摘要
利用Matlab系统建立了丝杠螺母副的摩擦系数与载荷之间关系的BP人工神经网络模型。并用trainbpx动量—自适应学习率调整算法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。采用Matlab神经网络工具函数建立神经网络的方法简单方便 。
It uses Matlab to establish the BP neural network with relationship of load and friction characteristic of screw-nut pairs. The network is adapted quicker and has less error with the Trainbpx rule. The comparison of the network output with the test shows the results are identical. With the function of the toolbox in Matlab the calculable measure could be provided for the tribology design.
出处
《机械设计与制造工程》
2000年第4期12-14,共3页
Machine Design and Manufacturing Engineering
基金
国家自然科学基金资助项目! ( 595750 34)
