Research on Instantaneous Angular Speed Signal Separation Method for Planetary Gear Fault Diagnosis

Planetary gear train is a critical transmission component in large equipment such as helicopters and wind turbines. Conducting damage perception of planetary gear trains is of great significance for the safe operation of equipment. Existing methods for damage perception of planetary gear trains mainly rely on linear vibration analysis. However, these methods based on linear vibration signal analysis face challenges such as rich vibration sources, complex signal coupling and modulation mechanisms, significant influence of transmission paths, and difficulties in separating damage information. This paper proposes a method for separating instantaneous angular speed (IAS) signals for planetary gear fault diagnosis. Firstly, this method obtains encoder pulse signals through a built-in encoder. Based on this, it calculates the IAS signals using the Hilbert transform, and obtains the time-domain synchronous average signal of the IAS of the planetary gear through time-domain synchronous averaging technology, thus realizing the fault diagnosis of the planetary gear train. Experimental results validate the effectiveness of the calculated IAS signals, demonstrating that the time-domain synchronous averaging technology can highlight impact characteristics, effectively separate and extract fault impacts, greatly reduce the testing cost of experiments, and provide an effective tool for the fault diagnosis of planetary gear trains.

Time Synchronous Averaging Based on Cross-power Spectrum

Periodic components are of great significance for fault diagnosis and health monitoring of rotating machinery.Time synchronous averaging is an effective and convenient technique for extracting those components.However,the performance of time synchronous averaging is seriously limited when the separate segments are poorly synchronized.This paper proposes a new averaging method capable of extracting periodic components without external reference and an accurate period to solve this problem.With this approach,phase detection and compensation eliminate all segments'phase differences,which enables the segments to be well synchronized.The effectiveness of the proposed method is validated by numerical and experimental signals.

Fault feature extraction of planet gear in wind turbine gearbox based on spectral kurtosis and time wavelet energy spectrum

Planetary transmission plays a vital role in wind turbine drivetrains, and its fault diagnosis has been an important and challenging issue. Owing to the complicated and coupled vibration source, time-variant vibration transfer path, and heavy background noise masking effect, the vibration signal of planet gear in wind turbine gearboxes exhibits several unique characteristics： Complex frequency components, low signal-to-noise ratio, and weak fault feature. In this sense, the periodic impulsive components induced by a localized defect are hard to extract, and the fault detection of planet gear in wind turbines remains to be a challenging research work. Aiming to extract the fault feature of planet gear effectively, we propose a novel feature extraction method based on spectral kurtosis and time wavelet energy spectrum （SK-TWES） in the paper. Firstly, the spectral kurtosis （SK） and kurtogram of raw vibration signals are computed and exploited to select the optimal filtering parameter for the subsequent band-pass filtering. Then, the band-pass filtering is applied to extrude periodic transient impulses using the optimal frequency band in which the corresponding SK value is maximal. Finally, the time wavelet energy spectrum analysis is performed on the filtered signal, selecting Morlet wavelet as the mother wavelet which possesses a high similarity to the impulsive components. The experimental signals collected from the wind turbine gearbox test rig demonstrate that the proposed method is effective at the feature extraction and fault diagnosis for the planet gear with a localized defect.

Research on intelligent fault diagnosis based on time series analysis algorithm

Aiming to realize fast and accurate fault diagnosis in complex network environment, this article proposes a set of anomaly detection algorithm and intelligent fault diagnosis model. Firstly, a novel anomaly detection algorithm based on time series analysis is put forward to improve the generalized likelihood ratio （GLR） test, and thus, detection accuracy is enhanced and the algorithm complexity is reduced. Secondly, the intelligent fault diagnosis model is established by introducing neural network technology, and thereby, the anomaly information of each node in end-to-end network is integrated and processed in parallel to intelligently diagnose the fault cause. Finally, server backup solution in enterprise information network is taken as the simulation scenario. The results demonstrate that the proposed method can not only detect fault occurrence in time, but can also implement online diagnosis for fault cause, and thus, real-time and intelligent fault management process is achieved.

AUTO-EXTRACTING TECHNIQUE OF DYNAMIC CHAOS FEATURES FOR NONLINEAR TIME SERIES

The main purpose of nonlinear time series analysis is based on the rebuilding theory of phase space, and to study how to transform the response signal to rebuilt phase space in order to extract dynamic feature information, and to provide effective approach for nonlinear signal analysis and fault diagnosis of nonlinear dynamic system. Now, it has already formed an important offset of nonlinear science. But, traditional method cannot extract chaos features automatically, and it needs man＇s participation in the whole process. A new method is put forward, which can implement auto-extracting of chaos features for nonlinear time series. Firstly, to confirm time delay r by autocorrelation method; Secondly, to compute embedded dimension m and correlation dimension D; Thirdly, to compute the maximum Lyapunov index λmax; Finally, to calculate the chaos degree Dch of Poincare map, and the non-circle degree Dnc and non-order degree Dno of quasi-phase orbit. Chaos features extracting has important meaning to fault diagnosis of nonlinear system based on nonlinear chaos features. Examples show validity of the proposed method.

