Lightweight optimization of space reactor loop heat pipe radiation radiator based on surrogate models

Loop heat pipes, as a type of two-phase thermal control device, can not only eliminate the reliance on pump transportation technology in single-phase fluid circuits, but also utilize the structural features of gas-liquid pipeline separation to reasonably achieve the evaporation and condensation process. Considering that the loop heat pipes have good application prospects in the field of heat dissipation system in space environment, the condensation pipeline layout on its heat dissipation plate has been optimized in this paper to achieve lightweight while meeting the operational constraints. Based on previous research results, different static surrogate model training methods were compared and combined with genetic algorithms to complete preliminary optimization. After observing insufficient accuracy from static surrogate model, dynamic surrogate model technology was introduced. The results show that the dynamic surrogate model based on hybrid update strategy proposed in this paper can balance the local prediction accuracy, the global convergence ability, and the optimization calculation efficiency, which was more efficient and accurate in lightweight optimization.