Optimization Design of Laminated Functionally Carbon Nanotube-Reinforced Composite Plates Using Deep Neural Networks and Differential Evolution

Abstract

This study presents a new approach as an integration of deep neural networks (DNN) into differential evolution (DE) to give the so-called DNN-DE for frequency optimization of laminated functionally graded carbon nanotube (FG-CNT)-reinforced composite quadrilateral plates under free vibration. In the presented approach, the DNN is applied to predict the objective and constraints during the optimization process instead of using the time-consuming finite element analysis (FEA) procedures while the DE is used as an optimizer for solving the optimization problem. Several numerical examples are performed to illustrate the performance of the proposed method. Optimal results obtained by the DNN-DE are compared with those achieved by other methods in order to show the reliability and effectiveness of the proposed methodology. Additionally, the influence of various parameters such as the boundary condition, the carbon nanotube (CNT) volume fraction, the CNT distribution on the optimal results is also investigated. The obtained results indicate that the proposed DNN-DE is an effective and promising method in solving optimization problems of engineering structures.