DE算法优化CNN的滚动轴承故障诊断研究

针对卷积神经网络(Convolutional Neural Network,CNN)在滚动轴承故障诊断应用中所存在超参数难以确定、网络输出不稳定等问题,提出一种基于差分进化算法(Differential Evolution,DE)优化卷积神经网络的故障诊断模型(CNN-DE)。首先,该故障诊断模型利用DE算法来自适应调节CNN中的超参数,同时将CNN的诊断精度和稳定性一起作为DE算法优化的目标,使得CNN在保证精度的同时降低网络的波动;其次,使用自适应噪声完备集合经验模态分解(Complete Ensemble Empirical Mode Decomposition with Adaptive Noise,CEEMDAN)结合多参数的诊断方法,从原始信号中提取出更加有效的故障特征;最后,根据提取出的特征采用CNN-DE、CNN和BP等算法进行故障诊断,结果表明所提出的算法模型拥有更高的精度和更稳定的性能,也具有优异的抗噪能力,显示了CNN-DE在故障诊断应用中的可靠性。

探究英语词汇学习

词汇量是衡量语言学习者的重要标尺。有效扩大词汇量需要学习者具有端着的学习态度,遵循记忆的规律,把词汇学习和句法篇章阅读相结合,并掌握一套有效的学习方法。语言学习总是发生在特点的社会环境之中。它是社会环境、学习者心智特点、物资条件(内容、目标、环节等),学习过程(策略、技巧和思维运作)、成果评价五个部分。一个语言学习者如果能充分发挥各个要素的优势,他的学习就会获得最大限度的收获。

Ga-based liquid metal as an extreme pressure lubricant for steel-ceramic pairs

Extreme pressure lubrication is a crucial engineering problem for many mechanical assemblies.Herein,an extraordinary and novel functional material,Ga-based liquid metal(GLM),is found to be a promising extreme pressure lubricant for steel-ceramic sliding pairs and bears applied load above 1500 N.A Ga-rich metallic tribofilm forms on the frictional interface during sliding,providing good lubricity and bearing capability.In addition,the lubrication performance of GLM strongly depends on its composition and the contacting materials.Ternary Ga-In-Sn alloy has superior lubrication to binary Ga-In alloy.GLM lubrication is more effective on the interfaces of AISI52100/Si_(3)N_(4)than on those of AISI52100/SiC sliding pairs.