检索结果-内蒙古大学图书馆

IEEE-Power-and-Energy-Society General Meeting (PESGM)

作者： Islam, Md Maidul Faruque, Md Omar Butterfield, Joshua Singh, Gaurav Cooke, Thomas A. FAMU FSU Coll Engn Elect & Comp Engn Tallahassee FL 32310 USA Elect Power Res Inst Palo Alto CA USA

ISBN: (纸本)9781665464413

Power quality (PQ) events are recorded by PQ meters whenever anomalous events are detected on the power grid. Using neural networks with machine learning can aid in accurately classifying the recorded waveforms and help power system engineers diagnose and rectify the root causes of problems. However, many of the waveforms captured during a disturbance in the power system need to be labeled for supervised learning, leaving a large number of data recordings for engineers to process manually or go unseen. This paper presents an autoencoder and K -means clustering-based unsupervised technique that can be used to cluster PQ events into categories like sag, interruption, transients, normal, and harmonic distortion to enable filtering of anomalous waveforms from recurring or normal waveforms. The method is demonstrated using three-phase, field -obtained voltage waveforms recorded in a distribution grid. First, a convolutional autoencoder compresses the input signals into a set of lower feature dimensions which, after further processing, is passed to the K -means algorithm to identify data clusters. Using a small, labeled dataset, numerical labels are then assigned to events based on a cosine similarity analysis. Finally, the study analyzes the clusters using the t-distributed stochastic neighbor embedding (t-SNE) visualization tool, demonstrating that the technique can help investigate a large number of captured events in a quick manner.

关键词： convolutional autoencoder data visualization power quality unsupervised clustering waveform clustering

来源：评论

学校读者我要写书评

暂无评论

Dynamic Clustering for Wafer Map Patterns Using Self-Supervised Learning on convolutional autoencoders

引用

IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING 2021年第4期34卷 444-454页

作者： Kim, Donghwa Kang, Pilsung Korea Univ Sch Ind & Management Engn Seoul 02841 South Korea

Defect pattern analysis in wafer bin maps (WBM) plays a significant role in the semiconductor manufacturing process because it helps identify problematic steps or equipment so that process engineers can take appropriate actions to improve the overall yield. Clustering algorithms have been widely used to detect different defect patterns. However, most clustering algorithms, such as K-means clustering and self-organizing map, are required to determine the number of clusters in advance. To resolve this issue, we propose a self-supervised learning-based dynamic WBM clustering method. The proposed model first uses pseudo-labeled data, of which, the labels are dynamically determined by the Dirichlet process mixture model (DPMM). Thereafter, it is fine-tuned using pseudo-labels in a self-supervised manner. Experimental results based on the WM-811K dataset indicate that the proposed model not only outperforms the benchmark models but also demonstrates robustness to hyperparameters. In addition, the defect patterns identified by our model are more accurately and distinctively localized than those identified by the benchmark models.

关键词： Feature extraction Semiconductor device modeling Image reconstruction Clustering algorithms Visualization Data models Clustering methods convolutional autoencoder deep clustering Dirichlet process pseudo-labels self-supervised learning wafer maps

来源：评论

学校读者我要写书评

暂无评论

A Convolution Neural Network and Genetic Algorithm-Based Abnormity Detection Model for Cyber Attacks

引用

JOURNAL OF CIRCUITS SYSTEMS AND COMPUTERS 2025年第4期34卷

作者： Wang, Fang Yu, Kan Yang, Lei Wenhua Coll Fac Informat Sci & Technol Wuhan 430074 Hubei Peoples R China

With the rapid development of the Internet and network technology, network security has become increasingly prominent. Regarding the important issue of network attack detection, traditional methods often find it difficult to effectively capture and identify new types of network attack behaviors, while models based on deep learning and evolutionary algorithms are considered to better adapt to complex and ever-changing network attack environments. This paper aims to explore how to combine convolutional autoencoder (CAE) and genetic algorithm (GA) to construct an efficient network attack detection model and improve the ability of network security defense. First, the convolutional autoencoder is used to effectively learn the feature representation of network data, and combined with hierarchical attention mechanism, appropriate weights are assigned to classification tasks under different features and feature fusion is performed. Using GA to adaptively optimize random forest and improve its performance and robustness in network attack detection. Ultimately, achieving better network security protection results and effectively preventing and combating various network security threats. And conduct experimental verification on multiple datasets and compare with other benchmark methods. The results show that the model has achieved significant improvement in the detection of text network attacks, and can more effectively classify various types of attacks, bringing new technological breakthroughs and application prospects for network attack detection. It will also provide new ideas and methods for the further development of network security, effectively ensuring the safe and stable operation of network systems.

