检索结果-内蒙古大学图书馆

A hybrid convolutional autoencoder training algorithm for unsupervised bearing health indicator construction

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 2025年 139卷

作者： Milani, Ali Eftekhari Zappala, Donatella Watson, Simon J. Delft Univ Technol Kluyverweg 1 NL-2629 HS Delft Netherlands

Conventional Deep Learning (DL) methods for bearing health indicator (HI) adopt supervised approaches, requiring expert knowledge of the component degradation trend. Since bearings experience various failure modes, assuming a particular degradation trend for HI is suboptimal. Unsupervised DL methods are scarce in this domain. They generally maximise the HI monotonicity built in the middle layer of an autoencoder (AE) trained to reconstruct the run-to-failure signals. The backpropagation (BP) training algorithm is unable to perform this maximisation since the monotonicity of HI subsections corresponding to input sample batches does not guarantee the monotonicity of the whole HI. Therefore, existing methods achieve this by searching AE hyperparameters so that its BP training to minimise the reconstruction error also leads to a highly monotonic HI in its middle layer. This is done using expensive search algorithms where the AE is trained numerous times using various hyperparameter settings, rendering them impractical for large datasets. To address this limitation, a small convolutional autoencoder (CAE) architecture and a hybrid training algorithm combining Particle Swarm Optimisation and BP are proposed in this work to enable simultaneous maximisation of the HI monotonicity and minimisation of the reconstruction error. Asa result, the HI is built by training the CAE only once. The results from three case studies demonstrate this method's lower computational burden compared to other unsupervised DL methods. Furthermore, the CAE-based HIs outperform the indicators built by equivalent and significantly larger models trained with a BP-based supervised approach, leading to 85% lower remaining useful life prediction errors.

关键词： Particle Swarm Optimisation convolutional autoencoder Bearing Health indicator Prognostics Remaining useful life prediction

来源：评论

学校读者我要写书评

暂无评论

CAE SynthImgGen: Revolutionizing cancer diagnosis with convolutional autoencoder-based synthetic image generation

引用

ALEXANDRIA ENGINEERING JOURNAL 2025年 115卷 343-354页

作者： Hangaragi, Shivalila Neelima, N. Venugopal, Vivek Ganguly, Somnath Mudi, Joyti Choi, Joon-Ho Amrita Vishwa Vidyapeetham Dept Elect & Commun Engn Amrita Sch Engn Bengaluru India Chonnam Natl Univ Dept Elect Engn Gwangju South Korea Bankura Unnayani Inst Engn Dept Elect Engn Bankura W Bengal India

Medical imaging is pivotal in modern healthcare, offering a visual window into the human body's intricate structures and functions. However, the scarcity of diverse and representative medical images significantly limits research progress. Today, deep learning algorithms are increasingly incorporated into the medical imaging domain to automate the diagnostic process. The success of these algorithms relies heavily on vast and diverse datasets. Insufficient data hampers the training and validation of these algorithms, resulting in suboptimal performance, biased results, and reduced generalizability. This paper introduces a pioneering approach that employs a convolutional autoencoder (CA) to synthetically generate medical images. The synthetic images produced are then added to the existing database to create an augmented dataset. This augmented dataset is subsequently used for classification with a convolutional neural network. Experiments were conducted on publicly available datasets-the chest CT-scan dataset and the IQ-OTH/NCCD lung cancer dataset. The synthetic image generation capability of the CA was compared with traditional augmentation methods such as flipping, rotating, shearing, shifting, zooming, and sub-sampling. Results showed that the CA-based augmented dataset achieved an accuracy of 91 %, compared to 83 % with the traditional augmentation-based dataset.

