检索结果-内蒙古大学图书馆

Depth from defocus technique with convolutional neural networks for high particle concentrations

EXPERIMENTS IN FLUIDS 2025年第1期66卷 1-11页

作者： Xu, Rixin Huang, Zuojie Zhou, Wu Tropea, Cameron Cai, Tianyi Univ Shanghai Sci & Technol Sch Energy & Power Engn Shanghai Peoples R China Shanghai Key Lab Multiphase Flow & Heat Transfer P Shanghai Peoples R China Tech Univ Darmstadt Inst Fluid Mech & Aerodynam Darmstadt Germany

Recent advantages in the depth from defocus technique for the size and location determination of particles in dispersed two-phase flows have enabled the technique to detect and analyze spherical particle images in flow systems with high number concentrations. In the present study, the use of convolutional neural networks for this task will be explored, with comparisons to the conventional analyses in terms of accuracy, tolerable concentration limits and computational speed. This approach requires a large teaching dataset of images, which is only practical and feasible if the dataset can be synthetically generated. Thus, the first development to be presented is an image generation procedure for out-of-focus neighboring spherical particles resulting in a known blurred image overlap. This image generation procedure is validated using laboratory images of known particle size distribution, position and image overlap, before creating a teaching dataset. The trained processing scheme is then applied to both synthetic datasets and to experimental data. The synthetic datasets allow limits of image overlap and tolerable volume concentration limits of the technique to be evaluated as a function of particle size distribution.(https://***/xu200911/Generate-overlapping-out-of-focus-particles)

关键词： convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Multiperspective Temporal Pooling convolutional neural networks for Fault Diagnosis of Mechanical Transmission Systems

引用

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2025年 74卷

作者： Xu, Yadong Li, Sheng Feng, Ke Sun, Beibei Yang, Xiaolong Kou, Linlin Zhao, Zhiheng Huang, George Q. Hong Kong Polytech Univ Dept Ind & Syst Engn Hong Kong Peoples R China Nanjing Forestry Univ Coll Civil Engn Nanjing 210037 Peoples R China Xi An Jiao Tong Univ Sch Mech Engn Xian 710049 Peoples R China Southeast Univ Sch Mech Engn Nanjing 211189 Peoples R China Nanjing Univ Sci & Technol Sch Mech Engn Nanjing 210094 Peoples R China Beijing Mass Transit Railway Operat Corp Ltd Beijing 100044 Peoples R China Hong Kong Polytech Univ Res Inst Adv Mfg Hong Kong Peoples R China Huazhong Univ Sci & Technol State Key Lab Intelligent Mfg Equipment & Technol Wuhan 430072 Peoples R China

The rapid development of convolutional neural networks (CNNs) has significantly contributed to the progress of intelligent fault diagnosis of mechanical transmission systems. Nevertheless, a significant number of prevailing CNN-based diagnostic models may suffer from two notable constraints. First, the existing models often employ fixed temporal pooling for feature extraction, which restricts their ability to effectively capture and analyze a comprehensive range of temporal information. Second, these models may struggle to precisely forecast the operational state of the monitored machinery amidst nonstationary circumstances, such as time-varying or disturbed environments. These challenges limit their feature extraction capabilities and hinder their practical implementation and utilization. To tackle the aforementioned issues, this study develops a multiperspective temporal pooling CNN (MTPCNN). The main contributions encompass: 1) a multikernel feature perception module (MFPM) and a balanced attention module (BAM) are established for multilevel information exploration and optimal feature selection and 2) an innovative multiperspective temporal pooling learning (MTPL) strategy is introduced to aid the model in dynamically selecting the optimal temporal pooling method for the input data. A laboratory dataset collected from a gearbox fault simulator and an industrial dataset collected from a high-speed rail are used for the validation of the proposed approach. The extensive experimental results validate the superiority of the developed MTPCNN model over seven competitive approaches.

