基于模糊神经网络与主成分分析的磨削表面粗糙度在线预测

基于声发射和振动信号提出了一种模糊神经网络和主成分分析的表面粗糙度预测方法,以提高磨削过程中工件表面粗糙度识别的准确性。首先,采集磨削程中声发射与振动信号,提取相关时域特征、频域特征和小波包特征参数,利用主成分分析对特征量进行降维优化;然后,构建表面粗糙度模糊神经网络预测模型,将信号特征量与表面粗糙度作为模糊神经网络的输入和输出;最后,对模型进行训练,并对表面粗糙度预测精度进行验证。实验结果表明:通过主成分分析(PCA)方法对声发射和振动信号特征量进行降维得到5个主成分,以此建立的模糊神经网络表面粗糙度预测模型的效果精度可达到91%以上,与局部线性嵌入和多维标度法降维方法相比,PCA方法降维后的特征所含信息更优,预测准确度更高。     

Tool condition monitoring by acoustic analysis through a solid path

A sensing method using an acoustic signal obtained in a relative low frequency range through a solid path for the monitoring of tool wear has been investigated. Such acoustic signals could be in the form of stress waves that are released during a machining process, which can be picked up by a regular ferroelectric microphone. Data analysis was conducted in both time and frequency domains. A clear pattern in such signals corresponding to the tool wear conditions has been identified. Several components in spectra were found in the pattern for indicating sudden changes of tool wear or breakage occurring at major cutting edges. It was also observed that the RMS and variance values of the signals could indicate the specific wear condition of the tool. Therefore, this kind of acoustic signal carries sensitive information about the progress of tool wear and can be implemented on line for monitoring tool wear.     

融合经验模态分解与时频分析的单通道振动信号分离研究

通过单通道振动信号分离研究探讨机械振动源信号提取问题。采用集合经验模态分解方法将单通道信号构造成多通道信号,再通过主分量分析方法得到多通道构造信号的特征值分布情况并以此进行源信号数目估计,进而利用基于时频分析的盲源分离技术获取振动源信号。实验表明,该方法能有效实现单通道振动信号分离,具有较强实际应用价值。     

振动环境中螺栓连接结构状态声发射试验研究

螺栓连接应用广泛,螺栓松动可能造成灾难性的后果。结合声发射技术,在振动台上进行螺栓连接状态检测初步试验。对试验中所采集的大量声发射信号进行分析,找出不同扭矩下螺栓连接状态所对应的信号特征,实现对连接螺栓松动状态的辨识。     

面向机械振动信号采集的无线传感器网络节点设计

针对现有无线传感器网络节点的不足和机械振动信号采集的需求,设计了一种可实现机械振动信号采集和片上处理的无线传感器网络节点。该节点采用MEMS加速度传感器拾取振动信号,以低功耗FPGA芯片为控制核心,采用了可编程片上系统技术,将数据采集、存储、处理和传输等控制功能模块集成到单个FPGA芯片中。本文着重介绍了节点数据采集、存储和传输部分的硬件电路和控制逻辑设计,对比实验表明该节点能有效采集和监测机械振动信号。     

Laplace小波相关滤波法与冲击响应提取

冲击响应信号的出现标志着机械设备发生松动，碰撞，冲击等故障。在复杂的振动信号中准确提取出冲击响应信息在机械故障诊断中具有重要的意义，介绍了用来识别冲击响应信号的Lapalce小波相关滤波法，通过对含有噪声的模拟冲击响应信号的识别，证明该方法能够在强大噪声干扰中准确捕捉到冲击响应信号。工程应用结果表明，相关滤波法可以从复杂的内燃机缸盖振动信号中准确定位进气阀关闭时冲击发生的时刻及频率，从而成功地诊断出因进气阀磨损而导致的漏气故障。     

基于盲源分离和AR谱估计的旋转机械故障诊断方法

准确的AR模型能够较好地揭示信号中蕴含的状态特征变化的信息,然而,AR模型对系统的状态变化十分敏感,多个动态变化的源信号的耦合必然会影响其估计结果。基于此,提出了一种基于盲源分离和AR谱估计的旋转机械故障诊断方法。首先,利用盲源分离的方法从混合观测信号中恢复各机械振动源信号;然后,将非平稳性的故障信号通过经验模态分解得到各本征模态函数;最后,对经验模态分解得到的平稳的本征模态函数进行AR谱估计,提取振动信号的故障特征信息。通过仿真研究和实验分析验证了该方法在旋转机械故障诊断中的有效性和可行性。     

基于VMD-PNN的砂轮钝化声发射检测

在磨削加工过程中,加工刀具即砂轮会发生钝化现象,砂轮表面磨损影响加工精度和工件质量,需要及时检测并修整.磨粒的塑性变形、破碎、断裂等会产生声发射信号,能够作为精确识别砂轮钝化状态的依据,且不易被噪声干扰,因此提出一种基于变分模态分解(Variational Mode Decomposition,VMD)和概率神经网络(...     