Demodulation Based on Harmonic Wavelet and Its Application into Rotary Machinery Fault Diagnosis

The harmonic wavelet transform（HWT） and its fast realization based on fast Fourier transform（FFT） are introduced. Its ability to maintain the same amplitude-frequency feature is revealed. A new method to construct the time-frequency（TF） spectrum of HWT is proposed, which makes the HWT TF spectrum able to correctly reflect the time-frequency-amplitude distribution of the signal. A new way to calculate the HWT coefficients is proposed. By zero padding the data taken out, the non-decimated coefficients of HWT are obtained. Theoretical analysis shows that the modulus of the coefficients obtained by the new calculation way and living at a certain scale are the envelope of the component in the corresponding frequency band. By taking the cross section of the new TF spectrum, the demodulation for the component at a certain frequency band can be realized. A comparison with the Hilbert demodulation combined with band-pass filtering is done, which indicates for multi-components, the method proposed here is more suitable since it realizes ideal band-pass filtering and avoids pass band selecting. In the end, it is applied to bearing and gearbox fault diagnosis, and the results reflect that it can effectively extract the fault features in the signal.

Online Fault Diagnosis of Modern Process Industry System Based on Color-Spectrum

This paper proposes a new method to diagnose the system fault of the process industry based on the monitor data set of distributed control system(DCS). Firstly, it defines a homeomorphism product space named color phase space which is a Cartesian product of two-dimensional Euclidean space and three-dimensional color phase space. Secondly, it maps the DCS data to the color phase space in order to get a system color-spectrum which displays the inherent relationship of the whole system. Then, it diagnoses the system fault by observing the color change on the color-spectrum depending on the physiological characteristics that human's eyes are more sensitive for the color change than data change.

FAULT DIAGNOSIS FOR OSCILIATING COMB

Fault diagnosis is generally considered. Its technology on basis of time series analysis is stressed. And the oscillating comb of F.O.R.comber is tested, diagnosed and researched.

基于定子电流法的绕线转子无刷双馈发电机偏心故障研究

绕线转子无刷双馈电机在船舶轴带发电以及风力发电等领域中具有良好的应用前景。转子偏心是一种常见的机械故障,威胁着现代化工业体系的稳定运转,因此实现其状态检测和故障诊断非常关键。由于无刷双馈电机具有特殊的定、转子绕组结构以及复杂的气隙磁场,因此传统的故障检测方法无法直接用于该类电机。该文通过定子电流法提取了功率绕组电流的特征频率分量作为故障指标,实现对静偏心、动偏心以及混合偏心的无创诊断。建立故障电机的时步有限元模型,并通过详细的故障机理分析,得出不同类型的偏心故障可能导致的特征频率信号。最后,以一台2/4对极无刷双馈电机为例,并通过有限元仿真、样机试验验证解析法的结论,为绕线转子无刷双馈电机偏心故障的精准检测提供了理论依据。

适用小样本的并网光伏阵列故障诊断方法

该文提出一种适用小样本的并网光伏阵列故障诊断方法。首先,使用光伏阵列的稳态输出电信号时间序列构建特征向量,论证该特征向量可以表征正常、开路故障、短路故障、阴影等不同状态。其次,针对光伏阵列运行环境多变的情况,提出一种将实际气象条件下光伏阵列输出值转换到统一工况下的数据处理方法。然后,为适用小样本情况,将线性判别分析方法与有偏差的协方差估计、公共奇异值分解相结合,解决样本高维低量导致的样本协方差矩阵估计奇异和判别函数求解困难的问题。最后,在上海市某高校楼顶搭建实验平台,采集光伏阵列不同状态下的实验数据,验证了所提数据处理方法对使用稳态电信号的必要性,及该故障分类算法在小样本场景中的有效性。