关键词： convolutional autoencoder Synthetic image generation convolutional neural network SSIM Lung cancer classification

来源：评论

学校读者我要写书评

暂无评论

Reduced-order modeling via convolutional autoencoder for emulating combustion of hydrogen/methane fuel blends

引用

COMBUSTION AND FLAME 2025年 274卷

作者： Ding, Siyu Ni, Chenxu Chu, Xu Lu, Qingzhou Wang, Xingjian Tsinghua Univ Dept Energy & Power Engn Beijing 100084 Peoples R China

Numerical simulations are essential in comprehending the processes of flow and combustion in aerospace and power-generation systems. Yet, substantial computational demands render these simulations prohibitively expensive for design and optimization, owing to the large amount of design parameters to be surveyed in a wide design space. This study presents parametric reduced order models (ROMs) that leverage deep neural network- based dimension reduction through a convolutional autoencoder (AE) to emulate spatial distributions of physical flowfields in combustion problems. The proposed AE-based ROMs integrate multiple advancements, including design of experiment, nonlinear dimension reduction, and regression methods like kriging and deep neural networks (DNN) to enable parametric predictability. For comparison, proper orthogonal decomposition (POD)- based ROMs are also developed. Two distinct test scenarios are outlined: steam-diluted methane/hydrogenblending oxy-combustion from a triple-coaxial nozzle and hydrogen-enriched combustion in a practical aeroderivative combustor. Results suggest that ROMs with kriging show superior performance against those with DNN. In both test scenarios, AE-based ROMs exhibit better prediction accuracy in emulating spatial distributions for physical variables of interest than POD-based ROMs. This is mainly due to the effective capture of nonlinear features by AE through intricate network structure. Such nonlinearity effect is examined by introducing interpretable "AE modes", which demonstrate multiscale characteristics, distinct from POD modes, enabling a more detailed representation of local flow features. Both AE- and POD-based ROMs achieve a dramatic acceleration in prediction process, 8-9 orders of magnitude faster than traditional combustion simulations. While POD-based ROMs slightly underperform concerning prediction accuracy, they remain competitive in terms of training and prediction efficiency with a limited number of reduced ba

关键词： Reduced-order model (ROM) Kriging convolutional autoencoder Nonlinear feature Proper orthogonal decomposition

来源：评论

学校读者我要写书评

暂无评论

A Fuzzy-Operated convolutional autoencoder for Classification of Wearable Device-Collected Electrocardiogram

引用

IEEE TRANSACTIONS ON FUZZY SYSTEMS 2025年第1期33卷 267-277页

作者： Xing, Lumin Li, Xin Liu, Wenjian Wang, Xing Shandong First Med Univ Jinan 250014 Peoples R China Shandong Prov Qianfoshan Hosp Jinan 250014 Peoples R China City Univ Macau Taipa 999078 Peoples R China Shandong Univ Polit Sci & Law Jinan 250014 Peoples R China Shandong First Med Univ & Shandong Acad Med Sci Jinan 250117 Peoples R China

Utilizing electrocardiogram (ECG) to assess patients' cardiac health is currently the popular diagnostic method. This approach demonstrates high accuracy and helps alleviate the workload of medical professionals. However, the ECG signals consist of physiological signals with low-frequency and low-amplitude features, including various interference noises. Traditional feature extraction methods of classification models have proven ineffective in handling this data type. Moreover, previous classification models fail to capture the spatial and local structures within ECG data, resulting in learned features that do not include the key features of ECG signals. In addition, traditional models often adopt neural network as prediction modules, which are likewise inefficient in handling noise within the data. To end these issues, we propose a fuzzy-operated convolutional autoencoder (FOCAE) for the classification of ECG collected from wearable devices. The FOCAE model integrates convolutional autoencoder and fuzzy neural network, enabling it to learn high-quality features from noisy data to ensure the model's inferential capability. Furthermore, the FOCAE model inputs the learned features into the fuzzy neural network and fully connected network to obtain the classification results of ECG. Using fuzzy rules in the fuzzy neural network to describe the mapping between features and classification results enhances the robustness of the predictive outcomes. Some experimental results on a real ECG dataset validate the superior performance of the FOCAE model. The performance of the FOCAE model is superior to other three baseline models. Specifically, the Precision, Accuracy, and F1-Score of the FOCAE model are 0.971, 0.968, and 0.965, respectively.

