Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic Algorithm
Sandwich structures offer significant opportunities to improve the performance of many industrial applications such as aerospace, automotive, and marine. The design of a composite sandwich structure is often challenging because it is driven by the balance between the weight and cost of these structu...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2024-03-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123024001907 |
| _version_ | 1797247709405511680 |
|---|---|
| author | Mortda Mohammed Sahib György Kovács |
| author_facet | Mortda Mohammed Sahib György Kovács |
| author_sort | Mortda Mohammed Sahib |
| collection | DOAJ |
| description | Sandwich structures offer significant opportunities to improve the performance of many industrial applications such as aerospace, automotive, and marine. The design of a composite sandwich structure is often challenging because it is driven by the balance between the weight and cost of these structures. In this paper, a multi-objective optimization model using a Genetic Algorithm (GA) and an Artificial Neural Network (ANN) is presented to evaluate a new optimization approach in terms of weight and cost minimization for the composite sandwich structures. Classical lamination and beam bending theories were used, along with Monte Carlo simulation, to generate the design data for the proposed composite sandwich structure, which is applied as a floor panel for a high-speed train. Multilayer feedforward neural networks were used for predicting safety factors, cost, and weight of the designed structure based on the following inputs: core density, core thickness, face sheet materials’ combinations, and applied load. The trained Neural Network model was able to predict the considered results with a good performance metric, namely the coefficient of determination (R2 = 0.99) and the Mean Square Error (MSE = 1.3·10−5).Multi-objective optimization for cost and weight minimization was performed with a Genetic Algorithm using the derived ANN model. The obtained Pareto front provided several non-dominated optimal points leading to insights on the optimization process. The Finite Element Method (FEM) was used to model the key points of the optimal designs (i.e. the design with the lowest cost, weight, and Pareto optimal points). The FEM and optimization results had a maximum deviation of about 8.9%, indicating a good agreement between the two techniques.The newly elaborated methodology demonstrates a new approach for obtaining the optimum design of the investigated composite sandwich structure constructed from honeycomb core and laminated face sheets in terms the cost and weight. The study concluded that the use of Carbon Fiber-Reinforced Plastic (CFRP) or Fiber-Metal Laminate (FML) face sheets results in significant weight savings of about 59.5% and 48.6%, respectively, compared to an all-aluminum sandwich structure. |
| first_indexed | 2024-03-07T21:28:17Z |
| format | Article |
| id | doaj.art-f1424e044178424fae493cf41af32adb |
| institution | Directory Open Access Journal |
| issn | 2590-1230 |
| language | English |
| last_indexed | 2024-04-24T20:03:00Z |
| publishDate | 2024-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Results in Engineering |
| spelling | doaj.art-f1424e044178424fae493cf41af32adb2024-03-24T07:01:15ZengElsevierResults in Engineering2590-12302024-03-0121101937Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic AlgorithmMortda Mohammed Sahib0György Kovács1Faculty of Mechanical Engineering and Informatics, University of Miskolc, Hungary; Basrah Technical Institute, Southern Technical University, Basrah, IraqFaculty of Mechanical Engineering and Informatics, University of Miskolc, Hungary; Corresponding author.Sandwich structures offer significant opportunities to improve the performance of many industrial applications such as aerospace, automotive, and marine. The design of a composite sandwich structure is often challenging because it is driven by the balance between the weight and cost of these structures. In this paper, a multi-objective optimization model using a Genetic Algorithm (GA) and an Artificial Neural Network (ANN) is presented to evaluate a new optimization approach in terms of weight and cost minimization for the composite sandwich structures. Classical lamination and beam bending theories were used, along with Monte Carlo simulation, to generate the design data for the proposed composite sandwich structure, which is applied as a floor panel for a high-speed train. Multilayer feedforward neural networks were used for predicting safety factors, cost, and weight of the designed structure based on the following inputs: core density, core thickness, face sheet materials’ combinations, and applied load. The trained Neural Network model was able to predict the considered results with a good performance metric, namely the coefficient of determination (R2 = 0.99) and the Mean Square Error (MSE = 1.3·10−5).Multi-objective optimization for cost and weight minimization was performed with a Genetic Algorithm using the derived ANN model. The obtained Pareto front provided several non-dominated optimal points leading to insights on the optimization process. The Finite Element Method (FEM) was used to model the key points of the optimal designs (i.e. the design with the lowest cost, weight, and Pareto optimal points). The FEM and optimization results had a maximum deviation of about 8.9%, indicating a good agreement between the two techniques.The newly elaborated methodology demonstrates a new approach for obtaining the optimum design of the investigated composite sandwich structure constructed from honeycomb core and laminated face sheets in terms the cost and weight. The study concluded that the use of Carbon Fiber-Reinforced Plastic (CFRP) or Fiber-Metal Laminate (FML) face sheets results in significant weight savings of about 59.5% and 48.6%, respectively, compared to an all-aluminum sandwich structure.http://www.sciencedirect.com/science/article/pii/S2590123024001907Sandwich structureFiber-reinforced plasticFiber-metal laminateHoneycomb coreStructural optimizationArtificial neural network |
| spellingShingle | Mortda Mohammed Sahib György Kovács Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic Algorithm Results in Engineering Sandwich structure Fiber-reinforced plastic Fiber-metal laminate Honeycomb core Structural optimization Artificial neural network |
| title | Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic Algorithm |
| title_full | Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic Algorithm |
| title_fullStr | Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic Algorithm |
| title_full_unstemmed | Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic Algorithm |
| title_short | Multi-objective optimization of composite sandwich structures using Artificial Neural Networks and Genetic Algorithm |
| title_sort | multi objective optimization of composite sandwich structures using artificial neural networks and genetic algorithm |
| topic | Sandwich structure Fiber-reinforced plastic Fiber-metal laminate Honeycomb core Structural optimization Artificial neural network |
| url | http://www.sciencedirect.com/science/article/pii/S2590123024001907 |
| work_keys_str_mv | AT mortdamohammedsahib multiobjectiveoptimizationofcompositesandwichstructuresusingartificialneuralnetworksandgeneticalgorithm AT gyorgykovacs multiobjectiveoptimizationofcompositesandwichstructuresusingartificialneuralnetworksandgeneticalgorithm |