基于声发射和神经网络的风机叶片裂纹识别研究

提出一种对风力机叶片裂纹声发射信号进行模式识别的方法。该方法以叶片无裂纹、萌生裂纹、扩展裂纹和断裂四个阶段为声发射源的四个模式,基于声发射信号含有丰富的发射源信息的特点,通过大量采样获得叶片裂纹声发射信号参数,并依照叶片裂纹声发射参数分析的数值特点确定BP神经网络,用选定的网络对叶片裂纹阶段进行模式识别,以判断裂纹的危害程度。仿真结果表明,利用BP神经网络可以对声发射信号进行有效识别,识别准确率达到90%以上。     

Artificial neural network a tool for predicting failure strength of composite tensile coupons using acoustic emission technique

A series of 18 tensile coupons were monitored with an acoustic emission (AE) system, while loading them up to failure. AE signals emitted due to different failure modes in tensile coupons were recorded. Amplitude, duration, energy, counts, etc., are the effective parameters to classify the different failure modes in composites, viz., matrix crazing, fiber cut, and delamination, with several subcategories such as matrix splitting, fiber/matrix debonding, fiber pullout, etc. Back propagation neural network was generated to predict the failure load of tensile specimens. Three different networks were developed with the amplitude distribution data of AE collected up to 30%, 40%, and 50% of the failure loads, respectively. Amplitude frequencies of 12 specimens in the training set and the corresponding failure loads were used to train the network. Only amplitude frequencies of six remaining specimens were given as input to get the output failure load from the trained network. The results of three independent networks were compared, and we found that the network trained with more data was having better prediction performance.     

基于声发射和神经网络的木材受力损伤过程检测

对受力木材的声发射信号进行检测与分析,实现木材损伤过程的检测和监控;研究过程中采用时间序列结合神经网络建模方法,对声发射信号的累积能量时间序列和载荷时间序列进行仿真和预测.利用神经网络建模对声发射信号的累积能量时间序列进行预测,模型对44个样本进行检验的最大误差为5.6%,而且误差较大的样本对应声发射振幅参数值的局部极大值,而振铃计数与声发射率则对应局部极小值.对相应载荷时间序列预测分析中,模型输出与目标输出的最大误差不到0.1%.结构为5×5×1的网络能很好地根据序列前5个值准确预测即将发生的声发射累积能量值;结构为6×5×1的网络能很好地根据序列前6个值精确预测木材即将承受的载荷.     

基于神经网络的滚动轴承故障诊断方法的研究

神经网络是一种具有非线性映射能力强以及自学习、自组织、自适应等优点的智能方法,非常适合于滚动轴承的故障诊断。针对滚动轴承是机械设备重要的易损零件之一,大约有30%的故障是由轴承损坏引起的,提出了基于神经网络的滚动轴承故障诊断方法。以滚动轴承小波分解后的能量信息作为特征,通过神经网络作为分类器对滚动轴承故障进行识别、诊断。实验表明,该方法对于滚动轴承的故障诊断具有良好的效果和应用价值,并可方便地推广到其他类似的诊断领域。     

基于神经网络的滚动轴承振动预测

以神经网络与MATLAB实现理论为依据,提出了一种新的滚动轴承振动预测方法。这种方法根据轴承的加工质量试验数据,建立轴承振动预测的BP网络试验模型,在MATLAB开发环境下输入训练样本数据矩阵和目标矩阵。经过训练后,网络误差达到要求,预报结果的最大相对误差小于10％。     

基于TL-LSTM的轴承故障声发射信号识别研究

针对多工况下滚动轴承故障声发射信号智能识别问题,提出了一种长短时记忆网络(LSTM)与迁移学习(TL)相结合的故障识别新方法。该方法仅以单一工况下原始声发射信号参数作为训练样本,构建LSTM模型充分挖掘出声发射信号与故障之间的深层次映射关系,以识别与训练工况具有相近分布特征的其他工况下故障;引入并结合TL来应对相异分布特征的其它工况下故障识别问题,从而可完成多种类型工况下故障特征的自适应提取与智能识别。实验结果表明,对于转速、采集位置或滚动轴承型号工况改变时内圈、外圈及保持架故障的识别均具有较高的准确率,可端对端的实现多种类型工况下故障的实时在线智能监测任务,摆脱了对先验故障数据的过分依赖,验证了该方法的可行性与优越性。     

A neural network based approach for background noise reduction in airborne acoustic emission of a machining process

Tool wear prediction has become an indispensable technique to prevent downtime in manufacturing and production processes. Airborne emission from a machining process using a low-cost microphone may provide a vital signal of tool health. However, the effect of background noise results in anomaly in data that may lead to wrong prediction of tool health. The paper presents an adaptive approach using neural networks for background noise filtration in acoustic signal for a turning process. Acoustic signal of a turning process is mixed with background noise from four different machines and introduced at different RPMs and feed-rate at a constant depth of cut. A comparison of Backpropagation neural network (BPNN), Self-organizing map and k-means clustering algorithm for noise filtration is investigated in this paper. In this regard, back-propagation neural network showed better performance with an average accuracy for all the four sources. It shows 100 % accuracy for grinding machine signal, 94.78 % accuracy for background signal from 3-axis milling machine, 45.57 % and 12.69 % for motor and 4-axis milling machine, respectively. Signal reconstruction is then done using Discrete cosine transform (DCT). The proposed technique shows a promising future for noise filtration in airborne acoustic data of a machining process.     

