检索结果-内蒙古大学图书馆

A deep-learning-based surrogate modeling method with application to plasma processing

CHEMICAL ENGINEERING RESEARCH & DESIGN 2024年 211卷 299-317页

作者： Liu, Pengwei Wu, Qinxin Ren, Xingyu Wang, Yian Ni, Dong Zhejiang Univ Coll Control Sci & Engn Hangzhou 310027 Zhejiang Peoples R China Zhejiang Univ Polytech Inst Hangzhou 310027 Zhejiang Peoples R China Zhejiang Univ State Key Lab Ind Control Technol Hangzhou 310027 Zhejiang Peoples R China

Rapid and accurate system evolution predictions are crucial in scientific and engineering research. However, the complexity of processing systems, involving multiple physical field couplings and slow convergence of iterative numerical algorithms, leads to low computational efficiency. Hence, this paper introduces a systematic deep-learning-based surrogate modeling methodology for multi-physics-coupled process systems with limited data and long-range time evolution, accurately predicting physics dynamics and considerably improving computational efficiency and generalization. The methodology comprises three main components: (1) generating datasets using a sequential sampling strategy, (2) modeling multi-physics spatio-temporal dynamics by designing a heterogeneous convolutional autoencoder and Recurrent Neural Network, and (3) training high-precision models with limited data and long-range time evolution via a dual-phase training strategy. A holistic evaluation using a 2D low-temperature plasma processing example demonstrates the methodology's superior computational efficiency, accuracy, and generalization capabilities. It predicts plasma dynamics approximately 105 5 times faster than traditional numerical solvers, with a consistent 2% relative error across different generalization tasks. Furthermore, the potential for transferability across various geometries is explored, and the model's transfer capability is demonstrated with two distinct geometric domain examples.

关键词： Data-driven surrogate modeling Process systems Plasma processing convolutional autoencoder Recurrent neural network Transfer learning

来源：评论

学校读者我要写书评

暂无评论

Estimation of the Global Equivalence Ratio of a Swirl Combustor from a Small Number of Absorption Spectra Using Machine Learning

引用

APPLIED SPECTROSCOPY 2024年第10期78卷 1078-1088页

作者： Bong, Cheolwoo Im, Seong-kyun Do, Hyungrok Bak, Moon Soo Sungkyunkwan Univ Sch Mech Engn Suwon South Korea Korea Univ Dept Mech Engn Seoul South Korea Seoul Natl Univ Dept Mech Engn Seoul South Korea Sungkyunkwan Univ Dept Smart Fabricat Technol Suwon South Korea

A new optical diagnostic method that predicts the global fuel-air equivalence ratio of a swirl combustor using absorption spectra from only three optical paths is proposed here. Under normal operation, the global equivalence ratio and total flow rate determine the temperature and concentration fields of the combustor, which subsequently determine the absorption spectra of any combustion species. Therefore, spectra, as the fingerprint for a produced combustion field, were employed to predict the global equivalence ratio, one of the key operational parameters, in this study. Specifically, absorption spectra of water vapor at wavenumbers around 7444.36, 7185.6, and 6805.6 cm-1 measured at three different downstream locations of the combustor were used to predict the global equivalence ratio. As it is difficult to find analytical relationships between the spectra and produced combustion fields, a predictive model was a data-driven acquisition. The absorption spectra as an input were first feature-extracted through stacked convolutional autoencoders and then a dense neural network was used for regression prediction between the feature scores and the global equivalence ratio. The model could predict the equivalence ratio with an absolute error of +/- 0.025 with a probability of 96%, and a gradient-weighted regression activation mapping analysis revealed that the model leverages not only the peak intensities but also the variations in the shape of absorption lines for its predictions. Graphical abstract This is a visual representation of the abstract.

