摘要
基于分群粒子群算法对平面度误差判定进行了研究。首先建立平面度误差评定数学模型,对平面度误差最小求解转化成对目标函数的非线性最优化问题;接着改进粒子群算法把粒子群一分为二,在不增加粒子个数和粒子维度的情况下,两个粒子群分别用来全局搜索和局部搜索,通过阈值判断早熟现象;最后给出了算法流程。实例验证结果表明:该算法具有较强的优化能力,对测试函数求解的最优解值数据波动性比较小,平面度的公差值为0.0073 mm,相比LSM、DM、TPM、PSO、ABC算法公差值平均分别减少了0.0023 mm,0.0025 mm,0.0027 mm,0.0002 mm,0.0005 mm,评定精度较高。
Flatness error is determined based on grouped particle swarm optimization(GPSO).Firstly,mathematic model of flatness error evaluation was established,and flatness error minimum was transmuted into nonlinear optimization problem of objective function.Secondly,GPSO was divided one PSO into two,which wasn't increased particle number and dimension,and one searched global,the other searched local,threshold was prevented premature.Finally,the process was given.The simulation results show that GPSO has a strong ability of optimization,and the optimal solution value of the test function is relatively small,the tolerance value of flatness is 0.0073 mm,the average tolerances of GPSO are reduced 0.0023 mm,0.0025 mm,0.0027 mm,0.0002 mm and 0.0005 mm comparing with LSM,DM,TPM,PSO and ABC,and the precision is higher than others algorithms.
作者
毕立恒
朱彦齐
BI Li-heng;ZHU Yan-qi(Yellow River Conservancy Technical Institute,Kaifeng,Henan 453002,China;Kai Feng Institute of Technicians,Kaifeng,Henan 453002,China)
出处
《计量学报》
CSCD
北大核心
2019年第6期980-985,共6页
Acta Metrologica Sinica
基金
国家自然科学基金(61275154)
关键词
计量学
平面度
误差评定
粒子群算法
阈值
metrology
flatness
error evaluation
particle swarm optimization
threshold
