Rolling Bearing Fault Diagnosis Based on VMD-MPE and PSO-SVM

The goal of the paper is to present a solution to improve the fault detection accuracy of rolling bearings. The method is based on variational mode decomposition (VMD), multiscale permutation entropy (MPE) and the particle swarm optimization-based support vector machine (PSO-SVM). Firstly, the original bearing vibration signal is decomposed into several intrinsic mode functions (IMF) by using the VMD method, and the feature energy ratio (FER) criterion is introduced to reconstruct the bearing vibration signal. Secondly, the multiscale permutation entropy of the reconstructed signal is calculated to construct multidimensional feature vectors. Finally, the constructed multidimensional feature vector is fed into the PSO-SVM classification model for automatic identification of different fault patterns of the rolling bearing. Two experimental cases are adopted to validate the effectiveness of the proposed method. Experimental results show that the proposed method can achieve a higher identification accuracy compared with some similar available methods (e.g., variational mode decomposition-based multiscale sample entropy (VMD-MSE), variational mode decomposition-based multiscale fuzzy entropy (VMD-MFE), empirical mode decomposition-based multiscale permutation entropy (EMD-MPE) and wavelet transform-based multiscale permutation entropy (WT-MPE)).     

A Rolling Bearing Fault Classification Scheme Based on k-Optimized Adaptive Local Iterative Filtering and Improved Multiscale Permutation Entropy

The health condition of the rolling bearing seriously affects the operation of the whole mechanical system. When the rolling bearing parts fail, the time series collected in the field generally shows strong nonlinearity and non-stationarity. To obtain the faulty characteristics of mechanical equipment accurately, a rolling bearing fault detection technique based on k-optimized adaptive local iterative filtering (ALIF), improved multiscale permutation entropy (improved MPE), and BP neural network was proposed. In the ALIF algorithm, a k-optimized ALIF method based on permutation entropy (PE) is presented to select the number of ALIF decomposition layers adaptively. The completely average coarse-graining method was proposed to excavate more hidden information. The performance analysis of the simulation signal shows that the improved MPE can more accurately dig out the depth information of the time series, and the entropy value obtained is more consistent and stable. In the research application, rolling bearing time series are decomposed by k-optimized ALIF to obtain a certain number of intrinsic mode functions (IMFs). Then the improved MPE value of effective IMF is calculated and input into backpropagation (BP) neural network as the feature vector for automatic fault identification. The comparative analysis of simulation signals shows that this method can extract fault information effectively. At the same time, the experimental part shows that this scheme not only effectively extracts the fault features, but also realizes the classification and identification of different fault modes and faults of different degrees, which has a certain application prospect in the research and application direction of rolling bearing fault identification.     

A Bearing Fault Diagnosis Method Based on PAVME and MEDE

When rolling bearings have a local fault, the real bearing vibration signal related to the local fault is characterized by the properties of nonlinear and nonstationary. To extract the useful fault features from the collected nonlinear and nonstationary bearing vibration signals and improve diagnostic accuracy, this paper proposes a new bearing fault diagnosis method based on parameter adaptive variational mode extraction (PAVME) and multiscale envelope dispersion entropy (MEDE). Firstly, a new method hailed as parameter adaptive variational mode extraction (PAVME) is presented to process the collected original bearing vibration signal and obtain the frequency components related to bearing faults, where its two important parameters (i.e., the penalty factor and mode center-frequency) are automatically determined by whale optimization algorithm. Subsequently, based on the processed bearing vibration signal, an effective complexity evaluation approach named multiscale envelope dispersion entropy (MEDE) is calculated for conducting bearing fault feature extraction. Finally, the extracted fault features are fed into the k-nearest neighbor (KNN) to automatically identify different health conditions of rolling bearing. Case studies and contrastive analysis are performed to validate the effectiveness and superiority of the proposed method. Experimental results show that the proposed method can not only effectively extract bearing fault features, but also obtain a high identification accuracy for bearing fault patterns under single or variable speed.     

