期刊文献+
共找到3篇文章
< 1 >
每页显示 20 50 100
基于卷积神经网络的微电机装配故障诊断研究 被引量:7
1
作者 刘冰洁 陈炳发 丁力平 《机械制造与自动化》 2021年第3期171-174,189,共5页
针对微电机装配质量控制水平低、产线故障发现不及时且难以做出最佳决策等问题,提出一种基于卷积神经网络(CNN)的微电机装配故障诊断方法。该方法将实时采集的微电机装配过程质量特性数据绘制出控制图,采用数值转化为图像的数据预处理... 针对微电机装配质量控制水平低、产线故障发现不及时且难以做出最佳决策等问题,提出一种基于卷积神经网络(CNN)的微电机装配故障诊断方法。该方法将实时采集的微电机装配过程质量特性数据绘制出控制图,采用数值转化为图像的数据预处理方法实现CNN对控制图异常模式的识别,最后通过控制图异常模式向故障映射的方法完成故障诊断。基于该方法开发了一套微电机装配故障诊断原型系统,可用于微电机装配过程的实时监控与故障诊断。 展开更多
关键词 卷积神经网络 微电机装配 故障诊断 控制图 模式识别
下载PDF
基于声学特征的微电机故障诊断方法
2
作者 刘艳杰 陈炳发 丁力平 《机械制造与自动化》 2022年第2期190-194,共5页
针对微电机质量检测水平低、故障识别困难等问题,设计一种基于声学特征的微电机故障诊断方法。通过声音采集装置获得微电机转动时的正常声音信号和三种故障信号;从声音信号中提取39维梅尔频率倒谱系数和短时能量,搭建一维卷积神经网络... 针对微电机质量检测水平低、故障识别困难等问题,设计一种基于声学特征的微电机故障诊断方法。通过声音采集装置获得微电机转动时的正常声音信号和三种故障信号;从声音信号中提取39维梅尔频率倒谱系数和短时能量,搭建一维卷积神经网络模型进行识别。将声音信号转化成语谱图,建立二维卷积神经网络模型并识别。利用多模型融合技术中的加权平均算法将两个模型融合,融合后模型的准确率为93.58%,比单个模型平均提高2.43%。 展开更多
关键词 声学特征 模型融合 卷积神经网络 微电机 故障诊断
下载PDF
Optical Characteristics of Chromophoric Dissolved Organic Matter(CDOM)in Upstream and Downstream Lakes of Taihu Lake Basin:New Insights for Water Environmental Management
3
作者 REN Weixiang WU Xiaodong +4 位作者 chen bingfa CHAO Jianying GE Xuguang YANG Jiuyun YANG Hui 《Chinese Geographical Science》 SCIE CSCD 2022年第4期606-619,共14页
Chromophoric dissolved organic matter(CDOM)is a key component of organic matter that contributes to the ecological functioning of lakes.The lakes in Taihu Lake Basin play an important role in maintaining regional ecol... Chromophoric dissolved organic matter(CDOM)is a key component of organic matter that contributes to the ecological functioning of lakes.The lakes in Taihu Lake Basin play an important role in maintaining regional ecological stabilities;however,the optical characteristics of the CDOM in the upstream and downstream lakes in this basin have not yet been systematically studied.Here,the optical characteristics of CDOM in ten lakes of upstream and downstream of the Taihu Lake Basin were studied using UV-Visible and excitation-emission matrix spectroscopies.Three different fluorophores consisting of two humic-like components(C1,C2)and one protein-like component(C3)were identified by parallel factor analysis.Soil or surface erosion was responsible for the higher abund-ance of C1 in the upstream lakes,and increased biological activities accounted for the higher abundance of C3 in the downstream lakes.Rainfall erosion in the wet season led to an increase in CDOM.We also found that the photodegradation and flocculation degree,which played a significant role in reducing CDOM,were higher in downstream lakes than in upstream lakes.Optical analysis of CDOM provides a promising method for monitoring water qualities(e.g.,total phosphorus and potassium permanganate index)in each lake.Re-ductions in soil or surface erosion in the upstream are needed to improve water quality. 展开更多
关键词 Taihu Lake Basin chromophoric dissolved organic matter(CDOM) excitation-emission spectra parallel factor analysis(PARAFAC)
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部