关键词： Swirl combustor status monitoring tunable diode laser absorption spectroscopy TDLAS convolutional autoencoder CAE gradient-weighted regression activation mapping grad-RAM

来源：评论

学校读者我要写书评

暂无评论

1DCAE-TSSAMC: Two-Stage Multi-Dimensional Spatial Features Based Multi-View Deep Clustering for Time Series Data

引用

INTERNATIONAL JOURNAL OF UNCERTAINTY FUZZINESS AND KNOWLEDGE-BASED SYSTEMS 2024年第4期32卷 593-623页

作者： Chen, Jianglong Song, Weiwei Zuo, Xiaoqing Zhao, Kang Jin, Baoxuan Zhu, Daming Dai, Bolan Kunming Univ Sci & Technol Fac Land Resources Engn Kunming 650500 Yunnan Peoples R China Dept Nat Resources Yunnan Prov Kunming 650224 Yunnan Peoples R China China Shipbldg Corp Res Inst 722 Wuhan 430079 Hubei Peoples R China

At present, as a research hotspot for time series data (TSD), the deep clustering analysis of TSD has huge research value and practical significance. However, there still exist the following three problems: (1) For deep clustering based on joint optimization, the inevitably mutual interference existing between deep feature representation learning progress and clustering progress leads to difficult model training especially in the initial stage, the possible feature space distortion, inaccurate and weak feature representation;(2) Existing deep clustering methods are difficult to intuitively define the similarity of time series and rely heavily on complex feature extraction networks and clustering algorithms. (3) Multidimensional time series have the characteristics of high dimensions, complex relationships between dimensions, and variable data forms, thus generating a huge feature space. It is difficult for existing methods to select discriminative features, resulting in generally low accuracy of methods. Accordingly, to address the above three problems, we proposed a novel general two-stage multi-dimensional spatial features based multi-view deep clustering method 1DCAE-TSSAMC (One-dimensional deep convolutional auto-encoder based two-stage stepwise amplification multi-clustering). We conducted verification and analysis based on real-world important multi-scenario, and compared with many other benchmarks ranging from the most classic approaches such as K-means and Hierarchical to the state-of-the-art approaches based on deep learning such as Deep Temporal Clustering (DTC) and Temporal Clustering Network (TCN). Experimental results show that the new method outperforms the other benchmarks, and provides more accurate, richer, and more reliable analysis results, more importantly, with significant improvement in accuracy and spatial linear separability.

关键词： Multi-dimensional convolutional autoencoder clustering feature representation linear separability

来源：评论

学校读者我要写书评

暂无评论

Price prediction in China stock market: an integrated method based on time series clustering and image feature extraction (Aug, 10.1007/s11227-023-05562-z, 2023)

引用

JOURNAL OF SUPERCOMPUTING 2024年第7期80卷 10191-10191页

作者： Guan, Bowen Zhao, Chencheng Yuan, Xianghui Long, Jun Li, Xiang Xi An Jiao Tong Univ Sch Software Engn Xian Peoples R China Xi An Jiao Tong Univ Sch Econ & Finance Xian Peoples R China Xi An Jiao Tong Univ Jinhe Ctr Econ Res Xian Peoples R China

Stock time-series data has the characteristics of high dimensionality and nonlinearity, which brings great challenges to stock forecasting. Aiming at the impact of stock correlation and the prediction information contained in stock image features, we propose a long short-term memory model based on clustering and image feature extraction, named Kmeans-CAE-LSTM. Firstly, the Kmeans algorithm is used for stock clustering, where the most correlated stocks are found. Secondly, a convolutional autoencoder (CAE) is applied to extract stock price image features. Finally, the stock technical data and image features are respectively input into the double-layer long-term short-term memory network to predict the stock price of the next trading day. The empirical research results on 11 industries in China’s stock market show that the hybrid model has achieved the best prediction effect, which further proves the predictive ability of stock image data and can provide investors with new ideas for stock prediction and asset portfolio.

