Sistema Distribuido RNA-AG para la Optimización de Patrones de Diseño en Vigas de Concreto Armado

Published in: Industry, Innovation, and Infrastructure for Sustainable Cities and Communities: Proceedings of the 17th LACCEI International Multi-Conference for Engineering, Education and Technology
Date of Conference: July 24-26, 2019
Location of Conference: Montego Bay, Jamaica
Authors: Jose Sulla-Torres (Universidad Nacional de San Agustín de Arequipa, PE)
Paul Ventura Acero (Universidad Nacional de San Agustín de Arequipa, PE)
Justo Saico Saico (Universidad Nacional de San Agustín de Arequipa, PE)
Richard Tumailla Sanchez (Universidad Nacional de San Agustín de Arequipa, PE)
Hector Concha (Universidad Nacional de San Agustín de Arequipa, PE)
(Universidad Nacional de San Agustín de Arequipa)
Full Paper: #169


In this article, a system capable of carrying out the design of a reinforced concrete beam subject to normal bending and cutting is constructed, which provides results quickly and safely, using two combined techniques of Artificial Intelligence (AI), Artificial Neural Networks ( RNA) and Genetic Algorithms (AG), and a distributed Master-Slave model (MDME) to calculate the optimal weights of the RNA using distributed parallelism, implemented in Java with Remote Method Invocation (RMI), Sockets and Threads. The purpose is to obtain an RNA capable of coherently associating data used for the design of the beam (beam cant, reinforcement steel area and abutment spacing) with the empirical design patterns of similar reinforced concrete beams, the Professional School of Civil Engineering (EPIC) of the National University of San Agustin (UNSA) of Arequipa. As a result, a comparative graph of the convergence times of the proposed RNA is shown, concluding its superior efficiency in terms of speed. For example for populations close to one million, the time is acceptable, and the error is less than 1%.