scispace - formally typeset
Search or ask a question

Showing papers in "IEEE Transactions on Instrumentation and Measurement in 2021"


Journal ArticleDOI
TL;DR: In this article, a model-driven scheme with dual stages is proposed to compensate the dynamic pressure measurement, which is applied to an industrial hydraulic pipe system, and the experimental results show that the relative error is reduced greatly after the compensation is implemented, demonstrating the validity of the proposed method.
Abstract: Strain-based non-intrusive approaches for measuring the pressure of pipes have attracted widespread attention due to their great convenience and ability to avoid destroying the integrity of structures. However, the mentioned method usually measures the dynamic pressure based only on the static strain–pressure sensitivity coefficients (SSSCs) instead of the dynamic strain–pressure sensitivity coefficients (DSSCs) due to its complicated calibration, which will inevitably affect the accuracy significantly. To address this issue, a model-driven scheme with dual stages is proposed in the present study to compensate the dynamic pressure measurement. The DSSCs are analytically derived for the first time from the axial governing equations of the pipe, considering the general boundary conditions for the thin-walled and thick-walled pipes simultaneously. In the first stage, the physical parameters involved in the DSSCs are calibrated by minimizing the residual of the experimental results and the theoretical counterparts. In the second stage, the DSSCs calculated from the calibrated analytical model are utilized to compensate the dynamic pressure based on the relationship between the DSSCs and the SSSCs. The proposed method is applied to an industrial hydraulic pipe system, and the experimental results show that the relative error is reduced greatly after the compensation is implemented, demonstrating the validity of the proposed compensation method.

191 citations


Journal ArticleDOI
TL;DR: In this article, the authors proposed a one-stage object detection framework for improving the detection accuracy while supporting a true real-time operation based on the YOLOv4.
Abstract: The use of object detection algorithms has become extremely important in autonomous vehicles. Object detection at high accuracy and a fast inference speed is essential for safe autonomous driving. Therefore, the balance between the effectiveness and efficiency of the object detector must be considered. This article proposes a one-stage object detection framework for improving the detection accuracy while supporting a true real-time operation based on the YOLOv4. The backbone network in the proposed framework is the CSPDarknet53_dcn(P). The last output layer in the CSPDarknet53 is replaced with deformable convolution to improve the detection accuracy. In order to perform feature fusion, a new feature fusion module PAN++ is designed and five scales detection layers are used to improve the detection accuracy of small objects. In addition, this article proposes an optimized network pruning algorithm to solve the problem that the real-time performance of the algorithm cannot be satisfied due to the limited computing resources of the vehicle-mounted computing platform. The method of sparse scaling factor is used to improve the existing channel pruning algorithm. Compared to the YOLOv4, the YOLOV4-5D improves the mean average precision by 4.23% on the BDD data sets and 1.68% on the KITTI data sets. Finally, by pruning the model, the inference speed of YOLOV4-5D is increased 31.3% and the memory is only 98.1 MB when the detection accuracy is almost unchanged. Nevertheless, the proposed algorithm is capable of real-time detection at faster than 66 frames/s (fps) and shows higher accuracy than the previous approaches with a similar fps.

173 citations


Journal ArticleDOI
TL;DR: Zhao et al. as mentioned in this paper constructed a taxonomy and performed a comprehensive review of unsupervised deep transfer learning (UDTL)-based intelligent fault diagnosis (IFD) according to different tasks.
Abstract: Recent progress on intelligent fault diagnosis (IFD) has greatly depended on deep representation learning and plenty of labeled data. However, machines often operate with various working conditions, or the target task has different distributions with the collected data used for training (the domain shift problem). Besides, the newly collected test data in the target domain are usually unlabeled, leading to unsupervised deep transfer learning (UDTL)-based IFD problem. Although it has achieved huge development, a standard and open source code framework and a comparative study for UDTL-based IFD are not yet established. In this article, we construct a new taxonomy and perform a comprehensive review of UDTL-based IFD according to different tasks. Comparative analysis of some typical methods and datasets reveals some open and essential issues in UDTL-based IFD, which is rarely studied, including transferability of features, the influence of backbones, negative transfer, physical priors, and so on. To emphasize the importance and reproducibility of UDTL-based IFD, the whole test framework will be released to the research community to facilitate future research. In summary, the released framework and comparative study can serve as an extended interface and basic results to carry out new studies on UDTL-based IFD. The code framework is available at https://github.com/ZhaoZhibin/UDTL .

161 citations


Journal ArticleDOI
TL;DR: An image fusion-based algorithm to enhance the performance and robustness of image dehazing is proposed, based on a set of gamma-corrected underexposed images, and pixelwise weight maps are constructed by analyzing both global and local exposedness to guide the fusion process.
Abstract: Poor weather conditions, such as fog, haze, and mist, cause visibility degradation in captured images. Existing imaging devices lack the ability to effectively and efficiently mitigate the visibility degradation caused by poor weather conditions in real time. Image depth information is used to eliminate hazy effects by using existing physical model-based approaches. However, the imprecise depth information always affects dehazing performance. This article proposes an image fusion-based algorithm to enhance the performance and robustness of image dehazing. Based on a set of gamma-corrected underexposed images, pixelwise weight maps are constructed by analyzing both global and local exposedness to guide the fusion process. The spatial-dependence of luminance of the fused image is reduced, and its color saturation is balanced in the dehazing process. The performance of the proposed solution is confirmed in both theoretical analysis and comparative experiments.