BSPST:形变监测仪器故障分类算法

针对现有形变监测仪器发生故障时故障类别难以准确分类的问题,提出了一种基于最大区分子序列(Shapelet)转换的时间序列分类算法(best qualify Shapelet Transform,BSPST)。为了提升Shapelet质量,利用布隆过滤器和相似度匹配保留一组高质量的候选Shapelet来构建分类模型,BSPST利用布隆过滤器筛选出同类别中重复的符号聚合近似(symbolic aggregation approximation,SAX)单词。随后通过位图中标记的单词来评价SAX单词的重复度,以此去除类别中相似的SAX单词。最终将处理后的符号聚合近似单词转化为高质量的Shapelet。通过Shapelet转换技术,对数据进行转换。最后采取集成分类器进行分类。根据地震形变仪器故障数据建立了7个地震设备故障数据集,并结合东安格利亚大学和加州大学河滨分校时间序列分类仓库中选取的44个数据集和具代表性的最先进的方法进行了充分的实验验证。结果表明,BQST算法在分类精度、分类速度上稳居前列,有效解决了形变监测仪器的故障分类问题。

基于Transformer的分段供电故障诊断方法

分段供电系统是长初级直线电机的重要组成部分,对其应用有效的故障诊断方法有助于电机的正常工作和故障检修。根据实测三相电流波形特点,提出一种结合幅值特征序列提取算法和Transformer深度学习模型的分段供电故障诊断方法。除此之外,引入深度学习模型的解释性方法,实现模型对异常进行自监督辅助定位。最后,将以上方法在某型分段供电直线电机的试验数据上进行应用,并验证这些方法的有效性和可靠性。

强噪声干扰下采煤机行星齿轮故障诊断方法

采煤机摇臂截割部行星齿轮的健康状态直接影响截割效率。针对采煤机截割煤岩过程中受多重冲击引起的强噪声干扰、齿轮结构复杂且传递路径多变导致故障特征难以提取等特点,提出了一种基于频谱平均降噪和相关谱的采煤机行星齿轮故障诊断方法。根据信号频谱分布特征及噪声随机特性,采用频谱平均降噪方法抑制噪声对信号频谱的干扰,获得信号降噪频谱。构建相关谱以建立少样本降噪频谱和多样本降噪频谱的内在联系,减少频谱平均降噪对样本数量的需求。采用一维卷积神经网络(1D CNN)建立相关谱与故障类别之间的精确映射关系,以相关谱为输入、故障类别为输出,实现行星齿轮故障分类识别。在DDS传动系统故障诊断实验台对基于频谱平均降噪和相关谱的采煤机行星齿轮故障诊断方法进行实验验证,结果表明该方法能够增强表征故障特征的关键频率,对正常、断齿、磨损、缺齿和裂纹5种行星齿轮健康状态信号的整体识别率达96%,在信噪比不低于15 dB时可有效、准确地实现齿轮故障诊断。

DHSI筛选奇异值分量在齿轮故障诊断中的应用

为有效剥离传动系统齿轮故障信号中的噪声成分,提出基于差分谐波显著指数(Differential Harmonic Significance Index,DHSI)筛选奇异值分量的齿轮故障诊断方法。该方法首先对原始信号构造Hankel矩阵,并对该矩阵进行奇异值分解,然后利用提出的一种新的奇异值突变位置判别指数,即奇异分量的差分谐波显著指数筛选奇异值的个数,并由这些奇异值分量重构信号,得到故障信号的包络谱。应用该方法分析齿轮故障仿真信号以及某型直升机传动系统并车级齿轮掉块故障信号,与基于奇异值差分谱的奇异值分量筛选结果对比表明,基于差分谐波显著指数的奇异值分量筛选能够更好地消除噪声并提取齿轮振动信号中的故障特征。

改进时序灰度图和深度学习的齿轮箱故障诊断

针对齿轮箱实际工作环境复杂、传统方法提取特征以及灰度图提取特征性能不足,提出了一种基于改进时序灰度图和深度学习的齿轮箱故障诊断方法。用EEMD(ensemble empirical mode decomposition)将振动信号分解为若干个本征模态分量(IMF)后,通过累计均值准则将IMFs划分为高频和低频分量,其中高频分量采用小波阈值降噪进行处理;重构降噪后的高频IMFs与低频IMFs,并利用灰度图方法对重构信号进行编码。将二维改进时序灰度图送入卷积神经网络进行训练,以发挥卷积网络对图片特征提取优势,并由混淆矩阵显示结果。最后将模型结果和不同灰度图与传统诊断方法进行对比。结果表明:相对于普通灰度图、全局去噪灰度图,所提方法对齿轮箱故障诊断准确率分别提高4、1.8个百分点,且收敛速度明显加快;相对于BP神经网络以及ELM诊断方法,所提方法对齿轮箱故障诊断准确率显著提高。

