An efficient thermal optimization model with integrated force paths, fully-decomposed hierarchies and hybrid genetic operations for a flight wing

Lightweighting of the thermal system is the key issue for high-speed vehicles, and the integrated optimization of heat-force transfer path is one of the most promising techniques. However, the inconsistent transfer paths and multi-hierarchy variables lead to a complex and time-consuming optimization problem. In this work, the integrated path with I-type force path and rectangular heat transport path on both sides of the force web is pre-designed for a high-speed wing. In order to form a great optimization space, the variable is fully decomposed into six hierarchies of topology, angle, heat width, force width, width factor and height factor, and the two-, three- and five-hierarchy models are supplemented to verify its enhancing effect on the optimization performance. Due to the intricate relevance among multiple hierarchies, a hybrid strategy based on the genetic algorithm is developed instead of the conventional sequential strategy, in which a binary coding scheme is adopted to uniformly describe the discrete and continuous variables, and the selection, crossover and mutation of different hierarchies are operated simultaneously during the genetic optimization. To address the excessive population size and high optimization cost caused by multiple variable hierarchies, an efficient population reduction method with limited effectiveness loss is constructed based on the individual pre-evaluation and screening by RBF neural network surrogate model. Finally, the wing is optimized with objectives of the minimum mass and minimum aeroelastic influence quantity, and the model effectiveness is verified by the great objective reductions of 35.7 % and 19.3 %, respectively.