Abstract: Considering the multi-objective attribute and excessive computational cost of finite element analysis during the optimization of drilling fixture locating layout,a multi-objective optimization method of drilling fixture locating layout based on Back Propagation Neural Network (BPNN) and Particle Swarm Optimization (PSO) was proposed.The maximum deformation and the overall deformation of the sheet metal part after drilling under different fixture locating layouts were obtained by finite element simulation analysis.The simulation results were taken as training samples to establish a multi-objective neural network prediction model between the locating layout and the maximum and overall deformation of the sheet metal part.On this basis,the Pareto optimal solution set of the fixture locating layout for drilling a single hole was obtained by taking the fixture locating layout as design variable and the maximum deformation and the overall deformation of workpiece as minimum objectives,and combining BPNN prediction model with PSO algorithm.The feasibility and effectiveness of the proposed method were verified by two cases of flat and curved plate.

Key words: drilling fixture, sheet metal part, locating layout, deformation, multi-objective optimization, neural network, particle swarm optimization algorithm

摘要： 针对薄板件钻孔夹具定位布局设计存在多个目标和有限元仿真成本高的问题,提出一种基于反向传播神经网络(BPNN)与粒子群优化(PSO)算法的薄板件钻孔夹具定位布局多目标优化方法。通过有限元仿真分析获取不同夹具定位布局下薄板件钻孔后的最大变形和整体变形,将仿真结果作为训练样本,建立定位布局与薄板件最大变形和整体变形之间的多目标神经网络预测模型。在此基础上,以夹具定位布局为设计变量,以工件的最大变形和整体变形最小化为目标,结合神经网络预测模型与粒子群优化算法,得到钻单个孔时的夹具定位布局Pareto最优解集。最后,通过平板和曲板两个实例验证了所提方法的可行性和有效性。

关键词: 钻孔夹具, 薄板件, 定位布局, 变形, 多目标优化, 神经网络, 粒子群优化算法