基于强化层次模糊熵的柴油机故障诊断方法

针对多尺度模糊熵(multi-scale fuzzy entropy,简称MFE)算法中多尺度化过程采用滑动均值滤波器导致原始信号高频信息丢失的问题,提出强化层次模糊熵方法(enhanced hierarchical fuzzy entropy,简称EHFE),用于表征原始信号中富含的高低频故障模式信息。结合萤火虫算法优化支持向量机(firefly algorithm optimized support vector machine,简称FAOSVM),提出一种基于EHFE和FAOSVM的柴油机故障诊断方法。柴油机试验数据对比分析表明:相比于现有方法,所提出方法能够充分表征柴油机故障信号富含的模式信息,并且能够有效识别柴油机正时齿轮故障,识别精度达到99.6%,在极小样本下也能达到较好的识别精度。

基于时间序列特征提取的光伏组件老化故障诊断方法

通过分析老化故障的产生和演化机理以及对热斑、隐裂和电势诱导衰减(PID)故障时间序列特性的差异性比较,确定老化故障时间序列具有独特的变化规律;通过光伏组串等效电路模型参数计算,研究和验证这种变化规律对模型参数的影响作用和相关性,确定老化故障诊断特征向量;采用模糊C均值聚类算法,提出基于时间序列特征提取的光伏组件老化故障诊断方法。仿真和实验结果表明:所获得的模型参数计算结果能很好地描述时间序列的特性变化;所确定的故障诊断特征量,能有效表征老化故障的发生和演化过程;所提出的故障诊断方法能可靠地实现老化故障判定、程度等级划分和程度估算。

基于轻量化神经网络的齿轮箱声信号可视化故障诊断

齿轮箱是机械装置的主要部件之一,其中的齿轮因为长期在复杂环境中工作而容易产生故障。利用声波图谱的方式来对齿轮箱故障进行诊断有着无需停机、非接触式的优点。因此,如何提高声波图谱方法的故障诊断效率和诊断精度尤为重要。对采集到的齿轮箱声学信号进行小波变换从而得到时频谱图。提出了一种以MobileNetV3为骨干网络(backbone)进行特征提取,使用改进单次检测器(SSD)进行特征检测的轻量化故障特征诊断网络GSLDN。利用GSLDN识别时频谱图的故障特征,依据转动构件运动原理提出了告警判别模型。通过工程实例分析,对比了YOLOv5_S特征检测模型与GSLDN的识别精确度和运行效率,发现GSLDN有着近似于YOLOv5_S的识别精度,但在参数量和运行时间上大幅度降低。所提出的GSLDN以及告警判别模型能够保证信噪比在大于-7.33 dB时仍然能够准确判断故障发生。

基于叶尖定时技术的轮盘外缘裂纹故障分类方法研究

针对轮盘外缘裂纹信号的故障分类需求,设计了基于Simulink的叶尖定时信号仿真模型,利用MATLAB对不同裂纹程度进行分类。通过ANSYS有限元分析软件建立叶片轮盘系统模型,计算不同转速下叶片的固有频率,得到叶片动频与转速频率的函数关系,根据叶片振动理论,利用Simulink设计叶片振动系统仿真模型,考虑转子转速波动和不平衡振动对叶片振动的影响,建立对应的仿真模型,得到含扰动量的叶尖定时信号。对叶尖定时信号进行频谱分析,通过Simulink设计数字滤波器,去除信号中含扰动量的频率成分,构造不同裂纹程度的叶尖定时信号数据集,将其去噪后导入MATLAB分类学习器中,对不同分类方法进行对比并测试。结果表明,此方法可对不同裂纹程度的叶尖定时信号进行准确分类,为轮盘外缘裂纹故障分类方法研究提供参考。

基于时频分析和迁移学习的齿轮故障诊断方法

针对齿轮故障诊断问题,提出了基于平滑伪Wigner-Ville时频分布和AlexNet迁移学习的故障诊断方法。原始齿轮故障信号是一维振动信号,通过平滑伪Wigner-Ville时频分析+灰度级-RGB变换伪彩色增强变换可以转化为二维时频图像,从而获得比一维振动信号更直观和丰富的特征;进而利用AlexNet神经网络模型进行迁移学习,显著提升开发效率。采集了齿轮系统正常运转、3种单一故障(齿面点蚀、齿面磨损、断齿)和2种复合故障(点蚀+磨损、断齿+磨损)共6种特征的振动信号,构造了齿轮故障数据集,并通过AlexNet迁移学习获得迁移AlexNet齿轮故障诊断模型。研究结果表明,提出的基于时频分析和迁移学习诊断方法对齿轮故障诊断的总体准确率为95.79%,因此可用于工业现场的齿轮系统故障诊断。同时,试验结果表明,齿面磨损单一故障和含有齿面磨损的复合故障造成所提方法的诊断正确率相对较低,说明对于齿面磨损的信号特征的研究和提取方法仍有进一步提升的空间。