关键词： Financial time series LSTM neural network Hybrid deep learning model convolutional autoencoder Kmeans

来源：评论

学校读者我要写书评

暂无评论

Plant Disease Detection and Classification Using Hybrid Model Based on convolutional Auto Encoder and convolutional Neural Network

引用

Computers, Materials & Continua 2025年第6期83卷 5219-5234页

作者： Tajinder Kumar Sarbjit Kaur Purushottam Sharma Ankita Chhikara Xiaochun Cheng Sachin Lalar Vikram Verma Computer Science&Engineering Department Jai Parkash Mukand Lal Innovative Engineering&Technology InstituteRadaurYamunanagar135133India Department of Computer Science Government PG CollegeAmbala CanttAmbala134003India School of Computer Science&Engineering Galgotias UniversityGreater Noida203201India Department of Computer Science and Applications Kurukshetra UniversityKurukshetra136118India Computer Science Department Bay Campus Fabian WaySwansea UniversitySwanseaSA18ENUK Department of Engineering and Technology Gurugram UniversityGurugram122003India Department of CSE Panipat Institute of Engineering and TechnologyPanipat132103India

During its growth stage,the plant is exposed to various *** and early detection of crop diseases is amajor challenge in the horticulture *** infections can harmtotal crop yield and reduce farmers’income if not identified ***’s approved method involves a professional plant pathologist to diagnose the disease by visual inspection of the afflicted plant *** is an excellent use case for Community Assessment and Treatment Services(CATS)due to the lengthy manual disease diagnosis process and the accuracy of identification is directly proportional to the skills of *** alternative to conventional Machine Learning(ML)methods,which require manual identification of parameters for exact results,is to develop a prototype that can be classified without *** automatically diagnose tomato leaf disease,this research proposes a hybrid model using the convolutional Auto-Encoders(CAE)network and the CNN-based deep learning architecture of *** date,none of the modern systems described in this paper have a combined model based on DenseNet,CAE,and convolutionalNeuralNetwork(CNN)todiagnose the ailments of tomato leaves *** trained on a dataset obtained from the Plant Village *** dataset consisted of 9920 tomato leaves,and the model-tomodel accuracy ratio was 98.35%.Unlike other approaches discussed in this paper,this hybrid strategy requires fewer training ***,the training time to classify plant diseases with the trained algorithm,as well as the training time to automatically detect the ailments of tomato leaves,is significantly reduced.

关键词： Tomato leaf disease deep learning DenseNet-121 convolutional autoencoder convolutional neural network

来源：评论

学校读者我要写书评

暂无评论

Multi-stream pyramid collaborative network for spectral unmixing

引用

INTERNATIONAL JOURNAL OF REMOTE SENSING 2024年第8期45卷 2674-2701页

作者： Wang, Jie Ni, Mengying Wang, Zhixiang Yan, Yu Cheng, Xiang Xu, Jindong Yantai Univ Sch Comp & Control Engn Yantai Peoples R China Peking Univ Sch Informat Sci & Technol Beijing Peoples R China

convolutional autoencoder, which can well model the spatial correlation of the data, have been widely applied to spectral unmixing task and achieved desirable performance. However, the fixed geometric structure of convolution kernels makes it difficult to capture global context. To address this issue, strategies such as dilated convolution or transformer are often employed, but this may result in minor loss of local details. Therefore, we propose a collaborative unmixing network with a multi-scale pyramid structure to capture both global and local features simultaneously. To integrate features from different scales in the unmixing process, we employ a cross-stream fusion feature strategy, which not only promote collaborative representations but also capture long-range dependencies while preserving local details. Meanwhile, we also design the residual spectral attention mechanism to refine the features from different scales and facilitate their fine-grained fusion. In the proposed network, each convolutional stream undergoes effective collaborative training using a convolutional autoencoder structure. The collaborative strategies include cross-stream feature fusion mechanism and alternating training strategy with weight sharing for endmember information. Experiments over three real hyperspectral datasets indicate the effectiveness of our method compared to other unmixing techniques.

