渐进成形工艺参数的分阶段扰动粒子群算法优化  

Optimization on incremental forming process parameters based on staged perturbation particle swarm algorithm

作　　者：徐伟[1] 万轶 沙鑫美[1] Xu Wei;Wan Yi;Sha Xinmei(School of Mechanical and Electrical Engineering,Sanjiang University,Nanjing 210012,China;School of Environment Science,Nanjing Xiaozhuang University,Nanjing 211171,China)

机构地区：[1]三江学院机械与电气工程学院,江苏南京210012 [2]南京晓庄学院环境科学学院,江苏南京211171

出　　处：《锻压技术》2020年第12期116-121,共6页Forging & Stamping Technology

基　　金：江苏省高等学校自然科学研究项目(17KJB460011,19KJB460006)。

摘　　要：为了降低渐进成形制件的减薄率、增加制件的光滑度,提出了基于分阶段扰动粒子群算法的参数优化方法。选择了渐进成形制件的减薄率作为优化目标,设计了4因素4水平正交实验。采用智能算法对BP神经网络参数进行训练,提出了智能神经网络的数值拟合方法。以制件减薄率均值和标准差最小为目标,建立了带约束的多目标优化模型。将精英粒子的多阶段扰动策略引入到粒子群算法中,平衡了算法的多样性和收敛性,深化了算法的优化能力,从而提出了分阶段扰动粒子群算法的模型求解方法。对成形角为37°圆台件的渐进成形参数进行了优化,分阶段扰动粒子群算法优化后的制件减薄率均值和标准差均小于粒子群算法,从而提高了制件的质量和光滑度。In order to decrease the thinning rate of incremental formed part and increase the smoothness of part, the optimization method of parameters based on staged perturbation particle swarm algorithm was proposed, and taking the thinning rate of incremental formed part as optimization object, four-factor and four-level orthogonal experiment was designed. Then, BP neutral network parameters were trained by the intelligent algorithm, and the fitting method based on intelligent neutral network was put forward. Taking the minimum mean and the standard deviation of thinning rate of part as object, the multi-object optimization model with restriction was built. Furthermore, introducing elite particle staged perturbation strategy to particle swarm algorithm, the diversity and convergence of algorithm were balanced, and the optimizing ability of algorithm was deepened so that the model solving method based on staged perturbation particle swarm algorithm was proposed. Finally, the incremental forming parameters of circular table with the forming angle of 37°were optimized, and the mean and the standard deviation of thinning rate for part optimized by the staged perturbation particle swarm were smaller than those by the particle swarm algorithm. Thus, the quality and smoothness of part were improved.

关 键 词：渐进成形 光滑度 减薄率 智能神经网络 分阶段扰动粒子群算法 正交实验 

分 类 号：TG386[金属学及工艺—金属压力加工]