关键词： Electrocardiography Feature extraction Data models Predictive models Arrhythmia Noise Fuzzy neural networks Classification task convolutional autoencoder electrocardiography fuzzy neural network (FNN) wearable devices

来源：评论

学校读者我要写书评

暂无评论

convolutional autoencoder-based ground motion clustering and selection

引用

SOIL DYNAMICS AND EARTHQUAKE ENGINEERING 2025年 191卷

作者： Jia, Yiming Sasani, Mehrdad Univ Calif Berkeley Dept Civil & Environm Engn Berkeley CA 94720 USA Northeastern Univ Dept Civil & Environm Engn Boston MA 02115 USA

Ground motion selection has become increasingly central to the assessment of earthquake resilience. The selection of ground motion records for use in nonlinear dynamic analysis significantly affects structural response. This, in turn, will impact the outcomes of earthquake resilience analysis. This paper presents a new ground motion clustering algorithm, which can be embedded in current ground motion selection methods to properly select representative ground motion records that a structure of interest will probabilistically experience. The proposed clustering-based ground motion selection method includes four main steps: 1) leveraging domainspecific knowledge to pre-select candidate ground motions;2) using a convolutional autoencoder to learn low-dimensional underlying characteristics of candidate ground motions' response spectra - i.e., latent features;3) performing k-means clustering to classify the learned latent features, equivalent to cluster the response spectra of candidate ground motions;and 4) embedding the clusters in the conditional spectra-based ground motion selection. The selected ground motions can represent a given hazard level well (by matching conditional spectra) and fully describe the complete set of candidate ground motions. Three case studies for modified, pulse-type, and non-pulse-type ground motions are designed to evaluate the performance of the proposed ground motion clustering algorithm (convolutional autoencoder +k-means). Considering the limited number of pre-selected candidate ground motions in the last two case studies, the response spectra simulation and transfer learning are used to improve the stability and reproducibility of the proposed ground motion clustering algorithm. The results of the three case studies demonstrate that the convolutional autoencoder +k-means can 1) achieve 100 % accuracy in classifying ground motion response spectra, 2) correctly determine the optimal number of clusters, and 3) outperform established cluster

关键词： Ground motion selection convolutional autoencoder K -means clustering Response spectra Conditional spectra Machine learning

来源：评论

学校读者我要写书评

暂无评论

AGCTO: Attributed Graph Clustering With Transitive Order convolutional autoencoder

引用

IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTATIONAL INTELLIGENCE 2025年

作者： Xie, Ying Wang, Jixiang Liu, Weihua Wang, Hao Pan, Yushan Wen, Lijie Xu, Rongbin Ding, Weiping Yang, Yun Sch Mech Elect & Informat Engn Putian 351100 Peoples R China Putian Univ Fujian Key Lab Financial Informat Proc Putian 351100 Peoples R China China Mobile Res Inst Beijing 100053 Peoples R China Xidian Univ Sch Cyber Engn Xian 710071 Peoples R China Xian Jiaotong Liverpool Univ Dept Comp Suzhou 215123 Peoples R China Tsinghua Univ Sch Software Beijing 100084 Peoples R China Nantong Univ Sch Artificial Intelligence & Comp Sci Nantong 226019 Peoples R China City Univ Macau Facultyof Data Sci Macau 999078 Peoples R China Swinburne Univ Technol Dept Comp Technol Melbourne Vic 3022 Australia

Attributed graphs, which associate vertices or edges with attributes, are essential in applications like social network analysis, recommendation systems, bioinformatics, and healthcare analysis. However, their high-dimensional and non-Euclidean nature complicates tasks such as vertex classification, community detection, graph visualization, and graph embedding. Existing techniques often inadequately integrate structural and attribute data, and reconstruction-centric models can compromise graph topology. Additionally, many methods require extensive manual hyperparameter tuning. To address these challenges, we introduce Attributed Graph Clustering with Transitive Order convolutional autoencoder (AGCTO). AGCTO uses a Graph Transitive convolutional autoencoder (GTCAE) framework that integrates structural and attribute information. A key innovation is the Graph ORder Distance (GORD), capturing topological relationships to enhance clustering performance. AGCTO's unified loss function combines GTCAE loss on the original graph, GTCAE loss on a simplified graph via GORD, and a topology-preserving loss. This ensures effective fusion of attribute and structure data and preserves graph topology. Evaluations on four real-world and two synthetic attributed graphs show AGCTO's superiority in clustering accuracy, normalized mutual information, F1-score, and Q-modularity. AGCTO offers a robust solution for attributed graph clustering, maintaining graph topology and delivering superior performance.