关键词： Cyber attack detection convolutional autoencoder hierarchical attention mechanism genetic algorithm

来源：评论

学校读者我要写书评

暂无评论

Forward-backstepping design of phononic crystals with anticipated band gap by data-driven method

引用

MECHANICAL SYSTEMS AND SIGNAL PROCESSING 2025年 224卷

作者： Li, Yingli Yin, Guohui Yan, Gengwang Yao, Song Cent South Univ Sch Traff & Transportat Engn Key Lab Traff Safety Track Minist Educ Changsha 410075 Peoples R China Cent South Univ Joint Int Res Lab Key Technol Rail Traff Safety Changsha Peoples R China Cent South Univ Natl & Local Joint Engn Res Ctr Safety Technol Rai Changsha Peoples R China

Phononic crystals are novel artificial materials characterized by the periodic arrangement of elastic materials within another elastic medium, which imparts exceptional features to the material, including the presence of band gaps. In order to obtain the phononic crystal structure possessing the desired features, standard design approaches typically involve iterative searches within an extensive design space which necessitates significant processing resources and expenses. This study introduces a novel approach, combining a convolutional autoencoder with deep neural network, to extract topological features of phononic crystal samples and conduct forward prediction of band gap distributions. Subsequently, the trained forward prediction model is employed as a substitute for finite element simulations to expedite the computation of the genetic algorithm's fitness. For the first time, this study introduced a chromosome coding expansion strategy and a fitness gradient strategy which can effectively avoid the issue of stagnation occurring during the initial population evolution. In the on-demand design, the band gap average loss ratio amounts to 5.9%, while the average excess proportion stands at 8.92%. The computation time for this method is 7.5 times shorter than that of method using a genetic algorithm with a finite element model, based on the six cases prediction. The band gap frequency of the optimum structure with proposed method is 2.7 times lower and 73 times broader compared to the classical phononic crystal with square-shaped scatter with almost twice volume fraction, which less degrades the strength. This interactive approach offers a solution to the issue of the nonunique response-to-design mapping problem. This method is not solely restricted to ondemand design, but also has the capability to optimize the band gap of phononic crystals with specific configurations under constrained conditions. This advancement sets the stage for the development of a forward-ba

关键词： Phononic crystals Band gap convolutional autoencoder Deep learning Genetic algorithm

来源：评论

学校读者我要写书评

暂无评论

Multiresolution convolutional autoencoders

引用

JOURNAL OF COMPUTATIONAL PHYSICS 2023年 474卷

作者： Liu, Yuying Ponce, Colin Brunton, Steven L. Kutz, J. Nathan Univ Washington Dept Appl Math Seattle WA 98105 USA Lawrence Livermore Natl Lab Livermore CA 94550 USA Univ Washington Dept Mech Engn Seattle WA 98105 USA

We propose a multi-resolution convolutional autoencoder (MrCAE) architecture that integrates and leverages three highly successful mathematical architectures: (i) multigrid methods, (ii) convolutional autoencoders and (iii) transfer learning. The method provides an adaptive, hierarchical architecture that capitalizes on a progressive training approach for multiscale spatio-temporal data. This framework allows for inputs across multiple scales: starting from a compact (small number of weights) network architecture and low-resolution data, our network progressively deepens and widens itself in a principled manner to encode new information in the higher resolution data based on its current performance of reconstruction. Basic transfer learning techniques are applied to ensure information learned from previous training steps can be rapidly transferred to the larger network. As a result, the network can dynamically capture different scaled features at different depths of the network. The performance gains of this adaptive multiscale architecture are illustrated through a sequence of numerical experiments on synthetic examples and real -world spatial-temporal data.(c) 2022 Elsevier Inc. All rights reserved.

