On the wear and failure of high speed roller bearings

A laboratory bearing test rig was constructed to study wear and failure mechanisms in high speed roller bearings under different conditions of loading and lubrication. The design of the rig permitted three bearings to be tested simultaneously. A test run of 720 h duration was performed during which wear of the bearings was monitored using ferrography and atomic absorption oil analysis. Wear debris particles collected from the bearings by ferrography were studied with scanning electron microscopy using energy-dispersive X-ray analysis. The results of ferrography were correlated with the metallographic observations made on the worn bearings.It was found that for lightly loaded high speed roller bearings rolling contact fatigue including surface pitting is not a significant wear or failure mechanism. However, misalignment due to shaft deflection above a critical point cannot be tolerated. Wear of high speed engine bearings can successfully be monitored by ferrography but, as the wear condition accelerates, frequent oil sample collections are beneficial.     

Failure Analysis and Condition Monitoring of an Open-Loop Oil System Using Ferrography

Condition monitoring of dynamic systems based on oil analysis is well known for closed-loop systems. The motivation for this work stemmed from repeating failures of Wankel engines. Failure analysis identified contact fatigue as the failure mechanism, but could not identify the cause. Thus, the objective of the work was to develop a method for condition monitoring of open-loop oil systems. A variety of analytical techniques was evaluated, including direct-reading ferrography, analytical ferrography combined with computational image analysis, atomic emission spectroscopy, and scanning electron microscopy combined with energy dispersive X-ray spectroscopy. Procedures for collection and separation of oil samples were developed. Analytical ferrography was found most useful in condition monitoring. Six engines were detected in their early failure stage. Those engines were disassembled, and contact fatigue failures in the bearing needles were observed. The quantitative image analysis allowed for a fairly objective rating of the wear level. The method developed in this work has already been implemented on a daily basis for monitoring the health of Wankel engines, with much success.     

基于yolov5在线可视铁谱图像磨粒多目标识别方法研究

鉴于在线图像可视铁谱获取的磨粒谱片图像分辨率低,磨粒种类复杂多变,磨粒图像背景复杂等问题,使得磨粒在线智能识别面临挑战。为了实现在线可视铁谱图像磨粒多目标实时检测与识别,提出基于yolov5在线可视铁谱图像磨粒多目标识别方法。以正常磨损磨粒、疲劳磨损磨粒、滑动磨损磨粒、球形磨粒、氧化磨损磨粒、切削磨损磨粒6种磨粒作为研究对象,基于yolov5深度神经网络模型对复杂油液环境下的异常磨损磨粒进行分割与识别。结果表明：基于yolov5算法的磨粒智能识别模型能够实现复杂环境下多目标、多类型磨粒在线实时识别,其识别速度和准确率基本满足油液在线监测需求,为装备在线图像可视铁谱技术工业化应用提供了技术支撑。     

滚动轴承疲劳失效过程的分析与研究

本文应用铁道技术以辅以ＳＥＭ、ＥＤＡＸ，表面轮廓仪等分析仪器，对滚动轴承在疲劳寿命试验中的润滑状态，磨损过程及疲劳失效机理进行了探讨。并讨论了铁普技术应用於滚动轴承的工作状态监测及故障诊断方面的可靠性及可靠性。     

基于油样分析技术的钻井泵寿命预测

通过跟踪3台不同运行时间钻井泵的油样的铁谱和光谱分析数据,发现随着使用时间的增多,设备磨损加大,油样中Al、Cu、Fe等金属元素含量明显增加,在泵接近损坏时,铁谱显示油液中存在严重滑动磨粒和球形磨粒,磨粒链加密并伴有疲劳剥离磨屑;光谱显示铁质含量陡增。分析了磨损产生的机制并提出了减少磨损的方法,提出了将金属磨屑增长的速率作为状态考察的重要判断指标。分析结果表明,通过油样检测分析可预报泵动力端轴承等摩擦副的运行状态,为泵的大修时间的确定以及提高泵的工作寿命提供了科学的依据。     

冲击-滑动工况条件下套管磨损机制图像分析

为实验验证套管磨损机制的理论研究,在冲击-滑动工况条件下进行了套管磨损的试验研究。运用SK2005工业显微镜、铁谱显微镜和扫描电镜,对套管磨损试验后的磨擦磨损表面进行了显微图像分析。研究发现在冲击-滑动工况条件下,深井超深井套管磨损机制主要为接触疲劳磨损、粘着磨损和磨粒磨损,从试验角度验证了深井超深井套管磨损的主要原因是钻杆对套管的横向冲击。     