关键词： Feature extraction convolutional neural networks Monitoring Fault diagnosis Convolution Noise reduction Machinery Vibrations Condition monitoring Prediction algorithms Balanced attention module (BAM) convolutional neural network (CNN) fault diagnosis multikernel feature perception module (MFPM) multiperspective temporal pooling learning (MTPL)

来源：评论

学校读者我要写书评

暂无评论

STRUCTURED PRUNING FOR GROUP REGULARIZED convolutional neural networks VIA DYNAMIC REGULARIZATION FACTOR

引用

JOURNAL OF INDUSTRIAL AND MANAGEMENT OPTIMIZATION 2025年第2期21卷 1440-1455页

作者： Li, Feng Li, Bo Zhu, Meijiao Ma, Junchi Yuan, Jinlong Dalian Maritime Univ Sch Sci Dalian 116026 Liaoning Peoples R China Liaoning Normal Univ Sch Math Dalian 116029 Liaoning Peoples R China

High demand for computation and storage largely hinders the deployment of deep convolutional neural networks (CNNs) in resource constrained devices. Regularization-based pruning methods are effective for the model compression of CNNs by introducing a regularization term into the loss function. However, such methods typically compress weights to zero utilising a fixed regularization factor, which cannot impose different punishments on weights or structures with different importance. In this paper, a group regularization method is proposed to impose a dynamic regularization factor to each filter of convolutional layers, named GRDRF. The scaling factors of batch normalization (BN) layer are used to represent the importance of filters and construct updatable and dynamic regularization factors according to the importance representation. Moreover, a novel pruning strategy of ResNet is proposed, which can significantly improve the pruning performance. Numerical experiments on the popular CNN architectures and CIFAR-10 dataset are performed to verify the efficiency of the new group regularization method and the ResNet pruning strategy. The programs are publicly provided on https://***/xiaoli7766111/DRF.

关键词： Structured pruning convolutional neural networks group regularization method BN layer scaling factor dynamic regularization factor

来源：评论

学校读者我要写书评

暂无评论

An interpretable operational state classification framework for elevators through convolutional neural networks

引用

COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING 2025年

作者： Olaizola, Jon Izagirre, Unai Serradilla, Oscar Zugasti, Ekhi Mendicute, Mikel Aizpurua, Jose I. Mondragon Univ Elect & Comp Dept Gipuzkoa Spain Laboral Kutxa Treasury Gipuzkoa Spain Univ Basque Country UPV EHU Comp Sci & Artificial Intelligence Dept Donostia San Sebastian Gipuzkoa Spain Basque Fdn Sci Ikerbasque Gipuzkoa Spain

Ensuring the safe, reliable, and cost-efficient operation of transportation systems such as elevators is critical for the maintenance of civil infrastructures. The ability to monitor the health state and classify different operational states (elevator moving up/down, stopped, doors opening/closing) may lead to the development of intelligent solutions, such as diagnostics and predictive maintenance. Accordingly, downtime and maintenance costs can be significantly reduced with an accurate monitoring of the operation parameters and dynamics. In this context, this paper presents a novel approach for the operational state classification of elevator systems based on a one-dimensional convolutional neural network, using exclusively a single axis (Z) of an accelerometer signal. The proposed model utilizes a single accelerometer and addresses the challenge of distinguishing overlapping signal patterns, such as those produced by vertical displacement and door movements. The approach includes an interpretability stage, which demonstrates the data processing involved in extracting features from the underlying physical phenomena captured in the acceleration signal. Obtained results have been validated with an on-site captured dataset which contains 250 elevator journeys and compared with three other classification methods that have been conventionally used: generalized likelihood ratio test (GLRT), barometer-assisted GLRT, and three conventional machine learning modelss. It has been shown that the proposed approach is very accurate, with 96% of the average F1 score and, importantly, includes the analytic relation of the classification model features.