关键词： convolutional autoencoder Multiresolution analysis Multigrid Transfer learning Model scaling Multi-scale dynamics

来源：评论

学校读者我要写书评

暂无评论

Vibration-Based Wear Condition Estimation of Journal Bearings Using convolutional autoencoders

引用

SENSORS 2023年第22期23卷 9212页

作者： Ates, Cihan Hoefchen, Tobias Witt, Mario Koch, Rainer Bauer, Hans-Joerg Karlsruhe Inst Technol KIT Inst Thermal Turbomachinery D-76137 Karlsruhe Germany KS Gleitlager GmbH Business Unit Bearings Rheinmetall AG D-68789 Sankt Leon Rot Germany

Predictive maintenance is considered a proactive approach that capitalizes on advanced sensing technologies and data analytics to anticipate potential equipment malfunctions, enabling cost savings and improved operational efficiency. For journal bearings, predictive maintenance assumes critical significance due to the inherent complexity and vital role of these components in mechanical systems. The primary objective of this study is to develop a data-driven methodology for indirectly determining the wear condition by leveraging experimentally collected vibration data. To accomplish this goal, a novel experimental procedure was devised to expedite wear formation on journal bearings. Seventeen bearings were tested and the collected sensor data were employed to evaluate the predictive capabilities of various sensors and mounting configurations. The effects of different downsampling methods and sampling rates on the sensor data were also explored within the framework of feature engineering. The downsampled sensor data were further processed using convolutional autoencoders (CAEs) to extract a latent state vector, which was found to exhibit a strong correlation with the wear state of the bearing. Remarkably, the CAE, trained on unlabeled measurements, demonstrated an impressive performance in wear estimation, achieving an average Pearson coefficient of 91% in four different experimental configurations. In essence, the proposed methodology facilitated an accurate estimation of the wear of the journal bearings, even when working with a limited amount of labeled data.

关键词： vibration sensors journal bearings state space modeling state trajectory machine learning convolution convolutional autoencoder unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

Fast prediction of three-dimensional indoor flow fields by a reduced dimensional deep-learning approach

引用

BUILDING AND ENVIRONMENT 2025年 270卷

作者： Gao, Hu Zhuang, Lei Li, Chenxi Qian, Weixin Dong, Jiankai Liu, Lin Liu, Jing Harbin Inst Technol Sch Architecture & Design Harbin 150090 Peoples R China Minist Ind & Informat Technol Key Lab Cold Reg Urban & Rural Human Settlement En Harbin 150090 Peoples R China Guangdong Univ Technol Sch Civil & Transportat Engn Guangzhou 510006 Peoples R China

The prediction of flow fields is crucial in both scientific research and engineering, yet traditional Computational Fluid Dynamics (CFD) calculations demand substantial computational resources and time. Recently, deep learning has gained attention in fluid simulation for its suitability in nonlinear problems, but its application in practical three-dimensional building simulations remains limited. This study proposes a deep learning framework based on a non-intrusive reduced-order model for rapid predictions of 3D multi-physical fields. The framework involves three key steps: (1) using a convolutional autoencoder to extract low-dimensional features from similar flow fields;(2) incorporating geometric parameters and boundary conditions of the target flow field, computing a Hadamard product with extracted features;and (3) reconstructing the predicted flow field through an up- sampling network. Two model scenarios (case-3v and case-6v) with varying boundary conditions were validated using actual residential environments. Comparisons with CFD simulations and on-site measurements showed that the model could predict wind speed, temperature, and other parameters in under 5 seconds, with accuracy comparable to zero-equation models. This approach offers a rapid and efficient solution for indoor flow prediction in ventilation research and early engineering design.

