摩擦表面边界膜温度特性的神经网络模型  

Neural Network Model for Examination of the Temperature Characteristics of Boundary Film on Friction Surface

作　　者：徐建生[1] 李健[2] 邹岚[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磨损实验室,湖北武汉430030

出　　处：《摩擦学学报》2000年第6期469-471,共3页Tribology

基　　金：国家自然科学基金资助项目 !(5 95 75 0 34)

摘　　要：采用非线性变换单元组成的多层前馈神经网络建立了丝杆螺母摩擦副表面边界膜温度特性的磨损自补偿数学模型 ,该模型可用于准确地预测边界膜对摩擦学特性的影响 .采用 L- M规则进行神经网络学习训练使网络收敛快且误差小 ,所得网络输出结果与实验结果有较好的吻合性 .A mathematical model based on BP neural network has been established to examine the temperature characteristics of the boundary film on the friction surfaces of a screw nut pair which is characterized by wear self compensation feature. The network could be used to predict the effect of the boundary film on tribological behavior. It is also capable of learning and the error is small while being trained according to L M rule. The outputs of the network are precise and in good agreement with the experimental ones. The network model could be used as an effective calculation tool for the tribological design of engineers.

关 键 词：边界膜 温度特性 线杆-螺母摩擦副 神经网络模型 

分 类 号：TH117[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]