基于响应面法的爬壁机器人多目标优化设计

Multiobjective Optimization Design of Wallclimbing Robot Based on Response Surface Method

为满足爬壁机器人在工作过程中高强度、轻量化的设计要求,提出一种基于参数相关性筛选的多目标优化方法。将爬壁机器人底盘以及侧挡板的5个参数作为设计变量,经过参数相关性分析,筛选后得到对机器人结构性能影响灵敏度较高的3个设计变量,以质量最小、位移变形最小和应力最大为优化目标,经过多元回归分析和帕累托方差分析拟合得到二项式响应面模型,采用非支配排序遗传算法对爬壁机器人机身进行多目标优化。结果表明,经过参数相关性筛选后,计算量大大减少,机身质量减少1.7 kg,最大位移变形减小0.1 mm,最大等效应力减少1.36 MPa,模态固有频率增加,避免共振现象。

In order to meet the highstrength and lightweight design requirements of wallclimbing robots in the working process,a multiobjective optimization method based on parameter correlation screening is proposed.The five parameters of the wallclimbing robot chassis and side baffles are used as design variables.After parameter correlation analysis,three design variables with high sensitivity to the robot’s structural performance are obtained after screening.With the smallest mass,the smallest displacement and the largest stress as the optimization goals,the binomial response surface model is obtained through multiple regression analysis and Pareto variance analysis.The nondominated sorting genetic algorithm is used to optimize the body of the wallclimbing robot.The results show that after the parameter correlation screening,the amount of calculation is greatly reduced,the mass of the fuselage is reduced by 1.7 kg,the maximum displacement and deformation are reduced by 0.1 mm,the maximum equivalent stress is reduced by 1.36 MPa,and the modal natural frequency is increased to avoid resonance.