BP神经网络在摩擦学设计计算中的应用  

The application of BP neural network in the calculation of tribology design

作　　者：徐建生[1] 赵源[2] 

机构地区：[1]武汉化工学院机械系,湖北武汉430073 [2]武汉材料保护研究所

出　　处：《机械设计》2000年第10期16-18,共3页Journal of Machine Design

基　　金：国家自然科学基金!资助项目 (595750 34)

摘　　要：用双隐层BP人工神经网络 ,建立了丝杆螺母副的磨损率与滑动速度关系的数学模型。该模型可用于准确地计算丝杆螺母副和蜗轮蜗杆副的磨损率 ,可十分方便地用于摩擦学程序设计。采用L -M规则进行神经网络学习训练可使网络收敛快 ,误差小。网络输出结果与实验结果比较有极好的吻合性。该神经网络可为工程设计人员 ,在摩擦学设计时提供有效的计算工具。By the use of double hidden layered BP artificial neural network a mathematical model of relationship between wear rate of screw-nut pair and sliding speed is established.This model can be used in the accurate calculation of wear rate of screw-nut pair and worm-worm wheel pair,and can be applied quite conveniently to tribology programming as well.By adopting the L-M rule,the learning and training of neural network carried out in this paper enable the network a quicker convergence and a less error.The output result of network possesses extreme conincidence compared with the result of experiment.This neural network can provide an effective calculation means for personnels of engineering design while doing their tribology design.

关 键 词：摩擦学设计 神经网络 L-M训练规则 滑动速度 

分 类 号：TH117.1[机械工程—机械设计及理论] TP183[自动化与计算机技术—控制理论与控制工程]