关键词： autoencoders Feature extraction Topology Vectors Fuses Decoding Data mining Manuals Image edge detection convolutional codes GNNs clustering embedding convolutional autoencoder

来源：评论

学校读者我要写书评

暂无评论

Unsupervised defect detection for solar photovoltaic cells based on convolutional autoencoder

引用

NONDESTRUCTIVE TESTING AND EVALUATION 2025年

作者： Zhang, Yufei Zhang, Xu Tu, Dawei Shanghai Univ Sch Mechatron Engn & Automat Shanghai Peoples R China

Solar photovoltaic (PV) cells are inevitably subject to defects during the production process, affecting their power generation efficiency and life. Electroluminescence (EL) imaging is the mainstream non-destructive method for PV cell defect detection. Aiming at PV cell EL images, an unsupervised defect detection method was proposed. Specifically, an unsupervised convolutional autoencoder (CAE), the scale structure perception convolutional autoencoder (SSP_CAE), was constructed, whose Squeeze-and-Excitation Attention (SE Attention) and skip connections avoid the blurring of image structure information and the loss of pixel-level details in the encoding and decoding process. Furthermore, to balance the global and local information of the image, a scale perception loss function called SP_SSIM was proposed for model training. The defect segmentation was achieved by using Otsu thresholding method to binarize the obtained Mean Absolute Error (MSE) residual heat image in the testing stage of the model. Finally, the experiments were performed on the test dataset and the experimental results showed that the proposed SSP_CAE can effectively detect PV cell defects. The experimentally obtained defect detection performance metrics Precision, Recall, IoU, F1-score and AUROC values were 0.739, 0.886, 0.723, 0.764 and 0.841, respectively. Compared with other classical methods, the proposed SSP_CAE had a better comprehensive performance for defect detection.

关键词： Defect detection PV cell convolutional autoencoder unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

Anomaly detection using convolutional autoencoder with residual gated recurrent unit and weak supervision for photovoltaic thermal heat pump system

引用

JOURNAL OF BUILDING ENGINEERING 2025年 100卷

作者： John, Lukudu Samuel Yoon, Sungmin Li, Jiteng Wang, Peng Dalian Univ Technol Fac Infrastruct Engn Dalian 116024 Peoples R China Sungkyunkwan Univ Dept Global Smart City Suwon 16419 South Korea Sungkyunkwan Univ Sch Civil Architectural Engn & Landscape Architect Suwon 16419 South Korea Sungkyunkwan Univ Coll Engn Built Environm Res Ctr Suwon 16419 South Korea

Utilizing convolutional autoencoders with Residual Bi-Directional Gated Recurrent Unit bottleneck offers an advanced approach for detecting anomalies in heat pumps. Given the importance of heat pumps in achieving decarbonization goals for residential heating and cooling, accurate diagnosis of their health issues is essential for improving their efficiency and reducing environmental impact. Traditional diagnostic techniques, such as visual inspection, thermography, electrical testing, pressure measurements and temperature differentials require skilled technicians. These methods face some challenges due to the complexities of heat pump systems, which include components like compressors, coils, valves, and therefore require expertise in diagnosing their interactions. Our approach leverages convolutional autoencoders with Gated Recurrent Unit and Weak supervision to automatically detect anomalies and label significantly accelerating the diagnostic process. Our results show that thresholds based on the rolling mean outperform thresholds based on the actual errors by 6-10 %, depending on the random seeds. Additionally, the weak labels generated exhibit a positive correlation of at least 0.5 with error thresholds and 0.62 with rolling mean predictions.

