بهينه يابي دكل هاي انتقال نيرو به كمك الگوريتم ژنتيك و شبكه هاي عصبي

OPTIMAL DESIGN OF TRANSMISSION TOWERS VIA HYBRID GENETIC ALGORITHM AND NEURAL NETWORKS

Transmission towers are the most important part of the systems for transferring electrical power and for its dis¬tribution. These structures are designed to support the conductors and ground wires of electrical power lines and withstand mechanical forces in the worst, weather and under the worst atmospheric conditions. A linear anal¬ysis is generally sufficient for the analysis of a transmis¬sion tower. Since these structures contain a high number of elements, and due to their high cost, their optimiza¬tion is of great importance from both a material and a construction point of view. Transmission towers are often subjected to many combinations of loading and, in the process of optimization, one is faced with many analyses. In this paper, a genetic algorithm is employed for optimization. One of the main difficulties in using the genetic algorithm for optimization is that, for each chromosome in each generation, one analysis should be performed. This requires the inversion of matrices and, with an increase in the number of elements and in geo¬metric. optimization, the convergence becomes very slow. In order to overcome this difficulty, neural networks are trained as analyzers to take on part of the computa¬tional load. Using neural networks increases the rate of convergence of the optimization. A multi-population genetic algorithm is used in this article.