DESIGN AND OPTIMIZATION OF AIRCRAFT SUB-FLOOR CRASHWORTHY STRUCTURE USING MACHINE LEARNING METHOD

• Main Author: H. Jamaddhiha Napitupulu, Manav
• Format: Final Project
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/78093
• Institution: Institut Teknologi Bandung

The development of crashworthy aircraft structure is required to preserve the safety of its occupants in a event of crash. Aircraft sub-floor, which include crash boxes, are designed to absorb as much of the energy during impact. This research study intends to improve the crashworthiness of an aircraft by designing and optimizing the crash box in the aircraft sub-floor using numerical simulations and machine learning. A finite element drop test simulation of an A320 fuselage section using LS-DYNA software were carried out. The crash box structure in the fuselage section is then optimized by varying the cross-section, material type, width and thickness to achieve the highest Specific Energy Absorption (SEA). This optimization is achieved by utilizing Machine Learning method, through stages of Artificial Neural Network (ANN) and Non-dominated Sorting Genetic Algorithm II (NSGA-II). 100 design configurations were generated using the Latin Hypercube Sampling (LHS) and Continuous Uniform Distribution (CUD). Numerical simulations of the 100 configurations were performed and the data obtained is used to train the ANN model and the NSGA-II predicts the optimized configuration based on the ANN model. The optimization result shows that the optimum crash box configuration for the A320 fuselage section model is made using AA7075-T6 material, with circular cross-sectional size of 50 mm and thickness of 1 mm. The result is then validated using finite element simulation. The tests shows that the optimum crash box configuration improves the SEA value by 337.17% with the prediction difference at 4.11%.