齿轮磨损状态的在线铁谱监测试验研究

对ＯＬＦ１在线铁谱仪用于齿轮磨损状态的监测进行了研究。试验数据分析表明，ＯＬＦ－１在线铁谱仪对大磨粒有很强的敏感性，其采样值的大小和变化反映了齿面状态的变化。在正常磨损情况下，采样值曲线比较平稳；而一旦采样值出现无规律波动，则意味着不正常磨损的开始。与分析铁谱仪数据对比，也显示出非常近似的变化趋势。因而，ＯＬＦ－１在线铁谱仪用于齿轮磨损状态的实时和连续监测具有良好的性能。本文同时指出，ＯＬＦ－１在线铁谱与其它油液分析技术相结合，可以更准确地进行齿轮磨损状态的监测和诊断     

油液分析技术在振动筛故障诊断中的应用

振动筛具有振动幅度大，负荷重等特点，其运行状态好坏，将直接影响到生产线的正常运行，本文通过利用光谱和铁谱分析技术对某炼铁厂的3#振动筛进行状态监测，指出该振动筛发振器轴承在短时间内即发生疲劳失效的原因在于轴承的选型不合理，该结论得到了设计方的认可，并最终更换了新型号轴承，取得了良好的实际效果。     

齿轮箱磨粒在铁谱片上的识别特征

介绍了铁谱分析的原理、样品的制备方法和铁谱分析的步骤,概括了齿轮箱齿轮、滚动轴承、滑动接触3种主要摩擦副在磨合磨损、正常磨损、异常磨损(过载、过速、疲劳)3个阶段的磨粒的主要识别特征,包括尺寸、尺寸分布、数量、成分和形貌等。     

Advances in quantitative analytical ferrography and the evaluation of a high gradient magnetic separator for the study of diesel engine wear

Several sources of variation in quantitative analytical ferrography are investigated. A standard ferrography analysis procedure is developed. Normalization of ferrographic data to account for the amount of oil used to make the Ferrograms is discussed. Procedures to minimize the errors involved in calculating three quantitative ferrography parameters, i.e. the area A_L (% (ml of oil)⁻¹) covered by the large particles, the area A_S (% (ml of oil)⁻¹) covered by the small particles and the area under the curve (millimeters per cent per milliliter of oil) are outlined. Ferrographic data are presented which show that the volume and dilution ratio of the oil sample being analyzed have a major effect on the accuracy of the analysis. Several variables which influence the area-covered readings of the particle deposit on a Ferrogram are discussed. The accuracy of quantitative analytical ferrography is assessed.

Quantitative analytical ferrography is used to evaluate a high gradient magnetic separator (HGMS), both as a tool for engine wear research and as an oil filter. Laboratory tests indicate that the HGMS is an order of magnitude more efficient than conventional paper oil filters in removing magnetic wear debris from the lubricating oil.     

磨粒图象背景光照不均匀分布图象预处理

针对磨粒图片背景亮度呈现分布不均的问题，采用两种计算机图象处理方式进行了预处理，得到比较完整的二值图。第一种方式是：在没有铁磨粒情况下采集背景亮度变化，通过数字处理得到背景亮度变化的差值，然后将亮度补偿到铁谱图上。试验结果表明，该方法在边缘信息损失很小的情况下，得到磨粒二值图。第二种是：在不能得到背景光的条件下，通过一阶和二阶差分处理及其变形方法，来消除背景梯度变化，从而消除背景光的低频噪声。从试验效果来看，一阶和二阶差分不仅能进行边缘检测，而且能通过卷积运算消除背景光分布不均，磨粒边缘有点扩大，有些少量信息损失掉了，这种损失可以用图象形态学中的腐蚀的方法消除。两种方法相比，采用背景补偿的方法效果更好一些，在条件许可的情况下尽量采用背景光补偿方法：     

直读式旋转铁谱仪定量测试方法研究

在自行研制开发的DRF-II型直读式旋转铁谱仪的基础上,提出了利用光电传感器对包括斑点、油质、旋转谱片等进行测试的在用润滑油定量测试方法,并在机械设备的磨损状态检测中进行了试验应用研究。结果表明:提出的定量测试方法,数据可靠度高,对反映机械运行过程中的磨损状态和油质变化比较敏感;扩展了DRF铁谱仪的测试范围,为故障诊断提供了更多的依据。     