基于声发射信号EMD-WPD特征融合的航天器在轨泄漏辨识方法

长期运行在空间环境中的航天器可能由于撞击、振动、老化等因素而发生气体泄漏,在轨泄漏辨识对航天器安全保障具有重要意义.提出了一种基于声发射信号经验模态分解(empirical mode decomposition,EMD)和小波包分解(wavelet packet decomposition,WPD)特征融合的航天器泄漏...     

基于BSS的含噪声机械振动信号分离研究

水下航行器的噪声源识别面临的两个问题(:1)无法获得振源信号(,2)测得振动信号有环境噪声影响且振源之间相互耦合。将环境噪声作为一个独立的噪声源,给出瞬时混合信号的盲源分离(BSS)数学模型;利用基于二阶统计特性的两次去相关盲源分离算法,对机械振动加白噪声的混合信号和水池试验实测混合信号进行分离;通过试验验证两次去相关盲源分离方法可以用来解决上述问题。     

基于反馈混沌化方法的多线谱控制技术研究

船舶辐射水声中的线谱成分是被动声纳在水声对抗中检测、跟踪和识别目标的主要特征信号,混沌线谱控制技术利用混沌系统响应功率谱为宽频谱的特殊性质,把集中于某一频率的能量分散到较宽的频带上来抑制线谱成分,提高船舶的声隐身性能。利用反馈混沌化理论,对多谐波激励条件下隔振系统的混沌化进行了研究,提出了多谐波激励多自由度隔振系统的反馈混沌化方法,掌握了隔振系统进入混沌的条件和途径,突破了激励频率增加引起隔振系统混沌运动难以保持的技术难题。根据反馈混沌化原理设计了双层隔振系统试验装置,开展了隔振系统多线谱控制试验研究,试验结果证明了该方法的有效性,可作为混沌隔振系统设计的理论和试验依据。     

A Method for Incipient Fault Diagnosis of Roller Bearings Based on the Wavelet Transform Correlation Filter and Hilbert Transform

Noise is the biggest obstacle that makes the incipient fault diagnosis results of roller bearings uncorrected; a new method for diagnosing incipient fault of roller bearings based on the Wavelet Transform Correlation Filter and Hilbert Transform was proposed. First, the weak fault information features are picked up from the roller bearings fault vibration signals by use of a de-noising characteristic of the Wavelet Transform Correlation Filter as the preprocessing of the Hilbert Envelope Analysis. Then, in order to get fault features frequency, de-noised wavelet coefficients of high scales which represent high frequency signal were analyzed by Hilbert Envelope Spectrum Analysis. The simulation signals and diagnosing examples analysis results reveal that the proposed method is more effective than the method of direct wavelet coefficients-Hilbert Transform in de-noising and clarifying roller bearing incipient fault.     

Trennung von Schallemissionssignalen durch rechnerische Analysen

Separation of Acoustic Emission Signals by Computerized Analyses Features extracted from acoustic emission signals by means of different analysis methods can be evaluated by the application of pattern recognition computer programs. By these methods it can be established, whether the employed features – maximum amplitude, signal rise time, pulse sum, pulse area, pulse energy and mean amplitude as well as simple characteristics of frequency spectra – are capable to distinguish acoustic emission signals or whether combinations of features lead to better results when single features fail. The efficiency of different analyses and the pattern recongnition programs is examined by four simulated types of signals. According to the employed analysis method there are distinct differences regarding the separation of acoustic emission. However, by applying feature combinations signal groups not separable by single analysis methods can be distinguished.     

基于CHMM的齿轮箱状态识别研究

针对离散隐Markov模型(HMM)在状态识别中的不足,结合齿轮箱全寿命实验数据,研究了基于连续隐Markov模型(CHMM)的状态识别方法。建立了基于齿轮箱原始振动信号的CHMM状态识别框架,提出了基于K均值算法和交叉验证相结合的状态数优化方法,通过计算待确定观测数据的极大似然概率值来确定齿轮箱当前状态。结果表明,用原始振动信号作为CHMM的输入可以实现状态识别,验证了模型的有效性,为齿轮箱基于状态的维修提供了科学依据。     

Quantification of failure mechanisms in mode-I loading of fiber reinforced plastics utilizing acoustic emission analysis

