基于最大-最小蚂蚁系统优化ELM的电解槽故障诊断

针对铝电解槽故障特征种类繁多,难以快速准确的实现故障类型诊断,设计了一种基于最大-最小蚂蚁系统(MMAS)优化的极限学习机(ELM)故障诊断方法。介绍了电解槽常见的故障类型及其对槽电压的影响,对采集到的故障情况下的槽电压信号进行降噪处理,根据对降噪后故障信号的局域均值分解(LMD)结果得到故障特征。采用ELM算法辨识故障类型,针对ELM算法存在的参数问题,采用MMAS对ELM隐含层参数寻优。结果表明,MMAS优化的ELM既保证了较快的训练速度,同时获得了更高的故障测试正确率。     

基于深度学习的故障检测方法

针对传统故障诊断方法中特征提取技术难度大、故障样本获取困难等问题，在深度学习计算框架下提出了一种半监督训练的故障检测方法，利用深度信念网络中的受限波茨曼机堆栈结构实现了数据高层特征的自动提取，结合支持向量数据描述方法实现了异常数据检测，只需利用正常工况的数据样本进行网络训练和模型拟合，无需故障样本数据，也无需人工干预进行信号特征提取，即能实现对故障数据进行的实时检测和判别。经采用标准轴承实验数据的三组故障数据进行验证，故障识别率达到100%，具有很强的工程应用价值。     

基于灰色预测和KID离群点检测的AUV故障诊断

提高故障诊断能力对于确保水下机器人AUV系统的稳定运行具有重要意义。针对水下机器人推进器系统,提出一种基于离群点检测的AUV故障检测方法。首先,将传感器采集的数据进行灰色预测处理;然后,提出了一种结合K-mean和DBSCAN的改进迭代聚类(Iterative K-mean DBSCAN,IKD)算法进行离群点检测;最后,与K-mean和DBSCAN算法相比,仿真实验结果表明基于灰色预测和KID离群点检测算法的故障检测准确率高,能够有效地实现水下机器人AUV的无监督故障诊断。     

基于改进极根学习机的回转窑煅烧带温度预测方法

针对传统算法预测回转窑煅烧带温度存在精度低、速度慢的问题,提出了基于改进极限学习机(ELM)的回转窑煅烧带温度预测方法。对ELM输入权值矩阵定义了变换系数,采用黄金分割法在给定区间内搜寻变换系数的最佳值,改进了ELM网络参数的确定方式,弥补了随机确定输入权值并且不作调整的缺陷,在保证ELM训练速度的前提下提高预测精度、减小模型随机性。实验结果表明,改进的ELM预测精度高、训练速度快、模型性能优,可满足工况恶劣的回转窑的生产需要。     

基于长短时记忆神经网络的风电机组滚动轴承故障诊断方法

风能作为一种绿色能源在我国能源结构中发挥着越来越重要的作用。风电机组的滚动轴承作为传动系统的重要组成部分，是其主要故障部件之一。随着风电规模的不断增长，及时地发现风电机组滚动轴承的故障对风电场安全稳定运行具有重要意义。针对传统回归神经网络存在的梯度消失问题，提出了利用长短时记忆神经网络对风电机组滚动轴承进行故障诊断的模型。首先，利用小波包变换对风电机组滚动轴承振动信号进行处理，提取其特征向量，将其作为长短时神经网络的输入，从而诊断出风电机组滚动轴承的三种常见故障。通过算例分析，结果表明所提出的方法能够有效地对风电机组的滚动轴承进行故障诊断，并且在故障特征量差异不明显的情况下长短时记忆神经网络仍具有良好的故障诊断性能，说明了该方法的可行性和有效性。     

基于小波包—灰色神经网络的机电作动系统故障预测

针对健康状态难以预测的机电作动系统,提出一种基于小波包—灰色神经网络的机电作动系统故障预测方法;该方法利用小波包分解对机电作动系统的故障特征进行提取,利用灰色GM(1,1)模型对提取出的12种不同故障模式下的频带值进行预测,并作为神经网络的输入,采用BP三层结构,通过验证取隐含层数为10进行训练,预测结果相对误差只有-0.014 6,取得了较好的预测效果;仿真结果表明:小波包—灰色神经网络方法在机电作动系统故障预测中具有一定的有效性。     