150 citations


Journal ArticleDOI
Jiayi Ma1, Hao Zhang1, Zhenfeng Shao1, Pengwei Liang1, Han Xu1 
TL;DR: A new fusion framework called generative adversarial network with multiclassification constraints (GANMcC) is proposed, which transforms image fusion into a multidistribution simultaneous estimation problem to fuse infrared and visible images in a more reasonable way.
Abstract: Visible images contain rich texture information, whereas infrared images have significant contrast. It is advantageous to combine these two kinds of information into a single image so that it not only has good contrast but also contains rich texture details. In general, previous fusion methods cannot achieve this goal well, where the fused results are inclined to either a visible or an infrared image. To address this challenge, a new fusion framework called generative adversarial network with multiclassification constraints (GANMcC) is proposed, which transforms image fusion into a multidistribution simultaneous estimation problem to fuse infrared and visible images in a more reasonable way. We adopt a generative adversarial network with multiclassification to estimate the distributions of visible light and infrared domains at the same time, in which the game of multiclassification discrimination will make the fused result to have these two distributions in a more balanced manner, so as to have significant contrast and rich texture details. In addition, we design a specific content loss to constrain the generator, which introduces the idea of main and auxiliary into the extraction of gradient and intensity information, which will enable the generator to extract more sufficient information from source images in a complementary manner. Extensive experiments demonstrate the advantages of our GANMcC over the state-of-the-art methods in terms of both qualitative effect and quantitative metric. Moreover, our method can achieve good fused results even the visible image is overexposed. Our code is publicly available at https://github.com/jiayi-ma/GANMcC .

144 citations



Journal ArticleDOI
TL;DR: A deep neural network strategy is presented to ameliorate the difficulties faced in ECG-based CVD diagnosis and treatment and suggests that the proposed model could serve as an analytic module to alert users and/or medical experts when anomalies are detected.
Abstract: An electrocardiograph (ECG) is employed as a primary tool for diagnosing cardiovascular diseases (CVDs). ECG signals provide a framework to probe the underlying properties and enhance the initial diagnosis obtained via traditional tools and patient–doctor dialogs. Notwithstanding its proven utility, deciphering large data sets to determine appropriate information remains a challenge in ECG-based CVD diagnosis and treatment. Our study presents a deep neural network (DNN) strategy to ameliorate the aforementioned difficulties. Our strategy consists of a learning stage where classification accuracy is improved via a robust feature extraction protocol. This is followed by using a genetic algorithm (GA) process to aggregate the best combination of feature extraction and classification. Comparison of the performance recorded for the proposed technique alongside state-of-the-art methods reported the area shows an increase of 0.94 and 0.953 in terms of average accuracy and F1 score, respectively. The outcomes suggest that the proposed model could serve as an analytic module to alert users and/or medical experts when anomalies are detected.

130 citations


Journal ArticleDOI
Jiayi Ma1, Linfeng Tang1, Meilong Xu1, Hao Zhang1, Guobao Xiao2 
TL;DR: Li et al. as mentioned in this paper proposed an infrared and visible image fusion network based on the salient target detection, which can preserve the thermal targets in infrared images and the texture structures in visible images.
Abstract: In this article, we propose an infrared and visible image fusion network based on the salient target detection, termed STDFusionNet, which can preserve the thermal targets in infrared images and the texture structures in visible images. First, a salient target mask is dedicated to annotating regions of the infrared image that humans or machines pay more attention to, so as to provide spatial guidance for the integration of different information. Second, we combine this salient target mask to design a specific loss function to guide the extraction and reconstruction of features. Specifically, the feature extraction network can selectively extract salient target features from infrared images and background texture features from visible images, while the feature reconstruction network can effectively fuse these features and reconstruct the desired results. It is worth noting that the salient target mask is only required in the training phase, which enables the proposed STDFusionNet to be an end-to-end model. In other words, our STDFusionNet can fulfill salient target detection and key information fusion in an implicit manner. Extensive qualitative and quantitative experiments demonstrate the superiority of our fusion algorithm over the current state of the arts, where our algorithm is much faster and the fusion results look like high-quality visible images with clear highlighted infrared targets. Moreover, the experimental results on the public datasets reveal that our algorithm can improve the entropy (EN), mutual information (MI), visual information fidelity (VIF), and spatial frequency (SF) metrics with about 1.25%, 22.65%, 4.3%, and 0.89% gains, respectively. Our code is publicly available at https://github.com/jiayi-ma/STDFusionNet .

