Optimization of Shear Walls with the Combination of Genetic Algorithm and Artificial Neural Networks

Abstract

Objectives: In this study, the cost of a reinforced concrete structures containing medium ductile shear walls, which were similar in plan and height, has been optimized by the combination of Genetic Algorithm and Artificial Neural Networks. Methods: The selected structure was a 13-storey building with four types of shear walls, which was supposed to be built in Tehran. The costs estimated for this research included the cost of concrete and reinforcement used in the walls. By changing the measure of shear walls, some steel rods, and cement altered. Results: Accordingly, the ranges of shear walls were recognized as the independent variables and the concrete and steel rod results were dependent variables. Beams, columns and the width of the shear walls were the same on all types. The prices of concrete and support of shear walls have been used for back propagation neural network training. Before optimization by neural network in the section of neural network training, the Genetic Algorithm steel bar function is defined. The steel bar function included the costs of concrete and steel bars used in shear walls. Conclusion: Afterwards, the best combination of four types of shear walls was selected using Genetic Algorithm. Finally, the optimum shear wall area to the plan area of regular concrete buildings with medium ductile shear walls is presented.