一种飞行栖附机器人起降结构的多目标优化  

Multi-objective Optimization of a Take-off and Landing Structure for Flying and Perching Robot

作　　者：潘兆祾 顿向明 敬忠良[2] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200240 [2]上海交通大学航空航天学院,上海200240

出　　处：《机电一体化》2023年第1期53-61,共9页Mechatronics

基　　金：国防科工局基础科研项目(JCKY2021110B134)。

摘　　要：针对飞行栖附机器人的壁面起降问题,提出了一种借助结构实现的新型起降策略,可利用初始动能完成贴壁过程中的姿态转换。在该策略的基础上利用显式计算软件LS-DYNA对该贴壁一变形过程进行计算机仿真。结合理论分析与仿真结果,针对该系统的碰撞力优化、变形可靠性、轻量化等关键问题,提取了组合臂长度、缓冲垫厚度等优化设计变量。为减少计算时间,利用Hammersley采样法在优化空间中均匀采样,并使用移动最小二乘法构建代理模型,以接触力、质量为优化目标,变形前后重心变化量为优化约束,采用遗传算法NSGA-I对结构参数进行多目标优化。优化后转轴处接触力峰值下降11.99%,关键结构质量下降3.72%,优化目标均得到改善。Aiming at the problem of wall take-off and landing of the flying and perching robot,a new take-off and landing strategy based on structure is proposed,which can use the initial kinetic energy to complete the attitude conversion in the process of wall attachment.On the basis of this strategy,the computer simulation of the adhesion-deformation process is carried out by using the explicit calculation software LS-DYNA.Combined with the theoretical analysis and simulation results,aiming at the key problems of collision force optimization,deformation reliability and lightweight of the system,the optimization design variables such as the length of the combined arm and the thickness of the cushion are extracted.In order to reduce the calculation time,the Hammersley sampling method is used to uniformly sample in the optimization space,and the moving least squares method is used to construct the surrogate model.The contact force and mass are taken as the optimization objectives,and the change of the center of gravity before and after deformation is taken as the optimization constraint.The genetic algorithm NSGA-II is used to optimize the structural parameters.After optimization,the peak value of contact force at the shaft decreases by 11.99%,and the mass of key structure decreases by 3.72%.The optimization objectives are improved.

关 键 词：飞行栖附机器人 显式有限元 多目标优化 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]