Composite Multivariate Multi-Scale Permutation Entropy and Laplacian Score Based Fault Diagnosis of Rolling Bearing

As a powerful tool for measuring complexity and randomness, multivariate multi-scale permutation entropy (MMPE) has been widely applied to the feature representation and extraction of multi-channel signals. However, MMPE still has some intrinsic shortcomings that exist in the coarse-grained procedure, and it lacks the precise estimation of entropy value. To address these issues, in this paper a novel non-linear dynamic method named composite multivariate multi-scale permutation entropy (CMMPE) is proposed, for optimizing insufficient coarse-grained process in MMPE, and thus to avoid the loss of information. The simulated signals are used to verify the validity of CMMPE by comparing it with the often-used MMPE method. An intelligent fault diagnosis method is then put forward on the basis of CMMPE, Laplacian score (LS), and bat optimization algorithm-based support vector machine (BA-SVM). Finally, the proposed fault diagnosis method is utilized to analyze the test data of rolling bearings and is then compared with the MMPE, multivariate multi-scale multiscale entropy (MMFE), and multi-scale permutation entropy (MPE) based fault diagnosis methods. The results indicate that the proposed fault diagnosis method of rolling bearing can achieve effective identification of fault categories and is superior to comparative methods.     

Intelligent Fault Diagnosis of Rolling-Element Bearings Using a Self-Adaptive Hierarchical Multiscale Fuzzy Entropy

The fuzzy-entropy-based complexity metric approach has achieved fruitful results in bearing fault diagnosis. However, traditional hierarchical fuzzy entropy (HFE) and multiscale fuzzy entropy (MFE) only excavate bearing fault information on different levels or scales, but do not consider bearing fault information on both multiple layers and multiple scales at the same time, thus easily resulting in incomplete fault information extraction and low-rise identification accuracy. Besides, the key parameters of most existing entropy-based complexity metric methods are selected based on specialist experience, which indicates that they lack self-adaptation. To address these problems, this paper proposes a new intelligent bearing fault diagnosis method based on self-adaptive hierarchical multiscale fuzzy entropy. On the one hand, by integrating the merits of HFE and MFE, a novel complexity metric method, named hierarchical multiscale fuzzy entropy (HMFE), is presented to extract a multidimensional feature matrix of the original bearing vibration signal, where the important parameters of HMFE are automatically determined by using the bird swarm algorithm (BSA). On the other hand, a nonlinear feature matrix classifier with strong robustness, known as support matrix machine (SMM), is introduced for learning the discriminant fault information directly from the extracted multidimensional feature matrix and automatically identifying different bearing health conditions. Two experimental results on bearing fault diagnosis show that the proposed method can obtain average identification accuracies of 99.92% and 99.83%, respectively, which are higher those of several representative entropies reported by this paper. Moreover, in the two experiments, the standard deviations of identification accuracy of the proposed method were, respectively, 0.1687 and 0.2705, which are also greater than those of the comparison methods mentioned in this paper. The effectiveness and superiority of the proposed method are verified by the experimental results.     

Forest Fire Detection via Feature Entropy Guided Neural Network

Forest fire detection from videos or images is vital to forest firefighting. Most deep learning based approaches rely on converging image loss, which ignores the content from different fire scenes. In fact, complex content of images always has higher entropy. From this perspective, we propose a novel feature entropy guided neural network for forest fire detection, which is used to balance the content complexity of different training samples. Specifically, a larger weight is given to the feature of the sample with a high entropy source when calculating the classification loss. In addition, we also propose a color attention neural network, which mainly consists of several repeated multiple-blocks of color-attention modules (MCM). Each MCM module can extract the color feature information of fire adequately. The experimental results show that the performance of our proposed method outperforms the state-of-the-art methods.     

Evaluating the Learning Procedure of CNNs through a Sequence of Prognostic Tests Utilising Information Theoretical Measures

Deep learning has proven to be an important element of modern data processing technology, which has found its application in many areas such as multimodal sensor data processing and understanding, data generation and anomaly detection. While the use of deep learning is booming in many real-world tasks, the internal processes of how it draws results is still uncertain. Understanding the data processing pathways within a deep neural network is important for transparency and better resource utilisation. In this paper, a method utilising information theoretic measures is used to reveal the typical learning patterns of convolutional neural networks, which are commonly used for image processing tasks. For this purpose, training samples, true labels and estimated labels are considered to be random variables. The mutual information and conditional entropy between these variables are then studied using information theoretical measures. This paper shows that more convolutional layers in the network improve its learning and unnecessarily higher numbers of convolutional layers do not improve the learning any further. The number of convolutional layers that need to be added to a neural network to gain the desired learning level can be determined with the help of theoretic information quantities including entropy, inequality and mutual information among the inputs to the network. The kernel size of convolutional layers only affects the learning speed of the network. This study also shows that where the dropout layer is applied to has no significant effects on the learning of networks with a lower dropout rate, and it is better placed immediately after the last convolutional layer with higher dropout rates.     