113 citations


Journal ArticleDOI
TL;DR: In this article, a novel trigonometric cross-entropy function is developed to compute the sparsity cost, which introduces sparsity by avoiding unnecessary activation of neurons in the hidden layers of CNN.
Abstract: This work presents the development of novel convolutional neural network (NCNN) for effective identification of bearing defects from small samples. For effective feature learning from small training data, cost function of convolution neural network (CNN) is modified by adding additional sparsity cost in the existing cost function. A novel trigonometric cross-entropy function is developed to compute the sparsity cost. The proposed cost function introduces sparsity by avoiding unnecessary activation of neurons in the hidden layers of CNN. For identification of bearing defects from small training samples, NCNN-based transfer learning is applied in the following manner. First, the raw vibration signals as well as envelope signals from source domain machine are obtained. Thereafter, these envelope signals are applied to NCNN for the learning of features from the big training data acquitted from the source domain. After feature learning, knowledge gained from NCNN is transferred to do fine-tuning of NCNN from small training samples of target domain. Thereafter, defect identification is carried out by applying the test data of target domain to fine-tuned NCNN. The experimental result validates that the proposed cross-entropy function introduces sparsity in CNN and, hence, creates an effective deep learning which can even work under a situation when training data are not available in abundant.

110 citations


Journal ArticleDOI
TL;DR: In this paper, two sliding window techniques are proposed to enhance the binary classification of motor imagery (MI) brain-computer interface (BCI) signals, namely SW-LCR and SW-Mode.
Abstract: Accurate binary classification of electroencephalography (EEG) signals is a challenging task for the development of motor imagery (MI) brain–computer interface (BCI) systems. In this study, two sliding window techniques are proposed to enhance the binary classification of MI. The first one calculates the longest consecutive repetition (LCR) of the sequence of prediction of all the sliding windows and is named SW-LCR. The second calculates the mode of the sequence of prediction of all the sliding windows and is named SW-Mode. Common spatial pattern (CSP) is used for extracting features with linear discriminant analysis (LDA) used for classification of each time window. Both the SW-LCR and SW-Mode are applied on publicly available BCI Competition IV-2a data set of healthy individuals and on a stroke patients’ data set. Compared with the existing state of the art, the SW-LCR performed better in the case of healthy individuals and SW-Mode performed better on stroke patients’ data set for left- versus right-hand MI with lower standard deviation. For both the data sets, the classification accuracy (CA) was approximately 80% and kappa ( $\kappa $ ) was 0.6. The results show that the sliding window-based prediction of MI using SW-LCR and SW-Mode is robust against intertrial and intersession inconsistencies in the time of activation within a trial and thus can lead to a reliable performance in a neurorehabilitative BCI setting.

106 citations


Journal ArticleDOI
Xun Cheng1, Jianbo Yu1
TL;DR: A new deep neural network (DNN), RetinaNet with difference channel attention and adaptively spatial feature fusion (DEA_RetinaNet), is proposed for steel surface defect detection and has better recognition performance compared with other famous DNN-based detectors.
Abstract: Surface defect detection of products is an important process to guarantee the quality of industrial production. A defect detection task aims to identify the specific category and precise position of defect in an image. It is hard to take into account the accuracy of both, which makes it be challenging in practice. In this study, a new deep neural network (DNN), RetinaNet with difference channel attention and adaptively spatial feature fusion (DEA_RetinaNet), is proposed for steel surface defect detection. First, a differential evolution search-based anchor optimization is performed to improve the detection accuracy of DEA_RetinaNet. Second, a novel channel attention mechanism is embedded in DEA_RetinaNet to reduce information loss. Finally, the adaptive spatial feature fusion (ASFF) module is used for an effective fusion of shallow and deep features extracted by convolutional kernels. The experimental results on a steel surface defect data set (NEU-DET) show that DEA_RetinaNet achieved 78.25 mAP and improved by 2.92% over RetinaNet. It has better recognition performance compared with other famous DNN-based detectors.

Journal ArticleDOI
TL;DR: Wang et al. as mentioned in this paper proposed a domain generalization-based hybrid diagnosis network, which regularizes the discriminant structure of the deep network with both intrinsic and extrinsic generalization objectives such that the diagnostic model can learn robust features and generalize to unseen domains.
Abstract: The data-driven methods in machinery fault diagnosis have become increasingly popular in the past two decades. However, the wide applications of this scheme are generally compromised in real-world conditions because of the discrepancy between the training data and testing data. Although the recently emerging transfer fault diagnosis can learn transferable features from relevant source data and adapt the diagnostic model to the target data, these methods still only work on the target domain with a priori data distribution. The generalization capability of the transferred model cannot be guaranteed for unseen domains. Since the working conditions of machinery are varying during operation, the generalization capability of the diagnosis methods is crucial in this case. To tackle this challenge, this article proposes a domain generalization-based hybrid diagnosis network for deploying to unseen working conditions. The main idea is to regularize the discriminant structure of the deep network with both intrinsic and extrinsic generalization objectives such that the diagnostic model can learn robust features and generalize to unseen domains. The triplet loss minimization of intrinsic multisource data is implemented to facilitate the intraclass compactness and the interclass separability at the class level, leading to a more generalized decision boundary. The extrinsic domain-level regularization is achieved by using adversarial training to further reduce the risk of overfitting. Extensive cross-domain diagnostic experiments on planetary gearbox demonstrate the effectiveness of the proposed method.