关键词： convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Deep learning-based stochastic ground motion modeling using generative adversarial and convolutional neural networks

引用

SOIL DYNAMICS AND EARTHQUAKE ENGINEERING 2025年 194卷

作者： Masoudifar, Mohsen Mahsuli, Mojtaba Taciroglu, Ertugrul Sharif Univ Technol Ctr Infrastruct Sustainabil & Resilience Res Dept Civil Engn Tehran Iran Univ Calif Los Angeles Dept Civil & Environm Engn Los Angeles CA USA

This paper proposes a probabilistic framework for generating three-dimensional (3D) synthetic ground motions using deep learning techniques-specifically, generative adversarial networks (GAN) and convolutional neural networks (CNN). Deep learning methods have been shown to surpass classical model classes in performance when provided with large datasets, and the ever-increasing number of ground motion records provides an opportunity to design generative models to produce artificial ground motions that outperform classical models. In addition, these methods can directly extract features and patterns from ground motion data without loss of generality, enabling prediction and generation of synthetic ground motions. The proposed framework consists of two distinct deep learning modules. The first generates normalized 3D synthetic ground motions given source and site characteristics. For this purpose, a conditional Wasserstein GAN comprising a generator and a critic in an adversarial setup is designed in which they engage in a simultaneous competitive process. Through learning from the dataset of real ground motions, the generator attempts to generate artificial ground motions that are more convincing to the critic, whereas the critic seeks to improve its ability to identify the realness or artificialness of the motions and provide the generator with feedback. The second module produces peak ground accelerations (PGA) for the three spatial components of the generated normalized ground motion, given the normalized motion and the said characteristics. For this purpose, a CNN is designed with "inception" layers, each of which concurrently applies multiple convolution filters of varying sizes to the input and concatenates their outputs, enabling the network to efficiently capture features at various scales. The learning performance of both modules is improved by realistic data augmentation techniques that increase training data size and are specifically designed for 3D ground

关键词： Artificial Ground Motion Deep Learning Generative Adversarial Network convolutional neural networks Ground Motion Prediction Model

来源：评论

学校读者我要写书评

暂无评论

HARDCORE: H-Field and Power Loss Estimation for Arbitrary Waveforms With Residual, Dilated convolutional neural networks in Ferrite Cores

引用

IEEE TRANSACTIONS ON POWER ELECTRONICS 2025年第2期40卷 3326-3335页

作者： Kirchgaessner, Wilhelm Foerster, Nikolas Piepenbrock, Till Schweins, Oliver Wallscheid, Oliver Paderborn Univ Dept Power Elect & Elect Drives D-33095 Paderborn Germany Univ Siegen Chair Interconnected Automat Syst D-57076 Siegen Germany

The MagNet challenge 2023 called upon competitors to develop data-driven models for the material-specific, waveform-agnostic estimation of steady-state power losses in toroidal ferrite cores. The following HARDCORE (H-field and power loss estimation for arbitrary waveforms with residual, dilated convolutional neural networks in ferrite cores) approach shows that a residual convolutional neural network with physics-informed extensions can serve this task efficiently when trained on observational data beforehand. One key element is an intermediate model layer, which first reconstructs the $BH$ curve, and then, estimates the power losses based on the curve's area rendering the proposed topology physically interpretable. In addition, emphasis was placed on expert-based feature engineering and information-rich inputs in order to enable a lean model architecture. A model is trained from scratch for each material, while the topology remains the same. A Pareto-style tradeoff between model size and estimation accuracy is demonstrated, which yields an optimum at as low as 906 parameters and down to below 8% for the average 95th percentile of the relative power loss error across diverse materials. This contribution has won the first place in the performance category of the MagNet challenge 2023, which further highlights the effectiveness of the proposed model.