灰色极限学习机在滚动轴承故障预测中的应用

针对较强噪声环境下的滚动轴承故障预测问题,为提高轴承故障预测的精度,提出并研究了一种新的滚动轴承预测技术。采用将灰色模型和极限学习机（ELM）相结合的方法,针对轴承运行状态值的非线性特点,先将样本数据进行灰色处理,解决数据的随机性和波动性问题,然后代入学习速度快,泛化精度高的ELM神经网络进行训练。在训练完毕后,对未来的轴承运行状态数据进行分析,将其与轴承设备的理论诊断标准相比较以达到故障预测的目的。     

Hierarchical Amplitude-Aware Permutation Entropy-Based Fault Feature Extraction Method for Rolling Bearings

In order to detect the incipient fault of rolling bearings and to effectively identify fault characteristics, based on amplitude-aware permutation entropy (AAPE), an enhanced method named hierarchical amplitude-aware permutation entropy (HAAPE) is proposed in this paper to solve complex time series in a new dynamic change analysis. Firstly, hierarchical analysis and AAPE are combined to excavate multilevel fault information, both low-frequency and high-frequency components of the abnormal bearing vibration signal. Secondly, from the experimental analysis, it is found that HAAPE is sensitive to the early failure of rolling bearings, which makes it suitable to evaluate the performance degradation of a bearing in its run-to-failure life cycle. Finally, a fault feature selection strategy based on HAAPE is put forward to select the bearing fault characteristics after the application of the least common multiple in singular value decomposition (LCM-SVD) method to the fault vibration signal. Moreover, several other entropy-based methods are also introduced for a comparative analysis of the experimental data, and the results demonstrate that HAAPE can extract fault features more effectively and with a higher accuracy.     

基于轻量化卷积神经网络的光伏电池片缺陷检测方法研究

光伏电池片中的缺陷会影响整个光伏系统使用寿命及发电效率。针对现有电池片自动检测中尺寸弱小缺陷漏检率高的问题,建立了一种特征增强型轻量化卷积神经网络模型。针对性地设计了特征增强提取模块,提高了弱边界的提取能力,同时根据多尺度识别原理,增加了小目标预测层,实现了多尺度特征预测。在实验测试中,该模型平均精度均值（mAP）达到87.55%,比传统模型提高了6.78个百分点,同时检测速度达到40帧/s,满足精准性与实时性的检测要求。     

融合多尺度局部特征与深度特征的双目立体匹配

针对立体匹配中不适定区域难以找到精确匹配点的问题,提出一种融合多尺度局部特征与深度特征的立体匹配方法。特征融合阶段包括两部分,其一是融合不同尺度下Log-Gabor特征和局部二值模式特征组合的浅层次特征,其二是将多尺度浅层融合特征和卷积神经网络提取的深度特征进行级联,形成既包含语义信息又包含结构化信息的特征图像。通过在极线垂直方向添加不同强度的噪声来构造正负样本,减小图像中极线对齐欠准带来的误差。将该方法与两种变体方法(改变或舍弃部分模块)在KITTI数据集进行对比实验,结果表明各模块设置具有合理性;与一些经典方法相比,所提方法取得了有竞争力的匹配性能。     

Application of Generalized Composite Multiscale Lempel–Ziv Complexity in Identifying Wind Turbine Gearbox Faults

Wind turbine gearboxes operate in harsh environments; therefore, the resulting gear vibration signal has characteristics of strong nonlinearity, is non-stationary, and has a low signal-to-noise ratio, which indicates that it is difficult to identify wind turbine gearbox faults effectively by the traditional methods. To solve this problem, this paper proposes a new fault diagnosis method for wind turbine gearboxes based on generalized composite multiscale Lempel–Ziv complexity (GCMLZC). Within the proposed method, an effective technique named multiscale morphological-hat convolution operator (MHCO) is firstly presented to remove the noise interference information of the original gear vibration signal. Then, the GCMLZC of the filtered signal was calculated to extract gear fault features. Finally, the extracted fault features were input into softmax classifier for automatically identifying different health conditions of wind turbine gearboxes. The effectiveness of the proposed method was validated by the experimental and engineering data analysis. The results of the analysis indicate that the proposed method can identify accurately different gear health conditions. Moreover, the identification accuracy of the proposed method is higher than that of traditional multiscale Lempel–Ziv complexity (MLZC) and several representative multiscale entropies (e.g., multiscale dispersion entropy (MDE), multiscale permutation entropy (MPE) and multiscale sample entropy (MSE)).     

