Artificial Intelligence in Designing Non Geometric Brace Systems

In this paper, a new method for designing a particular brace system is proposed. This type of brace system, which is called Non Geometric Brace system are mostly used in seismic area and it allows architects to have more opening in the panel. Non-straight diagonal member of this system introduces eccentricity and it is connected to the corner of the frame by a third member.In designing this system, designers often use trial and error method to locate the connection point of the braced elements considering various parameters affecting the design such as opening and frame dimensions, cross section areas of brace elements and the location of brace element connection. Hence finding the best connection point with maximum stiffness and minimum weight of brace elements with conventional method is not trivial. In this paper, a multi-object genetic algorithm is proposed in determining the best selection for connection point. Equations of feasible area are determined, followed by utilization of some operators such as selection, mutation, crossover and elitist genetic algorithm. Based on the plain aggregation approach for transforming the objective vector in scalar, some modification are proposed considering the priority between the frame stiffness and structure´s weight in this transformation.