基于径向基函数神经网络的电机轴承智能故障诊断

本文提出了一种新的电机轴承故障诊断方法.首先把滚动轴承振动信号作为识别故障的特征向量,然后送入径向基函数神经网络中,进行故障类别的自动识别.试验结果表明,该诊断模型对电机轴承故障诊断具有良好的诊断效果,系统不仅能够检测到轴承故障的存在,而且能够比较准确地识别轴承的故障模式.     

A comparative experimental study on the use of acoustic emission and vibration analysis for bearing defect identification and estimation of defect size

Vibration monitoring of rolling element bearings is probably the most established diagnostic technique for rotating machinery. The application of acoustic emission (AE) for bearing diagnosis is gaining ground as a complementary diagnostic tool, however, limitations in the successful application of the AE technique have been partly due to the difficulty in processing, interpreting and classifying the acquired data. Furthermore, the extent of bearing damage has eluded the diagnostician. The experimental investigation reported in this paper was centred on the application of the AE technique for identifying the presence and size of a defect on a radially loaded bearing. An experimental test rig was designed such that defects of varying sizes could be seeded onto the outer race of a test bearing. Comparisons between AE and vibration analysis over a range of speed and load conditions are presented. In addition, the primary source of AE activity from seeded defects is investigated. It is concluded that AE offers earlier fault detection and improved identification capabilities than vibration analysis. Furthermore, the AE technique also provided an indication of the defect size, allowing the user to monitor the rate of degradation on the bearing; unachievable with vibration analysis.     

Application of cepstrum and neural network to bearing fault detection

This paper proposes an integrated system for motor bearing diagnosis that combines the cepstrum coefficient method for feature extraction from motor vibration signals and artificial neural network (ANN) models. We divide the motor vibration signal, obtain the corresponding cepstrum coefficients, and classify the motor systems through ANN models. Utilizing the proposed method, one can identify the characteristics hiding inside a vibration signal and classify the signal, as well as diagnose the abnormalities. To evaluate this method, several tests for the normal and abnormal conditions were performed in the laboratory. The results show the effectiveness of cepstrum and ANN in detecting the bearing condition. The proposed method successfully extracted the corresponding feature vectors, distinguished the difference, and classified bearing faults correctly.     

基于遗传算法的BP神经网络滚动轴承故障诊断

针对滚动轴承的故障诊断问题,提出了一种基于遗传算法的BP神经网络滚动轴承故障诊断方法。以BP神经网络的误差为目标函数,利用遗传算法进行BP神经网络的权值和阈值优化,并用优化后的BP神经网络进行故障诊断。通过MATLAB仿真,结果表明遗传算法优化的BP神经网络相比传统的BP神经网络具有更好的诊断效率和准确度。     

Artificial neural network based model for computation of injection mould complexity

The cost of injection mould construction depends primarily on the mould complexity. The ability to estimate the mould complexity before releasing the final drawings for construction purposes will greatly help the designers to understand the implications of their design on cost. Mould complexity depends on several factors such as part geometry, parting line, materials, and number of cavities per mould. In most industries, the mould complexity evaluation is performed manually based on past experiences of mould makers. Faced with a shortage of experienced mould makers, there is a pressing need for development of computer-aided tools for mould complexity evaluation. In this study, a neural network-based design tool for computing the mould complexity index, which represents the degree of difficulty of mould manufacturing, has been developed and implemented using a 14-3-1 backpropagation network running on the CNAPS neuro-computer.     

Fault diagnosis of rolling element bearing based on artificial neural network

Journal of Mechanical Science and Technology - This paper proposes the expert system for accurate fault detection of bearing. The study is based upon advanced signal processing method as wavelet...     

基于神经网络的管道泄漏声波信号特征识别

文中采用小波包和神经网络结合起来用于声波信号诊断的方法来提高泄漏检测的准确率.该方法首先对声波信号进行小波包分解,将各频带内的分解系数重构,得到在每个分解节点上构成的新时间序列的参数,将这些参数通过BP神经网络进行智能识别,来区分故障原因.文章最后对现场实验数据及其信号分析结果进行了研究,实验结果表明,该方法可以有效地提高管道泄漏信号识别的准确度.     