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...     

多局部残差连接注意网络的图像去模糊

针对现有的基于卷积神经网络的图像去模糊算法存在图像纹理细节恢复不清晰的问题,提出了一种基于多局部残差连接注意网络的图像去模糊算法。首先,采用一个卷积层进行浅层特征提取;其次,设计了一种新的基于残差连接和并行注意机制的多局部残差连接注意模块,用于消除图像模糊并提取上下文信息;再次,采用一个基于扩张卷积的成对连接模块进行细节恢复;最后,利用一个卷积层重建清晰图像。实验结果表明：在GoPro数据集上的PSNR (peak signal to noise ratio)和SSIM (structure similarity)分别为31.83 dB、0.927 5,在定性和定量两方面都表明所提方法能够有效地恢复模糊图像的纹理细节,网络性能优于对比方法。     

基于波段注意力卷积网络的近红外奶粉皮革水解蛋白掺假检测

近年来,深度学习技术在近红外光谱、拉曼光谱、荧光光谱等的光谱学数据建模上取得一系列突破。由于深度学习方法对于样本数量的需求高,而在分析化学领域获得大量有标签样本较为困难,因此过拟合问题一直是深度神经网络在化学计量学中应用时研究者高度关注的问题。该工作提出基于波段注意力卷积网络(WA-CNN)的近红外数据建模方法,并应用于婴儿配方奶粉皮革水解蛋白(HLP)掺假定量分析。WA-CNN在传统卷积网络的基础上加入波段注意力模块,该模块采用卷积操作自训练波段注意力权值,并以乘法加权形式对有效波段进行激活,从而有效缓解深度神经网络在近红外数据建模中的波段信息冗余问题,达到抑制过拟合,提升预测精度的目的。研究中共测试100个皮革水解蛋白掺假婴儿配方奶粉样本的近红外光谱数据,其中皮革水解蛋白的掺假比例范围是0%～20%。采用60%的样本训练,剩余40%样本测试,随机采样10次,通过测试集均方根误差(RMSEP)、决定系数(R²)以及相对分析误差(RPD)的均值来进行模型评价。并建立偏最小二乘回归(PLS)、支持向量机回归(SVR)和常规的一维卷积神经网络(CNN)三种传统模型用于...     

基于改进ViT的红外人体图像步态识别方法研究

针对卷积神经网络在步态识别时准确率易饱和现象,以及Vision Transformer(ViT)对步态数据集拟合效率较低的问题,提出构建一个对称双重注意力机制模型,保留行走姿态的时间顺序,用若干独立特征子空间有针对性地拟合步态图像块;同时,采用对称架构的方式,增强注意力模块在拟合步态特征时的作用,并利用异类迁移学习进一步提升特征拟合效率。将该模型运用在中科院CASIA C红外人体步态库中进行多次仿真实验,平均识别准确率达到96.8%。结果表明,本文模型在稳定性、数据拟合速度以及识别准确率3方面皆优于传统ViT模型和CNN对比模型。     

Estimating Phase Amplitude Coupling between Neural Oscillations Based on Permutation and Entropy

Cross-frequency phase–amplitude coupling (PAC) plays an important role in neuronal oscillations network, reflecting the interaction between the phase of low-frequency oscillation (LFO) and amplitude of the high-frequency oscillations (HFO). Thus, we applied four methods based on permutation analysis to measure PAC, including multiscale permutation mutual information (MPMI), permutation conditional mutual information (PCMI), symbolic joint entropy (SJE), and weighted-permutation mutual information (WPMI). To verify the ability of these four algorithms, a performance test including the effects of coupling strength, signal-to-noise ratios (SNRs), and data length was evaluated by using simulation data. It was shown that the performance of SJE was similar to that of other approaches when measuring PAC strength, but the computational efficiency of SJE was the highest among all these four methods. Moreover, SJE can also accurately identify the PAC frequency range under the interference of spike noise. All in all, the results demonstrate that SJE is better for evaluating PAC between neural oscillations.     

