摘要
为提升智能数控机床科技的加工质量和刀具耐磨度,并降低生产成本,提出一种自适应量子粒子群网络下的切削加工参数优化方法。为解决多目标数控切削参数优化的非线性与多约束问题,将粒子群方法(Particle Swarm Optimization,PSO)与改进Elman网络结合,并在粒子群方法中引入量子机制,通过自适应惯性权值调节适应度,利用自适应动量反向传播(Back Propagation,BP)方法完成网络训练,在网络学习的过程中,获得最优数控切削参数。采用KMC800SU五轴立式数控机床完成Matlab 2021a下的对比实验,以表面粗糙度为例,利用此方法加工的工件粗加工和精加工粗糙度能分别达到7.6μm、3.5μm,而PSO方法仅能分别达到8.6μm、3.9μm。结果表明,此方法相对于PSO方法的参数匹配更加合理,能够在较少的迭代次数下达到稳定和较优的表面粗糙程度、刀具耐磨度与平均最大完工时长。
For improving the machining quality and wear-resistance of intelligent numerically-controlled(NC)machine technology,and decreasing the production cost,an adaptive quantum particle swarm optimization method for machining parameters was proposed.Particle swarm optimization(PSO)method and improved Elman network are combined to solve the nonlinear and multi-constraint problems of multi-objective NC cutting parameter optimization.Then,quantum mechanism is introduced into PSO algorithm to adjust the fitness through adaptive inertia weight,and the network training is completed by using adaptive momentum back-propagation method.In the process of network learning,the optimal NC cutting parameters are obtained.A KMC800SU five-axis vertical NC machine tool was used to complete the comparison experiment under Matlab 2021a.Taking the surface roughness as an example,the roughing and finishing machining energy of the workpiece obtained by the proposed method can reach 7.6μm and 3.5μm respectively,while the PSO method can only reach 8.6μm and 3.9μm separately.The results show that the parameter matching of the proposed method is more reasonable than that of the PSO method,and it can achieve stable and better surface roughness,tool wear and average maximum completion time in less iterations.
作者
韩辉辉
付辉
HAN HuiHui;FU Hui(Chongqing Industry Polytechnic College,Chongqing 401120,China;College of Electrical and Information Engineering,Lanzhou University of Technology,Lanzhou 730050,China)
出处
《机械强度》
CAS
CSCD
北大核心
2023年第5期1117-1123,共7页
Journal of Mechanical Strength
基金
国家自然科学基金项目(62001198)
甘肃省青年科技基金计划(21JR7RA247)资助。
关键词
切削
量子机制
惯性权值
神经网络
PSO方法
Cutting
Quantum mechanism
Inertial weight
Neural network
PSO method