关键词： Spectral unmixing deep learning convolutional autoencoder multi-scale feature fusion collaborative strategies

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Learning Approach for Risky Driving Behavior Identification on Expressways in C-ITS Environments

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2025年第4期26卷 5146-5155页

作者： Kim, Dongmin Lee, Hwanpil Lee, Jooyoung Korea Adv Inst Sci & Technol Cho Chun Shik Grad Sch Mobil Daejeon 34051 South Korea Korea Expressway Corp Res Inst ICT Convergence Res Div Hwaseong 18489 Gyeonggi Do South Korea Hannam Univ Dept Ind & Management Engn Daejeon 34430 South Korea

Understanding the intricate relationship between driving behaviors, traffic crashes, and human factors is paramount in enhancing road safety. Human error, often stemming from risky driving behaviors, contributes significantly to traffic crashes. Identifying and mitigating these behaviors through advanced technologies and data analysis has become an important concern in the field of traffic safety management. This study introduces an unsupervised learning algorithm for detecting risky driving behaviors on expressways within Cooperative Intelligent Transport Systems (C-ITS) environments, employing deep clustering techniques to analyze individual driving patterns from Probe Vehicle Data (PVD). Utilizing data from 116 vehicles, including buses and heavy trucks, a convolutional Neural Network (CNN)-based autoencoder was employed to extract latent hierarchical features, facilitating the clustering of similar driving patterns. Elementary Driving Behaviors (EDBs) were identified for different vehicle types and driving statuses, serving as a foundation for detecting risky driving behaviors against the proposed criteria. The research revealed a clear positive correlation between detected risky driving behaviors and traffic crashes across the vehicle types. Furthermore, when comparing our model's criteria with traditional safety indexes, our proposed model demonstrated stronger correlations with traffic crashes, indicating its effectiveness in expressway driving environments. This research not only introduces a novel method for identifying risky driving behaviors but also underscores the importance of tailored traffic safety interventions in enhancing C-ITS environments.

关键词： Driver behavior Accidents Data mining Safety management Monitoring Feature extraction Unsupervised learning Probes Microscopy Intelligent transportation systems convolutional autoencoder cooperative intelligent transport systems elementary driving behavior expressways probe vehicle data risky driving behavior

来源：评论

学校读者我要写书评

暂无评论

A Deep autoencoder-Based Approach for Suspicious Action Recognition in Surveillance Videos

引用

ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING 2024年第3期49卷 3517-3532页

作者： Ahmed, Waqas Yousaf, Muhammad Haroon Univ Engn & Technol Dept Telecommun Engn Taxila 47050 Pakistan Univ Engn & Technol Dept Comp Engn Taxila 47050 Pakistan Natl Ctr Robot & Automat NCRA Swarm Robot Lab Taxila Pakistan

In the recent era of technological advancements, surveillance cameras are installed in crowded areas to ensure public protection. In the video surveillance context, contents belonging to suspicious actions are very less in course of the surveillance stream. Therefore, manual monitoring of suspicious actions may become very exhaustive, which effects reliability and speed during emergencies due to monitoring tiredness, so the importance of suspicious action detection is very clear. We first address the issue of detecting suspicious activities from the surveillance videos with our proposed CNN-based autoencoder. The features are extracted using a three-dimensional convolutional neural network (C3D) and fed to our proposed autoencoder framework, which detects the localization of activity based on high reconstruction loss. For normal video clips, we have seen low reconstruction loss and the converse is seen for video clips containing suspicious actions. Secondly, we extract these suspicious clips from the long surveillance videos and use them to classify various suspicious actions with the help of our proposed generative adversarial network (GAN). We evaluate the performance of our work with benchmark datasets, namely UT interaction, hybrid crime action (HCA), and UCF crime. The results show the effectiveness of our work and as achieved accuracies are 97.5%, 89.6%, and 47.34% on UT interaction, HCA and UCF crime dataset, respectively.