基于对数变换的六相永磁容错电机故障预测研究

研究电力作动系统用永磁容错电机故障预测问题,有利于准确监控飞行器健康状态,为飞机维修提供决策支持;TS(2*2永磁容错电机特征信号复杂无序,传统灰色模型故障预测精度不高,基于此,提出一种新的故障预测改进方法;TS(2*2首先对原始故障能量特征序列进行对数变换处理,对序列进行一次累加生成,建立GM(1,1)灰色模型,最后将得到的拟合还原成模拟值,得到预测数据;TS(2*2结果表明,故障原始序列经过对数函数变换处理后,预测误差相比于未经处理的基本灰色模型降低了4.63%,预测精度提高到96.5%以上,有效提高了永磁容错电机的故障预测精度。     

具有选择与遗忘机制的极端学习机在时间序列预测中的应用

针对训练样本贯序输入时的极端学习机 (ELM)训练问题,提出一种具有选择与遗忘机制的极端学习机 (SF-ELM),并研究了其在混沌时间序列预测中的应用. SF-ELM以逐次增加新训练样本的方式实现在线训练,通过引入遗忘因子以减弱旧训练样本的影响,同时以泛化能力为判断依据,对其输出权值进行选择性递推更新. 混沌时间序列在线预测实例表明,SF-ELM是一种有效的ELM在线训练模式. 相比于在线贯序极端学习机,SF-ELM具有更快的在线训练速度和更高的在线预测精度,因此更适于混沌时间序列在线预测. 关键词： 混沌时间序列 时间序列预测 神经网络 极端学习机     

基于遗传神经网络的旋转机械故障预测方法研究

许多大型旋转机械运行工况恶劣,非平稳、非线性特征明显,以及各种突发性、偶然性因素的影响,给基于振动信号处理的状态预测和状态维护分析带来困难。神经网络以其强大的处理非线性系统的能力在故障预测中得到广泛的应用,但由于其在追求高精度训练目标时易陷入局部极值,且收敛速度慢甚至发散。针对这个问题,提出了采用遗传算法对神经网络连接权值和阈值进行优化,这样不仅发挥了神经网络广泛的映射特性也使遗传算法的全局搜索优势尽显无疑。通过组合这两种算法,在提升网络学习的准确度方面,优点尤其突出,最终提高对旋转机械故障预测和寿命估计的性能,这在某环境模拟试验系统动力风机的轴承磨损故障预测中得到了验证。     

基于果蝇算法和灰色神经网络的电子装备故障预测方法研究

针对电子装备故障数据小样本、非线性的特点,在相空间重构处理原始时间序列数据的基础上,基于k折交叉验证和果蝇算法优化灰色神经网络模型参数,从而提出一种基于果蝇算法和灰色神经网络的故障预测方法,并以某型雷达高压电源监测数据仿真结果为例验证其模型性能;实验结果表明,相比已有方法,该方法在全局优化、收敛速度、预测精度方面都具有一定优势。     

利用声信号对滚动轴承进行故障诊断的研究

旋转机械在运行过程中产生的声信号包含了滚动轴承的运行状态信息,且可采用非接触式测量,本文应用它对滚动轴承进行故障诊断。基于morlet小波变换的包络分析对采集的声信号进行降噪及包络处理,然后变换到频域,提取出特征频率并经过转换后作为线性神经网路的输入向量,辨识滚动轴承的状态。实验表明,本方法对滚动轴承故障诊断是有效的。     

基于粒子滤波状态估计的滚动轴承故障识别方法

提出了一种基于粒子滤波状态估计的滚动轴承故障识别方法,该方法主要包括故障模型建立和故障识别两个步骤。在故障模型建立部分,首先依据滚动轴承不同故障状态下的振动信号,建立对应的自回归模型,作为故障模型；在故障识别部分,将正常状态下对应的模型,转化为状态空间模型,设计粒子滤波器,然后对不同的故障状态进行估计,提取其残差的相关特征,并结合模型参数特征应用BP神经网络识别算法进行故障识别。最后以美国凯斯西储大学的滚动轴承振动数据为例,验证了该方法的有效性。     