Journal ArticleDOI
TL;DR: In this article, a fast wind turbine abnormal data cleaning algorithm via image processing for wind turbine power generation performance measurement and evaluation is proposed, which includes two stages, data cleaning and data classification.
Abstract: A fast wind turbine abnormal data cleaning algorithm via image processing for wind turbine power generation performance measurement and evaluation is proposed in this paper. The proposed method includes two stages, data cleaning and data classification. At the data cleaning stage, pixels of normal data are extracted via image processing based on pixel spatial distribution characteristics of abnormal and normal data in wind power curve (WPC) images. At the data classification stage, wind power data points are classified as normal and abnormal based on the existence of corresponding pixels in the processed WPC image. To accelerate the proposed method, the cleaning operation is executed parallelly using graphics processing units (GPUs) via compute unified device architecture (CUDA). The effectiveness of the proposed method is validated based on real data sets collected from 37 wind turbines of two commercial farms and three types of GPUs are employed to implement the proposed algorithm. The computational results prove the proposed approach has achieved better performance in cleaning abnormal wind power data while the execution time is tremendously reduced. Therefore, the proposed method is available and practical for real wind turbine power generation performance evaluation and monitoring tasks.


Journal ArticleDOI
Tianfu Li1, Zhibin Zhao1, Chuang Sun1, Ruqiang Yan1, Xuefeng Chen1 
TL;DR: In this article, a domain adversarial graph convolutional network (DAGCN) is proposed to model the three types of information in a unified deep network and achieving UDA.
Abstract: Unsupervised domain adaptation (UDA)-based methods have made great progress in mechanical fault diagnosis under variable working conditions. In UDA, three types of information, including class label, domain label, and data structure, are essential to bridging the labeled source domain and unlabeled target domain. However, most existing UDA-based methods use only the former two information and ignore the modeling of data structure, which make the information contained in the features extracted by the deep network incomplete. To tackle this issue, a domain adversarial graph convolutional network (DAGCN) is proposed to model the three types of information in a unified deep network and achieving UDA. The first two types of information are modeled by the classifier and the domain discriminator, respectively. In data structure modeling, a convolutional neural network (CNN) is first employed to exact features from input signals. After that, the CNN features are input to the proposed graph generation layer to construct instance graphs by mining the relationship of structural characteristics of samples. Then, the instance graphs are modeled by a graph convolutional network, and the maximum mean discrepancy metric is leveraged to estimate the structure discrepancy of instance graphs from different domains. Experimental results conducted on two case studies demonstrate that the proposed DAGCN can not only obtain the best performance among the comparison methods, but also can extract transferable features for domain adaptation. The code library is available at: https://github.com/HazeDT/DAGCN .

Journal ArticleDOI
TL;DR: Wang et al. as discussed by the authors proposed a new feature extraction method based on the data-driven method, namely, Fitting Curve Derivative Method of Maximum Power Spectrum Density (FDMPD), to extract the performance degradation features of the same or similar rolling bearings from the historical state monitoring data.
Abstract: A variety of data-driven methods have been proposed to predict remaining useful life (RUL) of key component for rolling bearings. The accuracy of data-driven RUL prediction model largely depends on the extraction method of performance degradation features. The individual heterogeneity and working condition difference of rolling bearings lead to the different performance degradation curves of rolling bearings, which result in the mismatch between the established RUL prediction model by the training rolling bearings and the test rolling bearings. If a feature is found, which can reflect the consistency of the performance degradation curve of each rolling bearings, and give the indicator to determine the node and predictable interval of the declining period, the accuracy of the RUL prediction model will be improved. To solve this problem, a new feature extraction method based on the data-driven method, namely, Fitting Curve Derivative Method of Maximum Power Spectrum Density (FDMPD), is proposed to extract the performance degradation features of the same or similar rolling bearings from the historical state monitoring data in this article. The FDMPD can make the performance degradation feature curves of life cycle, which takes on consistency trend for different rolling bearings, and the starting point of the rolling bearings to enter the degenerating period is defined and the working stage of rolling bearings is divided. Based on this, the kernel extreme learning machine (KELM) and weight application to failure times (WAFT) are combined with FDMPD to establish a new RUL prediction model of rolling bearings, which can effectively realize the RUL prediction of rolling bearings. The whole life cycle data of rolling bearings are used to verify the validity of the RUL prediction model. The experimental results show that the established RUL prediction model can accurately predict the RUL of rolling bearings. It has the advantages of rapidity, stability, and applicability.

