Simulation-data-driven vibration optimization of deployable stepwise composite booms

Abstract

Bistable reeled composite (BRC) booms are suitable candidates for space applications owing to their light weight and high storage efficiency. However, their stiffness and vibration performance affect the dynamics of the space structures. In this study, a simulation-data-driven method was developed to optimize the frequency of a stepwise BRC boom subjected to bending-mode vibration with respect to fiber angles within bistable composite laminates. First, a specified constant coiled diameter and bistability criterion are taken as constraints for classifying the input datasets. A vibration frequency prediction model is constructed using a genetic algorithm and back-propagation neural network trained by finite element simulation values. Then, constricted particle swarm optimization is used to maximize the bending-mode frequency of the BRCs. The optimization results are validated by published values showing that the BRC with a step ratio of 0.55 and a stacking sequence of [±α1 ∕ ±α2 ∕ ±α2 ∕ ±α1] at the fixed end and [±α3 ∕ ±α3] at the free end can be improved by 63.8 % at α1 = 44.8◦, α2 = 35◦, α3= 44◦ in comparison to the uniform-thickness benchmark boom. Additionally, the evaluation and validation indices demonstrate that the simulation-data-driven prediction and optimization approach yields high computational accuracy.

Keywords

• Data-driven vibration optimization
• Bistable reeled composite booms
• Genetic algorithm
• Back-propagation neural network
• Constricted article swarm optimization