Application of correlation matching for automatic bearing fault diagnosis

This paper introduces an automatic feature extraction algorithm for bearing fault diagnosis using correlation filtering-based matching pursuit. This algorithm is described and investigated in theory and practice on both simulated and real bearing vibration signals. First, the vibration model for rolling bearing with fault is derived. Then, the numerical simulation signal being taken as an example, the principle of matching pursuit is mathematically explained and its drawbacks are analyzed. Afterward, to enhance the similarity of model related to the bearing faulty impulses, the model shape parameters are optimized using spectrum kurtosis and smoothing index. After that, the model with optimum shape and period parameters is taken as a template to approximate the impulses in faulty bearing signal. Finally, based on maximizing correlation principle, the optimized cycle parameter being as impuls e repetition period is matched up. The proposed method has been successfully applied in actual vibration signals of rolling element bearing with different faults.     

Identification of multiple transient faults based on the adaptive spectral kurtosis method

This paper proposes a multi-fault detection method based on the adaptive spectral kurtosis (ASK) analysis of the vibration signal from single sensor. A theoretical model of multiple bearing faults is established in this paper. Compared with the kurtogram and protrugram techniques, the proposed method can more effectively extract signatures of multiple bearing faults even in the presence of strong background noise. The performance of the proposed method in fault detection of the rolling element bearings is validated using simulation data and experimental signals from a bearing with multiple faults and two faulty bearings.     

A fault diagnosis scheme of rolling element bearing based on near-field acoustic holography and gray level co-occurrence matrix

Vibration signal analysis is the most widely used technique in condition monitoring or fault diagnosis, whereas in some cases vibration-based diagnosis is restrained because of its contact measurement. Acoustic-based diagnosis (ABD) with non-contact measurement has received little attention, although sound field may contain abundant information related to fault pattern. A new scheme of ABD for rolling element bearing fault diagnosis based on near-field acoustic holography (NAH) and gray level co-occurrence matrix (GLCM) is presented in this paper. It focuses on applying the distribution information of sound field to bearing fault diagnosis. A series of rolling element bearings with different types of fault are experimentally studied. Sound fields and corresponding acoustic images in different bearing conditions are obtained by fast Fourier transform (FFT) based NAH. GLCM features are extracted for capturing fault pattern information underlying sound fields. The optimal feature subset selected by improved F-score is fed into multi-class support vector machine (SVM) for fault pattern identification. The feasibility and effectiveness of our proposed scheme is demonstrated on the good experimental results and the comparison with the traditional ABD method. Considering test cost, the quantized level and the number of GLCM features for each characteristic frequency is suggested to be 4 and 32, respectively, with the satisfactory accuracy rate 97.5%.     

基于XRF-CNN土壤重金属Zn元素含量预测模型研究

结合X射线荧光光谱法,针对土壤中重金属元素Zn含量的预测问题,提出基于深度卷积神经网络回归预测模型。对原始土壤进行相关预处理,用粉末压片法制作土壤压片,采用X射线荧光光谱法(X-Ray-fluorescence,XRF)获取土壤光谱,相比于传统检测方式,XRF法具有检测速度快、精度高、操作简单、不破坏样品属性并且可实现多种重金属元素同时检测等优点,故将XRF与深度卷积神经网络相结合,实现对土壤中重金属Zn元素含量的精确预测。采用箱型图来剔除X射线荧光光谱中的异常数据,采用熵权法结合多元散射校正来对样品盒数据进行校正,采用Savitzky-Golay平滑去噪法以及线性本底法对光谱数据进行预处理,可以有效地解决由外界环境和人为因素产生的噪声及基线漂移等问题。针对卷积神经网络结构的特殊性,将获取的一维光谱数据向量,采用构建光谱数据矩阵的方式来进行处理,将同一浓度、同一含水率下5组平行光谱数据向量转化为二维光谱信息矩阵,以该矩阵作为深度卷积神经网络预测模型的输入,以适应卷积层的操作要求,利用深度卷积神经网络特殊的结构模式,能有效提取土壤光谱数据特征,提高了深度卷积神经网络预测模型的学习能力,降低模型的训练难度。深度卷积神经网络预测模型采用3层卷积层搭建,使用ReLU激活函数激活,采用最大池化方式,减少数据的维度,增加Dropout层,防止过拟合,使用ADAM优化器对预测模型进行优化。实验以平均相对误差(mean relative error, MRE)、损失函数(LOSS)、平均绝对误差(mean absolute error, MAE)确定了模型的最优学习率为10-3以及最优迭代次数为3000,并将深度卷积神经网络预测模型与BP预测模型、ELM预测模型、PLS预测模型进行对比,以均方误差(mean square error, MSE)、均方根误差(root mean square error, RMSE)、以及拟合系数R2来分析比较预测模型的好坏,结果表明,基于深度卷积神经网络预测模型在对土壤中重金属Zn元素含量预测方面优于BP,ELM,PLS三种预测模型,提高了预测精度。     

