共查询到20条相似文献,搜索用时 187 毫秒
1.
2.
3.
4.
为了提高机械加工过程中刀具磨损在线监测的准确性,提出了一种基于长短时记忆卷积神经网络(LSTM-CNN)的刀具磨损在线监测模型。在该监测模型中,通过振动、力、声发射传感器对刀具切削过程中的振动、力和声发射信号进行采集,采集的数据其本质为时间序列数据。考虑采集数据的序列和多维度特性,采用LSTM-CNN网络对采集的数据进行序列和多维度特征提取,利用线性回归实现特征到刀具磨损值的映射。通过实验验证了该模型的有效性和可行性,模型的精度较其他几种方法有了较大的提高。 相似文献
5.
6.
7.
基于开放式数控系统的车刀磨损监控技术研究 总被引:2,自引:1,他引:2
研究了采用工业PC机、电机控制卡和数据采集卡构成的开放式数控系统的体系结构和功能。在加工过程中,数据采集单元实时采集刀具的振动信号,经过数字滤波,运用多元回归建模技术,建立了车削过程刀具磨损监测模型。该方法也可以方便地建立适合不同切削过程要求的监控模块。实验表明,在开放式数控系统平台上,可实现车刀磨损程度监控技术,保证了加工过程的稳定性。 相似文献
8.
刀具磨损监测及破损模式的识别 总被引:2,自引:0,他引:2
对于金属切削过程中的刀具磨损,提出了基于隐马尔可夫模型的模式识别理论来识别刀具的不同磨损状态,从而预报刀具破损.该方法对切削过程中切削力信号的动态分量和刀柄振动信号进行快速傅里叶变换特征提取,然后利用自组织特征映射对提取的特征矢量进行预分类编码,把矢量编码作为观测序列引入到隐马尔可夫模型中进行机器学习,建立了3个不同磨损状态的隐马尔可夫模型,并利用最大概率进行模式识别.试验表明,该方法对车刀磨损过程进行识别和预报是有效的. 相似文献
9.
10.
本文介绍并分析了当前的一些过程建模方法,描述了基于组装的过程建模方法。该建模方法能够对可重用过程库中的过程部件进行组装,生成新的过程模型。最后介绍了一个采用基于组装的过程建模方法的环境。 相似文献
11.
12.
13.
基于遗传算法的零件加工系统误差建模 总被引:5,自引:2,他引:5
系统误差补偿是提高零件加工精度的重要途径,其中系统误差的建模是关键。分析了几种常见的建模方法在系统误差建模中的不足,提出了基于遗传算法的加工系统误差建模方法,它是一个广义的多项式回归建模方法——多项式幂次为变参。通过实例分析及与其它几种方法的比较,说明了所提方法的有效性,并指出了它对不等间隔误差序列建模的适用性。 相似文献
14.
15.
16.
17.
人工髋关节超高分子量聚乙烯(UHMWPE)关节面磨损仍是影响置换关节远期寿命的主要因素,其仿真建模是对关节模拟机磨损测试手段的重要补充,也是实现置换关节临床前性能评估的有效方法。由多向运动产生的交叉剪切效应是影响UHMWPE磨损的主要原因之一,也是仿真建模的关键。现有理论方法将磨损深度确定为滑动距离的函数,并将90°交叉剪切运动条件下的磨损作为度量基准计算不同角度下的交叉剪切效应,但尚未考虑接触应力变量对磨损深度的影响。针对以上问题,提出了在垂直交叉剪切运动条件下将磨损深度表示为摩擦功函数的方法。该方法利用UHMWPE摩擦因数与接触压力的定量关系计算摩擦因数并确定摩擦功,解决了UHMWPE磨损交叉剪切效应中滑动距离与接触应力的耦合问题。基于磨损仿真新模型研究了36 mm直径的交联UHMWPE髋关节,并与已有ProSim模拟机试验结果进行了验证。结果显示该仿真模型可准确计算体积磨损和线性磨损等磨损量以及髋关节载荷方向改变对磨损的影响。磨损新模型为进一步仿真模拟奠定了有效基础。 相似文献
18.
Because wear is one of the most typical causes of decreasing performance in running machines, monitoring wear is regarded as a crucial technology in maintaining the health of machines. However, monitoring wear is not a fully mature process because quantifying the development of wear in real time is a challenging task because there is no universal indicator. To meet this need, wear-oriented dynamic modeling with online ferrographic images was used to investigate and then describe a real-time wear state. This investigation was carried out by combining three wear indices to describe the wear rate, the wear mechanism, and the severity of wear. A binary classifier method is also proposed to classify these wear stages in the three extracted indices. A strategy to identify the dynamic transition of wear states with adaptive parameters is also developed and then a four-ball wear test is carried out to verify the method. The results indicate that this modeling strategy can accurately identify a developing wear state that is characterized by stages. This proposed method is better at monitoring the health evolution of a machine system than just detecting faults. 相似文献
19.
Delamination wear constitutes one relevant wear process in many materials. In particular, wear failure through delamination is relevant for material systems where a coating is present on a substrate material. A computational modeling approach is presented that aims to describe the processes of formation and growth of wear delaminations. The model is based on the cohesive zone model approach. An irreversible cohesive zone model is used in a parametric numerical study of delamination wear in a coating system. The model predicts trends in agreement with the trends emerging from Archard's law. The proposed modeling approach has the advantage that details of the delamination wear process can numerically be studied, and that a unified framework from delamination initiation and propagation is provided. 相似文献