Journal ArticleDOI
TL;DR: Wang et al. as discussed by the authors proposed a triplet-graph reasoning network (TGRNet) for metal surface defect segmentation, which includes triplet encoder and trip loss to segment background and defect area, respectively.
Abstract: Metal surface defect segmentation can play an important role in dealing with the issue of quality control during the production and manufacturing stages. There are still two major challenges in industrial applications. One is the case that the number of metal surface defect samples is severely insufficient, and the other is that the most existing algorithms can only be used for specific surface defects and it is difficult to generalize to other metal surfaces. In this work, a theory of few-shot metal generic surface defect segmentation is introduced to solve these challenges. Simultaneously, the Triplet-Graph Reasoning Network (TGRNet) and a novel dataset Surface Defects- $4^{i}$ are proposed to achieve this theory. In our TGRNet, the surface defect triplet (including triplet encoder and trip loss) is proposed and is used to segment background and defect area, respectively. Through triplet, the few-shot metal surface defect segmentation problem is transformed into few-shot semantic segmentation problem of defect area and background area. For few-shot semantic segmentation, we propose a method of multi-graph reasoning to explore the similarity relationship between different images. And to improve segmentation performance in the industrial scene, an adaptive auxiliary prediction module is proposed. For Surface Defects- $4^{i}$ , it includes multiple categories of metal surface defect images to verify the generalization performance of our TGRNet and adds the nonmetal categories (leather and tile) as extensions. Through extensive comparative experiments and ablation experiments, it is proved that our architecture can achieve state-of-the-art results.

Journal ArticleDOI
TL;DR: In this paper, a DL-based convolutional neural network (CNN) called DeprNet was proposed for classifying the EEG data of depressed and normal subjects, where the Patient Health Questionnaire 9 score was used for quantifying the level of depression.
Abstract: Depression is a common reason for an increase in suicide cases worldwide Thus, to mitigate the effects of depression, accurate diagnosis and treatment are needed An electroencephalogram (EEG) is an instrument used to measure and record the brain’s electrical activities It can be utilized to produce the exact report on the level of depression Previous studies proved the feasibility of the usage of EEG data and deep learning (DL) models for diagnosing mental illness Therefore, this study proposes a DL-based convolutional neural network (CNN) called DeprNet for classifying the EEG data of depressed and normal subjects Here, the Patient Health Questionnaire 9 score is used for quantifying the level of depression The performance of DeprNet in two experiments, namely, the recordwise split and the subjectwise split, is presented in this study The results attained by DeprNet have an accuracy of 09937, and the area under the receiver operating characteristic curve (AUC) of 0999 is achieved when recordwise split data are considered On the other hand, an accuracy of 0914 and the AUC of 0956 are obtained, while subjectwise split data are employed These results suggest that CNN trained on recordwise split data gets overtrained on EEG data with a small number of subjects The performance of DeprNet is remarkable compared with the other eight baseline models Furthermore, on visualizing the last CNN layer, it is found that the values of right electrodes are prominent for depressed subjects, whereas, for normal subjects, the values of left electrodes are prominent

Journal ArticleDOI
TL;DR: In this article, a deep learning-based scheme is proposed for identifying the facial expression of a person, which consists of two parts: the former one finds out local features from face images using a local gravitational force descriptor, while, in the latter part, the descriptor is fed into a novel deep convolution neural network (DCNN) model.
Abstract: An image is worth a thousand words; hence, a face image illustrates extensive details about the specification, gender, age, and emotional states of mind. Facial expressions play an important role in community-based interactions and are often used in the behavioral analysis of emotions. Recognition of automatic facial expressions from a facial image is a challenging task in the computer vision community and admits a large set of applications, such as driver safety, human–computer interactions, health care, behavioral science, video conferencing, cognitive science, and others. In this work, a deep-learning-based scheme is proposed for identifying the facial expression of a person. The proposed method consists of two parts. The former one finds out local features from face images using a local gravitational force descriptor, while, in the latter part, the descriptor is fed into a novel deep convolution neural network (DCNN) model. The proposed DCNN has two branches. The first branch explores geometric features, such as edges, curves, and lines, whereas holistic features are extracted by the second branch. Finally, the score-level fusion technique is adopted to compute the final classification score. The proposed method along with 25 state-of-the-art methods is implemented on five benchmark available databases, namely, Facial Expression Recognition 2013, Japanese Female Facial Expressions, Extended CohnKanade, Karolinska Directed Emotional Faces, and Real-world Affective Faces. The databases consist of seven basic emotions: neutral, happiness, anger, sadness, fear, disgust, and surprise. The proposed method is compared with existing approaches using four evaluation metrics, namely, accuracy, precision, recall, and f1-score. The obtained results demonstrate that the proposed method outperforms all state-of-the-art methods on all the databases.