Matching pursuit of an adaptive impulse dictionary for bearing fault diagnosis

The sparse decomposition based on matching pursuit is an adaptive sparse expression of the signals. An adaptive matching pursuit algorithm that uses an impulse dictionary is introduced in this article for rolling bearing vibration signal processing and fault diagnosis. First, a new dictionary model is established according to the characteristics and mechanism of rolling bearing faults. The new model incorporates the rotational speed of the bearing, the dimensions of the bearing and the bearing fault status, among other parameters. The model can simulate the impulse experienced by the bearing at different bearing fault levels. A simulation experiment suggests that a new impulse dictionary used in a matching pursuit algorithm combined with a genetic algorithm has a more accurate effect on bearing fault diagnosis than using a traditional impulse dictionary. However, those two methods have some weak points, namely, poor stability, rapidity and controllability. Each key parameter in the dictionary model and its influence on the analysis results are systematically studied, and the impulse location is determined as the primary model parameter. The adaptive impulse dictionary is established by changing characteristic parameters progressively. The dictionary built by this method has a lower redundancy and a higher relevance between each dictionary atom and the analyzed vibration signal. The matching pursuit algorithm of an adaptive impulse dictionary is adopted to analyze the simulated signals. The results indicate that the characteristic fault components could be accurately extracted from the noisy simulation fault signals by this algorithm, and the result exhibited a higher efficiency in addition to an improved stability, rapidity and controllability when compared with a matching pursuit approach that was based on a genetic algorithm. We experimentally analyze the early-stage fault signals and composite fault signals of the bearing. The results further demonstrate the effectiveness and superiority of the matching pursuit algorithm that uses the adaptive impulse dictionary. Finally, this algorithm is applied to the analysis of engineering data, and good results are achieved.     

Intelligent Online Monitoring of Rolling Bearing: Diagnosis and Prognosis

This paper suggests a new method to predict the Remaining Useful Life (RUL) of rolling bearings based on Long Short Term Memory (LSTM), in order to obtain the degradation condition of the rolling bearings and realize the predictive maintenance. The approach is divided into three parts: the first part is the clustering to detect the damage state by the density-based spatial clustering of applications with noise. The second one is the health indicator construction which could give a better reflection of the bearing degradation tendency and is selected as the input for the prediction model. In the third part of the RUL prediction, the LSTM approach is employed to improve the accuracy of the prediction. The rationale of this work is to combine the two methods—the density-based spatial clustering of applications with noise and LSTM—to identify the abnormal state in rolling bearings, then estimate the RUL. The suggested method is confirmed by experimental data of bearing life cycle, and the RUL prediction results of the model LSTM are compared with the nonlinear au-regressive model with exogenous input model. In addition, the constructed health indicator is compared with the spectral kurtosis feature. The results demonstrated that the suggested method is more appropriate than the nonlinear au-regressive model with exogenous input model for the prediction of bearing RUL.     

采用优化极限学习机的多变量混沌时间序列预测

基于优化极限学习机理论, 提出一种多变量混沌时间序列预测方法. 该方法利用复合混沌和混沌变尺度算法对极限学习机的模型参数进行搜索和优化, 以提高极限学习机的泛化性能; 然后利用优化后的极限学习机对Rossler耦合系统的多变量混沌时序进行一步和多步预测, 并且与同类算法进行了比较, 结果表明了该方法的有效性, 且算法具有较强的抗噪能力; 最后讨论了预测结果和隐层神经元数目的关系.