关键词： Mathematical models Estimation Data models convolutional neural networks Accuracy Magnetic hysteresis Magnetic cores Temperature measurement Magnetic domains Temperature distribution convolutional neural network (CNN) machine learning (ML) magnetics

来源：评论

学校读者我要写书评

暂无评论

Improving multi-talker binaural DOA estimation by combining periodicity and spatial features in convolutional neural networks

引用

EURASIP JOURNAL ON AUDIO SPEECH AND MUSIC PROCESSING 2025年第1期2025卷 1-18页

作者： Varzandeh, Reza Doclo, Simon Hohmann, Volker Carl von Ossietzky Univ Oldenburg Dept Med Phys & Acoust D-26111 Oldenburg Germany Carl von Ossietzky Univ Oldenburg Cluster Excellence Hearing4all D-26111 Oldenburg Germany

Deep neural network-based direction of arrival (DOA) estimation systems often rely on spatial features as input to learn a mapping for estimating the DOA of multiple talkers. Aiming to improve the accuracy of multi-talker DOA estimation for binaural hearing aids with a known number of active talkers, we investigate the usage of periodicity features as a footprint of speech signals in combination with spatial features as input to a convolutional neural network (CNN). In particular, we propose a multi-talker DOA estimation system employing a two-stage CNN architecture that utilizes cross-power spectrum (CPS) phase as spatial features and an auditory-inspired periodicity feature called periodicity degree (PD) as spectral features. The two-stage CNN incorporates a PD feature reduction stage prior to the joint processing of PD and CPS phase features. We investigate different design choices for the CNN architecture, including varying temporal reduction strategies and spectro-temporal filtering approaches. The performance of the proposed system is evaluated in static source scenarios with 2-3 talkers in two reverberant environments under varying signal-to-noise ratios using recorded background noises. To evaluate the benefit of combining PD features with CPS phase features, we consider baseline systems that utilize either only CPS phase features or combine CPS phase and magnitude spectrogram features. Results show that combining PD and CPS phase features in the proposed system consistently improves DOA estimation accuracy across all conditions, outperforming the two baseline systems. Additionally, the PD feature reduction stage in the proposed system improves DOA estimation accuracy while significantly reducing computational complexity compared to a baseline system without this stage, demonstrating its effectiveness for multi-talker DOA estimation.

关键词： convolutional neural networks Spatial feature Periodicity feature Binaural DOA estimation Multiple talkers Feature reduction Hearing devices.

来源：评论

学校读者我要写书评

暂无评论

Improve the interpretability of convolutional neural networks with probability density function

引用

INFORMATION SCIENCES 2025年 699卷

作者： Chen, Yueqi Pan, Tingting Yang, Jie Dalian Univ Technol Sch Math Sci Dalian 116024 Liaoning Peoples R China Key Lab Computat Math & Data Intelligence Liaoning Dalian 116024 Liaoning Peoples R China Dalian Polytech Univ Dept Basic Courses Teaching Dalian 116034 Peoples R China

Currently, convolutional neural networks (CNNs) have demonstrated extensive success in numerous practical applications. Nevertheless, their limited interpretability remains a significant barrier to further advancement in certain crucial fields. Improving the interpretability of CNNs stands as an exceptionally compelling topic in the present time. This paper explores the interpretability of a basic CNN incorporating a convolution-pooled block and a fully connected layer from a statistical perspective. Assuming that the input variables adhere to a normal distribution and maintain independence from each other, the output variables subsequent to the convolution and pooling layers also conform to a normal distribution. Simultaneously, the probability density function (pdf) characterizing the final output variable belongs to an exponential family distribution. By introducing intermediate variables, the pdf of this output variable can be expressed as a linear combination of three distinct normal distributions. Furthermore, the likelihood of the predicted class label can be rewritten as a cumulative density function (cdf) of the standard normal distribution. The originality of this paper lies in its provision of a more innovative and intuitive perspective for dissecting the operational mechanism of CNNs, analyzing them layer by layer to improve their interpretability. Experimental results obtained from both an artificial dataset and the image datasets CIFAR-10 and ImageNet further validate the rationality of these conclusions.