Journal ArticleDOI
Hai Yang1, Lizao Zhang1, Li Li1, Haibo Liang1, Jialing Zou1 
TL;DR: In this article, a method to improve the accuracy of U-tube mass flow meters based on variable step-size least-mean-square (LMS) filter and Hilbert transform with interval shifting under pulsating flow was presented.
Abstract: Coriolis mass flowmeter (CMF) is a kind of flow measurement instrument, which can directly measure the high-precision transient mass flow parameters. Also, the vibration characteristic of the U-shaped measuring tube inside is one of the important factors that determine the measuring accuracy. The pulsating flow through the measuring tube will lead to the motion component except for the main vibration frequency, which will affect the phase difference calculation and reduce the measurement accuracy of the mass flowmeter. This article presents a method to improve the accuracy of U-tube CMF based on variable step-size least-mean-square (LMS) filter and Hilbert transform with interval shifting under pulsating flow. Experimental work was conducted on a dynamic experimental platform of pulsating flow. Experimental results show that the stability and accuracy of the proposed algorithm are better than that of the traditional CMF phase difference calibration method. The mean time difference error is $9.0525~\mu \text{s}$ , and the mean time difference relative error is 5.806%. The calibration effect is more than 88.0934% better than other traditional algorithms. It is verified that it has a good error calibration effect for pulsating flow at various frequencies.

Journal ArticleDOI
TL;DR: Experimental results demonstrate that the proposed fusion solution, i.e., SEDRFuse, outperforms the state-of-the-art fusion methods in terms of both subjective and objective evaluations.
Abstract: Image fusion is an important task for computer vision as a diverse range of applications are benefiting from the fusion operation. The existing image fusion methods are largely implemented at the pixel level, which may introduce artifacts and/or inconsistencies, while the computational complexity is relatively high. In this article, we propose a symmetric encoder–decoder with residual block (SEDRFuse) network to fuse infrared and visible images for night vision applications. At the training stage, the SEDRFuse network is trained to create a fixed feature extractor. At the fusing stage, the trained extractor is utilized to extract the intermediate and compensation features, which are generated by the residual block and the first two convolutional layers from the input source images, respectively. Two attention maps, which are derived from the intermediate features, are then multiplied by the intermediate features for fusion. The salient compensation features obtained through elementwise selection are passed to the corresponding deconvolutional layers for processing. Finally, the fused intermediate features and the selected compensation features are decoded to reconstruct the fused image. Experimental results demonstrate that the proposed fusion solution, i.e., SEDRFuse, outperforms the state-of-the-art fusion methods in terms of both subjective and objective evaluations.

Journal ArticleDOI
TL;DR: This FPGA-based design delivers sufficient performance to record eye movements at high spatial and temporal precision and accuracy using coils small enough for use with small animals.
Abstract: Vestibular and oculomotor research often requires measurement of 3-D eye orientation and movement with high spatial and temporal precision and accuracy. We describe the design, implementation, validation, and use of a new magnetic coil system optimized for recording 3-D eye movements using small scleral coils in animals. Like older systems, the system design uses off-the-shelf components to drive three mutually orthogonal alternating magnetic fields at different frequencies. The scleral coil voltage induced by those fields is decomposed into three signals, each related to the coil’s orientation relative to the axis of one field component. Unlike older systems based on analog demodulation and filtering, this system uses a field-programmable gate array (FPGA) to oversample each induced scleral coil voltage (at 25 Msamples/s), demodulate in the digital domain, and average over 25 ksamples per data point to generate 1-ksamples/s output in real time. Noise floor is <0.036° peak-to-peak and linearity error is <0.1° during 345° rotations in all three dimensions. This FPGA-based design, which is both reprogrammable and freely available upon request, delivers sufficient performance to record eye movements at high spatial and temporal precision and accuracy using coils small enough for use with small animals.

Journal ArticleDOI
TL;DR: In this article, a combination of time-frequency analysis and convolutional neural network (CNN) was used to detect schizophrenia using EEG data, which achieved an accuracy of 9336% using the smoothed pseudo-Wigner-Ville distribution (SPWVD) and CNN model.
Abstract: Schizophrenia (SZ) is a psychiatric disorder characterized by cognitive dysfunctions, hallucinations, and delusions, which may lead to lifetime disability Detection and diagnosis of SZ by visual inspection is subjective, difficult, and time-consuming Electroencephalogram (EEG) signals are widely used to detect the SZ as they reflect the conditions of the brain Conventional machine learning methods involve many lengthy manual steps, such as decomposition, feature extraction, feature selection, and classification In this article, automated identification of SZ is proposed using a combination of time–frequency analysis and convolutional neural network (CNN) to overcome the limitations of feature extraction-based methods Three press button tasks are analyzed to segregate normal subjects and SZ patients The EEG signals are subjected to continuous wavelet transform, short-time Fourier transform, and smoothed pseudo-Wigner–Ville distribution (SPWVD) techniques to obtain scalogram, spectrogram, and SPWVD-based time–frequency representation (TFR) plots, respectively These 2-D plots are fed to pretrained AlexNet, VGG16, ResNet50, and CNN We have obtained an accuracy of 9336% using the SPWVD-based TFR and CNN model In comparison to the benchmark AlexNet, ResNet50, and VGG16 networks, the developed CNN model with four convolutional layers not only requires fewer learnable parameters but also is computationally efficient and fast This clearly indicates that our proposed method combining the SPWVD-CNN model has performed better than the state-of-the-art transfer learning techniques Our developed model is ready to be tested with more EEG data and can aid psychiatrists in their diagnosis