铁谱磨粒图像多焦点融合算法研究

铁谱分析技术是一种常用的磨损监测技术。受限于高倍物镜下的景深限制,一张铁谱大磨粒图像往往只有局部聚焦清晰的特征。为了能够解决在高倍物镜下铁谱大磨粒图像的自动化清晰采集以及高质量图像融合问题,设计并构建一套自动化扫描显微系统,该系统可进行多焦点铁谱图像的自动扫描采集;同时,提出一种基于相位一致性的铁谱磨粒图像多焦点融合算法,对自动扫描的多焦点图像进行融合,得到清晰的磨粒图像。实验结果表明,设计的自动化扫描显微系统能快速完成多焦点铁谱图像的自动化采集流程,提出的图像融合算法相较于传统的小波图像融合算法具有更高的图像评价质量,并能获得更加清晰的图像边缘信息。     

铁谱技术在水-乙二醇难燃液压液检测中的应用

介绍了水-乙二醇难燃液压液在台架试验中,如何用铁谱技术确定叶片泵的跑合终点,以及铁谱分析中水基样品的制备技术。试验证明应用铁谱技术,对判断叶片泵跑合终点是行之有效的手段。该液压液对台架系统适应性铁谱分析结果与其它摩擦磨损试验仪器测试结果具有一致性。     

用于铁谱分析磁化液使用规范的研究

本文简述了铁谱分析技术和磁化液的基本原理。在大量试验的基础上，建立了用于铁谱分析技术磁化液的使用规范。     

The development of ferrography in China—some personal reflections

Ferrography was first introduced into China over 20 years ago. Since then, academic research on wear debris formation mechanisms, intelligent identification of wear particle and other aspects have been carried out. Successful application of ferrography to machine condition monitoring in industry has been achieved with attendant cost benefits. Academic research has accelerated the development of ferrography, in which British scientists have made a considerable contribution to it. The future of ferrography is encouraging.     

铁谱技术中非铁磁性磨损颗粒的监测研究

在使用铁谱技术进行机械工况监测过程中，由于铁谱仪的磁场很难收集到摩擦副产生的非铁磁性磨损颗粒，往往会导致监测结果产生偏差或不正确。以钢一铜试件对磨后获得的油样为例进行铁谱对比实验，研究了磁流体在非铁磁性磨粒监测方面的应用。结果表明，在磁流体的作用下，油液中的非铁磁性磨粒将被磁化而有效沉积到铁谱片上，从而使油液的监测分析结果更为全面和准确。此外，介绍了磁流体的组成，分析了磁流体使非铁磁性磨粒磁化的机理。     

基于Mask R-CNN的铁谱磨粒智能分割与识别

针对铁谱图像因背景复杂、尺寸分布广、颗粒重叠等导致难以精确分割与识别的问题,以相似度高的疲劳剥块、严重滑动磨粒、层状磨粒共3种异常磨粒作为研究对象,提出基于深度神经网络模型Mask R-CNN的对多目标铁谱磨粒进行智能分割与识别的方法,并对特征提取层分别选用深度不同的残差网络ResNet50和ResNet101进行对比试验。实验结果表明,基于迁移学习方法的Mask R-CNN+ResNet101模型能够在复杂背景下对多目标、多类型、多尺寸的相似磨粒进行有效分割与识别,测试集的平均精度高达76.2%,模型具有较好的泛化能力。     

润滑脂铁谱分析的研究与应用

本文介绍了一种新的分离润滑脂中磨粒的方法，并对利用分析式铁谱仪和旋转式铁谱仪所制得的谱片进行了比较，指出二者相结合起来的方法是最有效的分析方法，介绍了它们的操作步骤，并通过对大型滚动轴承润滑脂的铁谱分析证明了该方法的有效性。     

基于Faster R-CNN的齿轮箱铁谱磨粒识别

在利用铁谱技术对齿轮箱磨损状态进行评估时,存在磨粒特征提取困难且磨粒识别的数量有限的问题,基于铁谱磨粒图像特性,提出基于Faster R-CNN算法的铁谱磨粒识别。该算法采用ResNet-34网络完成铁谱磨粒特征自动提取,并采用RPN网络实现对图像中多个磨粒的识别。通过实验对Batch_Size和学习率超参数进行优化,使用超参数优化后的Faster R-CNN算法进行实验。结果表明:该方法克服了磨粒交叉引起的识别难点,能识别一副图像中的多个磨粒,能统计各类磨粒数量,且准确率较高;在磨粒背景颜色不同、存在噪声干扰等情况下,该方法仍能够准确判断磨粒类型,具有较好的稳定性。