关键词： Interpretability convolutional neural networks Probability density function

来源：评论

学校读者我要写书评

暂无评论

Fault diagnosis method for chemical processes based on variable correlation-guided convolutional neural networks

引用

CANADIAN JOURNAL OF CHEMICAL ENGINEERING 2025年

作者： Zhou, Zhe Yu, Hongwei Yang, Li Li, Zuxin Wen, Chenglin Huzhou Univ Sch Engn Huzhou Zhejiang Peoples R China Huzhou Univ Huzhou Key Lab Intelligent Sensing & Optimal Contr Huzhou Zhejiang Peoples R China Zhejiang Key Lab Ind Solid Waste Thermal Hydrolysi Huzhou Zhejiang Peoples R China Huzhou Coll Sch Intelligent Mfg Huzhou Zhejiang Peoples R China Guangdong Univ Petrochem Technol Sch Automat Maoming Guangdong Peoples R China

convolutional neural networks (CNNs) have been widely applied in chemical process fault diagnosis due to their superior feature extraction capabilities. However, the inherent complexity and variability of chemical environments, involving multivariable interactions, noise interference, and other factors, pose challenges that hinder the direct application of CNNs. These limitations may compromise the accuracy of fault diagnosis by hindering the full exploitation of CNNs' feature extraction capabilities. To address these challenges, this paper proposes a novel fault diagnosis method for chemical processes based on a variable correlation-guided CNN. The proposed method uses the Pearson correlation coefficient to identify strongly correlated variable groups, integrating them into the original variables. This integration facilitates the convolution of these strongly correlated variables, thereby enhancing the extraction of more discriminative features and optimizing fault diagnosis methods. This approach enables CNNs to more accurately extract fault-relevant features, thereby improving diagnostic performance. The effectiveness of the proposed method is validated through comprehensive numerical simulations and the Tennessee Eastman (TE) process dataset. The results demonstrate substantial enhancements in both the accuracy and reliability of fault detection, validating the superiority of the proposed method.

关键词： convolutional neural networks fault diagnosis Pearson correlation coefficient

来源：评论

学校读者我要写书评

暂无评论

Identification of Vaca Muerta shale microlithofacies using convolutional neural networks with characterization by electron microscopy

GAS SCIENCE AND ENGINEERING

引用

GAS SCIENCE AND ENGINEERING 2025年 134卷

作者： Froute, Laura Nazarova, Marfa Jolivet, Isabelle C. Creux, Patrice Chaput, Eric Kovscek, Anthony R. Stanford Univ Dept Energy Sci & Engn Stanford CA 94305 USA Univ Pau & Pays Adour E2S UPPA LFCR F-64000 Pau France TotalEnergies SE One Tech Geosci & Reservoir F-64000 Pau France

Identifying sedimentary lithofacies, including their characteristics and distribution, is a common method to describe geological heterogeneity. Characterizing microlithofacies within shale fabric requires the integration of multi-scale, multimodal data. In this study, we identify 6 representative microlithofacies within a Vaca Muerta shale sample using convolutional neural networks. We characterize the two main porous microfacies using Focused Ion Beam-Scanning Electron Microscopy (FIB-SEM) and Scanning Transmission Electron Microscopy (STEM) imaging to understand the characteristics and roles of connected pores. The coccolithic mud microfacies, rich in carbonates, contains well-connected organic pores, intraparticle pores within clays, as well as interparticle pores in between grains. The organic-rich clayey mud microfacies has a large and well connected organic pore network. In both microfacies, the larger pores playa key role in connecting the abundant organic nanopores to meso- and macro-scale flow paths. Importantly, 2D information in FIB-SEM is insufficient to characterize fully the nanopores, and 3D STEM tomography is needed to probe connectivity. Given the impact of nanoscale processes on hydrocarbon production, high-resolution imaging techniques are necessary to characterize fully the formation. In particular, we show that most of the porosity exists below 100 nm for both microfacies, and abundant organic nanopores can be as small as a few nanometers.

关键词： Microlithofacies Vaca Muerta convolutional neural networks Electron microscopy Image analysis

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：