Journal ArticleDOI
TL;DR: An auxiliary classier Wasserstein generative adversarial network with gradient penalty (ACWGAN-GP) is proposed in this article, which is capable of generating high-quality samples for the minority classes stably utilizing an imbalanced training set.
Abstract: In the real scenario of engineering, the failure time of rotating machinery is generally much less than when it is in a healthy condition. Considering the cost, it is unrealistic to conduct the large-sample and long-time failure tests. This results in the problem of data imbalance in fault diagnosis, i.e., the number of normal samples far exceeds that of the fault ones, which seriously affects the accuracy and stability of fault diagnosis. For the settlement of the above problem, an auxiliary classier Wasserstein generative adversarial network with gradient penalty (ACWGAN-GP) is proposed in this article, which is capable of generating high-quality samples for the minority classes stably utilizing an imbalanced training set. In the experiment of fault diagnosis, the generated samples first go through the availability verification and then are employed to augment the imbalanced data set gradually. The final results show that the proposed method is competent for the generation of data, which is highly similar to real samples, and the accuracy of fault diagnosis has effective improvement as the imbalanced data set is gradually expanded to equilibrium. In addition, the ACWGAN-GP model presents better performance in sample generation than other widely used methods.

Journal ArticleDOI
TL;DR: An integrated global navigation satellite system/light detection and ranging (GNSS/LiDAR)-based simultaneous localization and mapping (SLAM) pose estimation framework to perform large-scale 3-D map building in partially GNSS-denied outdoor environments is presented.
Abstract: This article presents an integrated global navigation satellite system/light detection and ranging (GNSS/LiDAR)-based simultaneous localization and mapping (SLAM) pose estimation framework to perform large-scale 3-D map building in partially GNSS-denied outdoor environments. The framework takes the advantage of the complementarity between GNSS positioning and LiDAR-SLAM to decompose the map building task according to the GNSS real-time kinematic (RTK) status. When mapping in GNSS-denied scenes, a 3-D LiDAR-SLAM algorithm is adopted to estimate poses and a correction algorithm is presented to correct drift errors. On the other hand, when mapping in open scenes, a GNSS-initialized LiDAR mapping algorithm (GL-mapping) is proposed to loosely couple GNSS positioning and LiDAR data registration. It can perform the orientation estimation without the use of either the high-cost inertial sensing device or the GNSS dual-antenna. Experiments are conducted in large-scale outdoor environments to demonstrate that the proposed framework can accomplish simultaneous pose estimation and map building with high precision in both open scenes and GNSS-denied scenes.

Journal ArticleDOI
TL;DR: Li et al. as discussed by the authors proposed degradation-aware long short-term memory (LSTM) autoencoder (DELTA) to predict the remaining useful life (RUL) of industrial bearing.
Abstract: The remaining useful life (RUL) prediction plays a pivotal role in the predictive maintenance of industrial manufacturing systems. However, one major problem with the existing RUL estimation algorithms is the assumption of a single health degradation trend for different machine health stages. To improve the RUL prediction accuracy with various degradation trends, this article proposes an algorithm dubbed degradation-aware long short-term memory (LSTM) autoencoder (AE) (DELTA). First, the Hilbert transform is adopted to evaluate the degradation stage and factor with the real-time sensory signal. Second, we adopt LSTM AE to predict RUL based on multisensor time-series data and the degradation factor. Distinct from the existing studies, the proposed framework is able to dynamically model the degradation factor and explore latent variables to improve RUL prediction accuracy. The performance of DELTA is evaluated with the open-source FEMTO bearing data set. Compared with the existing algorithms, DELTA achieves appreciable improvements in the RUL prediction accuracy.

Journal ArticleDOI
TL;DR: In this paper, a novel ensemble long short-term memory neural network (ELSTMNN) model for RUL prediction is proposed to enhance the RUL prognosis accuracy and improve the adaptive and generalization abilities under different prognostic scenarios.
Abstract: Remaining useful life (RUL) prognosis is of great significance to improve the reliability, availability, and maintenance cost of an industrial equipment. Traditional machine learning method is not fit for dealing with time series signals and has low generalization and stability in prognostic. In this article, a novel ensemble long short-term memory neural network (ELSTMNN) model for RUL prediction is proposed to enhance the RUL prognosis accuracy and improve the adaptive and generalization abilities under different prognostic scenarios. The ELSTMNN contains a series of long short-term memory neural networks (LSTMNNs), each of which is trained on a unique set of historical data. A novel ensemble method is first proposed using Bayesian inference algorithm to integrate multiple predictions of the LSTMNNs for the optimal RUL estimation. The effectiveness of the ELSTMNN-based RUL prognosis method is validated using two characteristically different turbofan engine data sets. The experimental results show a competitive performance of the ELSTMNN in comparison with other prognostic methods.

