摘要
提出了用非线性惯性因子ω改进的微粒群算法与BP神经网络相结合的方法,适当选择钻削刀具的切削用量,克服了BP网络训练时间长,因易陷入局部最优点而不利于全局最优点搜索的不足.通过相同的实验样本测试发现,与以前的BP和GA-BP算法相比,训练时间分别缩短了73s和21s,测试的正确率分别提高了0.83%和0.32%.
This paper proposes combining BP Network and an improved Particle Swarm Optimizer with nonlinear inertia factor, to choose drilling tool' s cutting dosage. This method resolves the BP Network' s long training time and difficulty to search the global minimum since plunging into local minimum. From the same experimental sample, compared with previous BP arithmetic and GA-BP arithmetic ,the training time with this algorithm has been shortened 73s and 52s respectively, meanwhile the accurate rate has been improved O. 83% and 0. 32% respectively.
出处
《哈尔滨理工大学学报》
CAS
2008年第5期57-60,共4页
Journal of Harbin University of Science and Technology
基金
黑龙江省"十五"科技攻关项目(GA02A401-6)
关键词
微粒群
BP神经网络
切削用量
钻削刀具
particle swarm optimizer(PSO)
BP network
cutting dosage
drill tool