Artificial neural network based on genetic learning for machining of polyetheretherketone composite materials

In this paper an artificial neural network (ANN) aiming for the efficient modelling of a set of machining conditions for orthogonal cutting of polyetheretherketone (PEEK) composite materials is presented. The supervised learning of the ANN is based on a genetic algorithm (GA) supported by an elitist strategy. Input, hidden and output layers model the topology of the ANN. The weights of the synapses and the biases for hidden and output nodes are used as design variables in the ANN learning process. Considering a set of experimental data, the mean relative error between experimental and numerical results is used to monitor the learning process obtaining the completeness of the machining process modelling. Also a regularization term associated to biases in hidden and output neurons are included in the GA fitness function for learning. Using a different set of experimental results, the optimal ANN obtained after learning is tested. The optimal number of nodes on the hidden layer is searched and the positive influence of the regularization term is demonstrated. This approach of ANN learning based on GA presents low mean relative errors in learning and testing phases.     

基于轻量级卷积神经网络的带钢表面缺陷识别

带钢表面缺陷识别对促进带钢生产质量提升至关重要。然而传统的图像处理与识别方法存在精度不高、且容易受到光线等因素影响;而新兴的基于深度学习的算法,则存在模型参数量大且难以部署等问题,无法在实际生产中得到广泛应用。本文提出了一种轻量级部分深度混合可分离网络(PDMSNet)用于解决以上问题,由于其模型小以及浮点型运算(FLOPs)少更易于部署在资源受限的平台。采用标准的带钢表面缺陷数据集NEU-CLS的测试结果表明,与其他缺陷分类器相比,在带钢表面缺陷检测方面,本文所提出的模型性能更加优越。识别准确率达到了99.78%,而参数量只有0.17 M以及272 M FLOPs,在单一低端的GeForce MX250图形处理单元(GPU)识别一张图片平均时间为0.47 ms,可以满足工业现场实时检测的要求。     

基于BP神经网络的轴承寿命预测程序设计

基于BP神经网络原理,利用轴承空载和负载运行下的各参数建立轴承剩余寿命的预测模型。在MATLAB中对轴承数据样本进行学习与训练,获得较精确的BP神经网络结构的权值和阈值,根据BP神经网络算法编写m函数文件,在MATLAB中生成COM组件。用Visual Basic 6.0编写的系统软件界面,在界面中调用COM组件中的DLL文件。解决算法优化和模块间调用等问题,成功开发出轴承寿命预测系统,用于机电产品的质量控制管理。     

Rolling bearing fault convolutional neural network diagnosis method based on casing signal

Affected by the transmission path, it is very difficult to diagnose the vibration signal of the rolling bearing on the aircraft engine casing. A fault diag     

基于一维卷积神经网络的轴承剩余寿命预测

现有的轴承剩余使用寿命预测模型多依赖于对轴承的时域特征或频域特征进行降维后构建特征工程,然而可能丢失重要的信号信息,因此尝试利用轴承的振动水平加速度信号和垂直加速度信号,构建一维卷积神经网络实现对特征的自动提取,无需人工提取特征,并且基于人工神经网络的局部连接和参数共享机制,大大减少了训练参数,减少了训练时间。训练模型之前,通过设置轴承样本的开始退化点,使训练样本的剩余使用寿命值更为准确。研究发现,该模型能较为准确的对轴承的退化状态进行预测,进而预测剩余寿命。数据集来自于FEMTO-ST研究所的轴承退化数据集。     

基于分支卷积神经网络的托辊轴承故障分级诊断研究

在对矿山机械装备中使用的轴承进行故障诊断时,易受噪声干扰及多变工况的影响,同时也难以适应不同诊断任务,针对这一系列问题,提出了一种基于分支卷积神经网络(B-CNN)的托辊轴承故障分级诊断方法。首先,根据具体的诊断任务故障的层级结构进行了划分,采用多层标签表示健康状态、故障类型和损伤程度;通过交替卷积和池化层,构建了一维卷积神经网络(1DCNN)特征提取块;然后,将层级结构和特征提取块融合,设计出了一种基于分支一维卷积神经网络(B-1DCNN)的轴承故障分级诊断模型;最后,使用美国凯斯西储大学轴承数据和自建的带式输送机托辊故障模拟实验台数据,对托辊轴承故障进行了模拟实验,对该方法在噪声干扰和多变工况下的诊断性能进行了验证。研究结果表明：该方法成功实现了对托辊轴承故障从粗到精的分级诊断,对噪声干扰和变工况具有较好的鲁棒性,且与支持向量机(SVM)和反向传播神经网络(BPNN)模型相比,该方法的故障诊断性能更好。