Journal ArticleDOI
TL;DR: A deep adversarial capsule network (DACN) is proposed to embed multidomain generalization into the intelligent compound fault diagnosis, so that the DACN not only can intelligently decouple the compound fault into multiple single faults for industrial equipment (IE) but also can be generalized from certain working conditions to another new.
Abstract: With advanced measurement technologies and signal analytics algorithms developed rapidly, the past decades have witnessed large amount of successful breakthroughs and applications in the field of intelligent fault diagnosis (IFD). However, the historical IFD methods have difficulties for compound fault diagnosis, when labeled data cannot be collected in advance for new or extreme working conditions. Facing with such challenges, a deep adversarial capsule network (DACN) is proposed to embed multidomain generalization into the intelligent compound fault diagnosis, so that the DACN not only can intelligently decouple the compound fault into multiple single faults for industrial equipment (IE) but also can be generalized from certain working conditions to another new. First, a DACN including feature extractor (FE), decoupling classifier (DC) and multidomain classifier (MC) is constructed for feature learning, fault decoupling, and unsupervised multidomain adaptation, respectively. Second, adversarial training is introduced into the DACN in the training stage via a gradient reversal layer that can build the connection between the FE and MC, which can force the DACN to learn the domain-invariance features. Finally, the DACN is trained using the single-fault data collected under multiple working conditions, and then applied to monitor the health condition of IE under new working conditions. The cross-validation experiments have been implemented on an automobile transmission (AT), which illustrates that the DACN obtains an optimum performance with the highest average accuracy of 97.65% for compound fault diagnosis of IE under multidomain generalization task and outperforms other related methods.

Journal ArticleDOI
Kaicheng Feng1, Hengji Qin1, Shan Wu1, Weifeng Pan1, Guanzheng Liu1 
TL;DR: An SA detection model based on frequential stacked sparse auto-encoder (FSSAE) and time-dependent cost-sensitive (TDCS) classification model is proposed by combining the hidden Markov model (HMM) and the MetaCost algorithm to improve the performance of the classifier by considering temporal dependence and the imbalance problem.
Abstract: Sleep apnea (SA) is a harmful respiratory disorder that has caused widespread concern around the world. Considering that electrocardiogram (ECG)-based SA diagnostic methods were effective and human-friendly, many machine learning or deep learning methods based on ECG have been proposed by prior works. However, these methods are based on feature engineering or supervised and semisupervised learning techniques, and the feature sets are always incomplete, subjective, and highly dependent on labeled data. In addition, some related studies ignored the data imbalance problem which leads to poor performance of classifier on minority classes. In this study, an SA detection model based on frequential stacked sparse auto-encoder (FSSAE) and time-dependent cost-sensitive (TDCS) classification model was proposed. The FSSAE extracts feature set automatically with unsupervised learning technique, and the TDCS classification model is proposed by combining the hidden Markov model (HMM) and the MetaCost algorithm to improve the performance of the classifier by considering temporal dependence and the imbalance problem. In the test set, the result of per-segment classification achieved 85.1%, 86.2%, and 84.4% for accuracy, sensitivity, and specificity, respectively, proving that our method is helpful for SA detection.

Journal ArticleDOI
TL;DR: Wang et al. as discussed by the authors proposed a decoupled two-stage object detection framework based on convolutional neural networks (CNNs), wherein the localization task and the classification task are decouple through two specific modules, namely multi-hierarchical aggregation (MHA) block and locally non-local (LNL) enhancement module.
Abstract: In the integrated circuit (IC) packaging, the surface defect detection of flexible printed circuit boards (FPCBs) is important to control the quality of IC. Although various computer vision (CV)-based object detection frameworks have been widely used in industrial surface defect detection scenarios, FPCB surface defect detection is still challenging due to non-salient defects and the similarities between diverse defects on FPCBs. To solve this problem, a decoupled two-stage object detection framework based on convolutional neural networks (CNNs) is proposed, wherein the localization task and the classification task are decoupled through two specific modules. Specifically, to effectively locate non-salient defects, a multi-hierarchical aggregation (MHA) block is proposed as a location feature (LF) enhancement module in the defect localization task. Meanwhile, to accurately classify similar defects, a locally non-local (LNL) block is presented as a SEF enhancement module in the defect classification task. What is more, an FPCB surface defect detection dataset (FPCB-DET) is built with corresponding defect category and defect location annotations. Evaluated on the FPCB-DET, the proposed framework achieves state-of-the-art (SOTA) accuracy to 94.15% mean average precision (mAP) compared with the existing surface defect detection networks. Soon, source code and dataset will be available at https://github.com/SCUTyzy/decoupled-two-stage-framework .