Transmission expansion planning using neuro-computing hybridized with genetic algorithm

The aim of power transmission planning is to determine which right-of-way to construct new lines in order to meet the future load in the most economical way. This problem has been solved by mathematical programming techniques, which require considerable computational efforts, or by successive planning based on sensitivity analysis, which find a single nonoptimal solution. Although another method that has efficiency for combinatorial problems is neuro-computing, this approach obtains poor solutions while it saves computational effort. The most desirable approach for this planning problem can find many good solutions in a reasonable time, because experts of planning will easily plan economical and reliable expansion according to these solutions by comparison with each other. This paper presents an approach for solving power transmission expansion planning based on neuro-computing hybridized with genetic algorithms. This approach generates suitable initial states, which include past information, of neural networks utilizing genetic algorithms. Mingling neuro-computing and genetic algorithms, the proposed approach can find many good solutions in a reasonable time making full use of their merits. Computational examples show the effectiveness of the proposed approach by comparison with conventional techniques