关键词： convolutional autoencoder convolutional neural network Deep learning GAN UCF-crime

来源：评论

学校读者我要写书评

暂无评论

A deep autoencoder based approach for the inverse design of an acoustic-absorber

引用

ENGINEERING WITH COMPUTERS 2024年第1期40卷 279-300页

作者： Mahesh, K. Ranjith, S. Kumar Mini, R. S. Coll Engn Trivandrum Govt Kerala Dept Mech Engn Thiruvananthapuram 695016 Kerala India Coll Engn Trivandrum Govt Kerala Dept Mech Engn Micro Nanofluid Res Lab Thiruvananthapuram 695016 Kerala India APJ Abdul Kalam Technol Univ Thiruvananthapuram 695016 Kerala India

This paper proposes an algorithm to perform the inverse design of a low-frequency acoustic absorber using a deep convolutional autoencoder network. A hybrid sound-absorber configuration based on Helmholtz resonators with inserted curvy neck and microperforated panel is suggested and its geometrical properties are inversely forecasted from the targeted signal. A mathematical model is put forwarded to evaluate the absorption characteristics of the introduced geometry by employing the effective medium theory and the electro-acoustic analogy. The large dataset required to train, validate and test the deep neural network is extracted through this analytical procedure. Initially, the proposed inverse technique is successfully applied on a standard Helmholtz resonator based absorber setup with great accuracy. This prediction approach is further extended to suit the inverse design of a hybrid sound absorber with complex geometrical attributes. The encoder maps the input acoustic absorption spectrum to geometrical features of the absorber, and the subsequent decoder recreates the absorption characteristics using convolutional layers. Once the training and testing of the neural network are over, the deep autoencoder inversely predicts the geometrical parameters. In comparison with earlier inverse models which employed deep neural networks, the accuracy of the current scheme is very high and no pre-design information on absorber geometry is required as well. Since the relevant learnable parameters involved are very low, the computational load is also very less for this autoencoder based method. Later, using the new inverse scheme, four representative absorber designs with specific acoustic functionality are deduced. Most importantly, these four compact absorber models produce quasi-perfect absorption in the frequency bands 200-315 Hz, 255-400 Hz, 300-530 Hz, and 350-650 Hz. Notably, the developed absorber versions have great potential in noise reduction applications owing to t

关键词： Low frequency sound absorption Helmholtz resonator Analytical modelling Electro-acoustic analogy convolutional autoencoder Inverse design

来源：评论

学校读者我要写书评

暂无评论

A fault diagnosis method for rolling bearings in open-set domain adaptation with adversarial learning

引用

SCIENTIFIC REPORTS 2025年第1期15卷 1-19页

作者： Lei, Tongfei Pan, Feng Hu, Jiabei He, Xu Li, Bing Xijing Univ Sch Mech Engn Xian 710123 Peoples R China Xian Aeronaut Comp Tech Res Inst Xian 710068 Peoples R China Chinese Acad Sci Xian Inst Opt & Precis Mech Xian 710119 Peoples R China Shaanxi Inst Elect Technol Co Ltd Xian 710000 Peoples R China

The closed-set assumption often fails in practical industrial applications, especially considering diverse working conditions where the data distribution may differ significantly. In light of this, a domain adaptation method with adversarial learning is designed for open-set fault diagnosis. Firstly, convolutional autoencoder is developed to distill the fault features;Secondly, an unknown boundary by weighting the similarity between known and unknown classes is established, to ensure shared class alignment between domains while classifying known classes across domains and identifying unknown fault samples. Finally, the diagnostic performance is evaluated using three sets of rolling bearing datasets. The proposed method achieved average diagnostic F1-scores of 96.60%, 96.56%, and 96.62% on these datasets, respectively. The results demonstrate that the method effectively rejects unknown fault data in the target domain while aligning known classes, validating its fault diagnosis capability under the open-world assumption.

关键词： Open-set fault diagnosis Adversarial learning convolutional autoencoder Known classes Unknown classes

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

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

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：