Acoustic emission signals originating from interlaminar crack propagation in fiber reinforced composites were recorded during double cantilever beam testing. The acoustic emission signals detected during testing were analyzed by feature based pattern recognition techniques. In previous studies it was demonstrated that the presented approach for detection of distinct types of acoustic emission signals is suitable. The subsequent correlation of distinct acoustic emission signal types to microscopic failure mechanisms is based on two procedures. Firstly, the frequency of occurrence of the distinct signal types is correlated to different specimens’ fracture surface microstructure. Secondly, a comparison is made between experimental signals and signals resulting from finite element simulations based on a validated model for simulation of acoustic emission signals of typical failure mechanisms in fiber reinforced plastics. A distinction is made between fiber breakage, matrix cracking and interface failure. It is demonstrated, that the feature values extracted from simulated signals coincide well with those of experimental signals. As a result the applicability of the acoustic emission signal classification method for analysis of failure in carbon fiber and glass fiber reinforced plastics under mode-I loading conditions has been demonstrated. The quantification of matrix cracking, interfacial failure and fiber breakage was evaluated by interpretation of the obtained distributions of acoustic emission signals types in terms of fracture mechanics. The accumulated acoustic emission signal amplitudes show strong correlation to the mechanical properties of the specimens. Moreover, the changes in contribution to the different failure types explain the observed variation in failure behavior of the individual specimens quantitatively.     

人体长骨振动的在体测试

振动分析技术用来诊断骨病和估计骨折愈合程度的可行性已得到证实。测试时常用的激振方式有电动、人工和超声等，拾振装置有传声器、力传感器和加速度计等。文章采用医用叩诊锤激振、双加速度计拾振，在不同情况下在体测试了人体长骨的振动。通过分析测试信号的功率谱及传递函数，发现骨组织中振动传播具有方向性，拾振器的质量、测点及激振点位置对测试结果有影响。实际应用中应选用质量较小的拾振器，且应尽量保持待测肢与对照肢的测试条件一致。     

基于粒子滤波的多模态振动信号在线跟踪

针对多模态振动信号的在线监测和跟踪,提出基于随机子空间（SSI）和粒子滤波（PF）算法的仿真振动信号在线监测和跟踪方法。通过SSI算法提取得到振动系统的模态主频和阻尼比,根据振动系统模型模态主频和阻尼比的计算公式,得到系统的状态矩阵和输出矩阵。将计算所得状态矩阵和输出矩阵代入状态方程,利用PF算法进行信号的在线监测和跟踪,实现信号的降噪处理和预测分析。对于大型机械、桥梁等建筑物,对其进行在线监测保障其正常营运对社会经济发展具有深远影响。文中利用SSI算法提取系统的模态参数,进一步构建振动系统的状态矩阵和输出矩阵,并利用PF算法进行信号滤波抑噪和预测,在此基础上可以对结构状态实施在线监测及预警控制,实际大桥斜拉索振动信号测试也表明本文算法可以提供稳定可靠的信号跟踪与预测技术。     

阵列声波测井信号调理与首波提取技术研究

本文分析阵列声波测井中声波全波列信号的特点，采用前置通道信号调理技术对声波信号进行预处理以便于首波提取，并阐述一种首波到时的提取技术及实现算法。前置通道信号调理电路的设计采用了自动增益控制技术，可实现增益的自动调节。首波提取采用了短窗一长窗能量比算法，能够精确地检测到首波。     

机械振动信号内在模式的复杂性评估

针对机械振动信号非平稳特性,利用经验模式分解法将其分解为若干个内在的振荡模式（即基本模式分量）,从而使得不同的基本模式包含有不同的设备状态信息。借助近似熵的概念,可定理描述原始信号和各振荡模式的复杂性,实现对机械振动信号内在模式复杂性的定理评估。该方法不仅有助于揭示和认识转子系统的复杂动力学行为,还能有效地监测系统状态的早期变化,及时捕捉机组潜在的隐患,预防故障的升级恶化。工程应用实例表明,该方法可有效提取机组的故障信息,从而为机械设备状态监测和故障诊断提供一种行之有效的新方法。     

基于振动分布特征的电力变压器绕组故障诊断

振动分析法是实现电力变压器带电监测与故障诊断的重要手段,而基于振动分析法的故障诊断方法的关键在于从复杂的油箱壁振动信号中提取出状态特征(值或矢量)。传统的状态特征提取方法大多选取单个测点的振动信号进行时域或频域特征的提取,往往忽略了各测点间的振动分布特征。从振动重心的角度对振动分布的幅值重心及重心轨迹进行研究与分析,能够提出四个量化参数。在四个量化参数的基础上结合支持向量机分类算法提出基于振动分布特征的变压器绕组故障诊断模型。实际变压器的绕组故障实验以及十余台台电力变压器现场实测数据样本的分析与测试结果均表明,提出的振动分布特征及量化参数能够有效反映变压器绕组变形、压紧力松弛等机械结构变化,而基于振动分布特征的绕组故障诊断模型也可准确的对变压器绕组机械结构状态进行检测与诊断。