基于卷积神经网络的焊缝缺陷图像分类研究

为有效地对焊缝缺陷进行分类,从而判断焊接质量的等级,对传统卷积神经网络进行改进,提出一种多尺度压缩激励网络模型（SINet）。将4组两两串联的3×3卷积模块与Inception模块、压缩激励模块（SE block）相结合。通过多尺度压缩激励模块（SI module）将卷积层中的特征进行多尺度融合和特征重标定以提高分类准确率,并用全局平均池化层代替全连接层减少模型参数。此外考虑到焊接缺陷数量不平衡对准确率的影响,采用深度卷积对抗生成网络（DCGAN）进行数据集的平衡处理,并在该数据集上验证模型的有效性。与传统卷积神经网络相比,该模型具有良好的性能,在测试集上准确率达到96.77%,同时模型的参数个数也明显减少。结果表明该方法对焊缝缺陷图像能进行有效地分类。     

卷积神经网络用于近红外光谱预测土壤含水率

近红外光谱分析技术在土壤含水率预测方面具有独特的优势,是一种便捷且有效的方法。卷积神经网络作为高性能的深度学习模型,能够从复杂光谱数据中自主提取有效特征结构进行学习,与传统的浅层学习模型相比具有更强的模型表达能力。将卷积神经网络用于近红外光谱预测土壤含水率,并提出了有效的卷积神经网络光谱回归建模方法,简化了光谱数据的预处理要求,且具有更高的光谱预测精度。首先对不同含水率下土壤样品的光谱反射率数据进行简单的预处理,通过主成分分析减少光谱数据量,并将处理后的光谱数据变换为二维光谱信息矩阵,以适应卷积神经网络特殊的学习结构。然后基于卷积神经网络算法,设置双层卷积和池化结构逐层提取光谱数据的内部特征信息,并采用局部连接和权值共享减少网络参数、提高泛化性能。通过试验优化网络结构和各项参数,最终获得针对土壤光谱数据的卷积神经网络土壤含水率预测模型,并与传统的BP,PLSR和LSSVM模型进行对比实验。结果表明在训练样本达到一定数量时,卷积神经网络的预测精度和回归拟合度均高于三种传统模型。在少量训练样本参与建模的情况下,模型预测表现高于BP神经网络,但略低于PLSR和LSSVM模型。随着参与训练样本量的增加,卷积神经网络的预测精度和回归拟合度也随之稳定提升,达到并显著优于传统模型水平。因此,卷积神经网络能够利用近红外光谱数据对土壤含水率做出有效预测,且在较多样本参与建模时取得更好效果。     

Permutation Entropy-Based Interpretability of Convolutional Neural Network Models for Interictal EEG Discrimination of Subjects with Epileptic Seizures vs. Psychogenic Non-Epileptic Seizures

The differential diagnosis of epileptic seizures (ES) and psychogenic non-epileptic seizures (PNES) may be difficult, due to the lack of distinctive clinical features. The interictal electroencephalographic (EEG) signal may also be normal in patients with ES. Innovative diagnostic tools that exploit non-linear EEG analysis and deep learning (DL) could provide important support to physicians for clinical diagnosis. In this work, 18 patients with new-onset ES (12 males, 6 females) and 18 patients with video-recorded PNES (2 males, 16 females) with normal interictal EEG at visual inspection were enrolled. None of them was taking psychotropic drugs. A convolutional neural network (CNN) scheme using DL classification was designed to classify the two categories of subjects (ES vs. PNES). The proposed architecture performs an EEG time-frequency transformation and a classification step with a CNN. The CNN was able to classify the EEG recordings of subjects with ES vs. subjects with PNES with 94.4% accuracy. CNN provided high performance in the assigned binary classification when compared to standard learning algorithms (multi-layer perceptron, support vector machine, linear discriminant analysis and quadratic discriminant analysis). In order to interpret how the CNN achieved this performance, information theoretical analysis was carried out. Specifically, the permutation entropy (PE) of the feature maps was evaluated and compared in the two classes. The achieved results, although preliminary, encourage the use of these innovative techniques to support neurologists in early diagnoses.