关键词： Anomaly detection convolutional autoencoder Weak supervision Photovoltaic thermal heat pump system Machine learning Unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

convolutional autoencoder Based Feature Extraction and Clustering for Customer Load Analysis

引用

IEEE TRANSACTIONS ON POWER SYSTEMS 2020年第2期35卷 1048-1060页

作者： Ryu, Seunghyoung Choi, Hyungeun Lee, Hyoseop Kim, Hongseok Sogang Univ Dept Elect Engn Seoul 04107 South Korea Encored Technol Inc Seoul 06109 South Korea

As the number of smart meters increases, compression of metering data becomes essential for data transmission, storing and processing perspectives. Specifically, feature extraction can be used for the compression of metering data and further be utilized for smart grid applications such as customer clustering. So far, there are many studies for compression and clustering based on daily load profiles. However, in order to account for long-term characteristics of electricity load, utilizing yearly load profiles (YLPs) is vital for customer load clustering and analysis. In this paper, we propose a deep learning-based YLP feature extraction that jointly captures daily and seasonal variations. By leveraging convolutional autoencoder (CAE), YLPs in 8,640-dimensional space are compressed to 100-dimensional vectors. We apply the proposed CAE framework to YLPs of 1,405 residential customers and verify that the proposed CAE outperforms other dimensionality reduction methods in terms of reconstruction errors, e.g., by 19-40%, or the compression ratio is increased by 130% or higher than other methods for the same reconstruction error. In addition, clustering analysis is performed on the encoded YLPs. Our results confirm that year-round characteristics are well captured during the clustering process and also clearly visualized with load images.

关键词： convolutional autoencoder convolutional neural network clustering deep learning load profile smart grid

来源：评论

学校读者我要写书评

暂无评论

convolutional autoencoder based reconstruction of vascular structures in photoacoustic images

Convolutional autoencoder based reconstruction of vascular s...

引用

Conference on Biomedical Spectroscopy, Microscopy, and Imaging / SPIE Photonics Europe Conference

作者： Haq, Israr Ul Kawahara, Yoshinobu RIKEN Ctr Adv Intelligence Project Wako Saitama Japan Kyushu Univ Inst Math Ind Fukuoka Japan

ISBN: (纸本)9781510634909;9781510634916

In medical imaging, a number of imaging modalities have been used to visualize the structural information of different internal organs in the human body and some modalities can even visualize structures at a cellular level for diagnostic and treatment purposes. Optical resolution photoacoustic microscopy (OR-PAM) is one of the emerging imaging modalities to analyze the anatomy and functionality of tissues non-invasive. It is a hybrid imaging technology that combines photoacoustic (PA) contrast and acoustic resolution to reconstruct images of tissues in humans and animals. However, in OR-PAM the received ultrasonic signal by the thermal expansion of tissues has a low signal-to-noise ratio because most of the signal power is lost in the conversion process from light to acoustic waves which makes it difficult to visualize the structural information in the PA images. Traditional denoising methods such as wiener filter, bandpass filter, wavelet-based denoising, noise reduction by SVD and dictionary-based denoising methods can denoise PA images to some extent but it is still a difficult task to preserve structural information in the images by such methods. In this research, a convolutional autoencoder (CAE) based model is proposed to denoise and learn the structural patterns of blood vessels in PA images. To achieve this task, a CAE model is first trained between noisy and Gabor filtered sub-images, those contain the patterns of different vascular structures. Then, the trained model is used to approximate the denoise version of the input noisy sub-images of blood vessels. The proposed model is trained and tested on PA images of blood vessels of a mouse ear, acquired by the OR-PAM imaging system and the results show that our proposed method can effectively approximate and reconstruct the noisy vascular structures than traditional images denoising/filtering methods.

关键词： convolutional autoencoder Blood vessel reconstruction Denoise vascular structures Photoacoustic imaging Optical imaging

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：