the spanning trees, it becomes easier to explore better solutions, as well as, by evaluating constraints in order to provide better convergence towards the optimality. Then the most fitted solutions are further proceeded to the crossover and mutation op- erations. The multi-parent crossover operation is applied to the selected solutions so that considerably better solutions can be produced. Then the resulted ospring from the crossover is further muted pairwise. The customized GA is evaluated over different sizes of fully connected test networks. It has been observed that the designed GA performs optimally with less computational effort in comparison to standard GA. Moreover, with the increase of the population size the solution is going close to the optimum value.

zurück zur Terminübersicht