关键词： Deep learning Three-dimensional flow field Fast prediction Non-intrusive reduced-order model convolutional autoencoder

来源：评论

学校读者我要写书评

暂无评论

An audio-based framework for anomaly detection in large-scale structural testing

引用

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 2025年 142卷

作者： Munko, Marek J. Cuthill, Fergus Camacho, Miguel A. Valdivia Bradaigh, Conchur M. o Dubon, Sergio Lopez Univ Edinburgh Sch Engn Sanderson Bldg Robert Stevenson Rd Kings Bldg Edinburgh EH9 3FB Scotland Univ Sheffield Fac Engn Western Bank Sheffield S10 2TN England

FastBlade is a research facility that tests large-scale composite and metal structures. To maximise its throughput by uninterrupted running of experiments, unmanned operation of the site is desired. One of its key enablers is anomaly detection, where microphones are used as a non-specific, affordable, and well-established sensing method. The dataset collected during the operation of the system consists of both normal and anomalous samples, which we need to classify. The problems associated with the dataset involve significant intraclass variability of the normal operation samples, as well as the scarcity of anomalous data, increasing the complexity of the classification problem. In this work, we evaluate the performance of several tools for time-frequency signal analysis, which are used to extract features from the original high-dimensional signal. We choose to apply the wavelet scattering transform (WST) due to its remarkable performance. Based on the findings from the literature review, we first rely on the reconstruction error of the processed WST images to detect anomalous samples. However, due to the nature of the dataset, both the convolutional autoencoder (CAE) and the principal component analysis (PCA) transform turn out to be unsuccessful. We then investigate the hidden layers of the CAE in search of features that can be used to separate normal and anomalous samples. Having identified the most suitable candidates, we discover that applying the normalised cross-correlation (NCC) to measure the similarity of the generic features generated and our dataset results in satisfactory separation. We train a number of classifiers and test the method on unseen data. The model's accuracy is 99.58%, with a recall of 100% and 92% on normal and anomalous operation samples, respectively. The model's accuracy and low latency prove the WST's suitability for robust, real-time detection of different anomaly types. Therefore, the method can be deployed in systems with limited i

关键词： Anomaly detection Fatigue testing Artificial neural network convolutional autoencoder Wavelet scattering

来源：评论

学校读者我要写书评

暂无评论

Heuristic-aided fusion serial cascaded deep network for handwritten character recognition from handwritten images using optimization strategy

引用

APPLIED SOFT COMPUTING 2025年 174卷

作者： Banavatu, Triveni Parthasarathy, Govindaswamy Reva Univ Sch Comp & Informat Technol Bengaluru 560064 Karnataka India

Handwritten Character Recognition (HCR) identifies and interprets handwritten text, converting it into machinereadable characters. However, it faces challenges with South Indian languages due to their complex scripts and intricate characters. So, in this work, a novel HCR model was introduced for automatic recognition of handwritten characters in South Indian languages. At first, required Telugu and Kannada handwritten images are collected from standard sources, and given into the segmentation phase using Active Contour. Then, the textural pattern generation is performed using the Adaptive Local Weber Pattern (ALWP). The weights in the ALWP are optimally tuned using the Improved Snow Leopard Optimization Algorithm (ISLOA). Finally, the ALWP generates a textural pattern, and then the generated textural pattern is given to the Hybrid Serial Cascaded Deep Network (HSCDNet) for the final handwritten character recognition process. Here, the convolutional autoencoder (CAE) and Deep Belief Network (DBN) are serially connected to the network. Finally, the implemented HCR model's performance is evaluated by comparing it with various traditional approaches. The developed model attained the accuracy of dataset 1 is 93.45%, dataset 2 is 93.11 %, and dataset 3 is 94.13 %. Thus, it proved that the developed model can effectively and easily recognize the curves in the Telugu and Kannada handwritten scripts, making it suitable for a wide range of applications.

关键词： Handwritten character recognition South Indian languages Adaptive local weber pattern Improved snow leopard optimization algorithm Hybrid serial cascaded deep network convolutional autoencoder Deep belief network

来源：评论

学校读者我要写书评

暂无评论

Enhanced background reconstruction of X-ray angiography frames through a guided inpainting method built on an end-to-end deep learning approach

引用

BIOMEDICAL SIGNAL PROCESSING AND CONTROL 2025年 103卷

作者： Akhoondi, Fateme Almasganj, Farshad Amirkabir Univ Technol Biomed Engn Fac Tehran Iran

Achieving clear images during X-ray angiography is sometimes a challenge for cardiologists due to unwanted shadows of various fixed and moving organs inside the chest. To face this problem, many traditional works have been reported. The major drawback of these methods is that the corresponding algorithms should process all the frames of an input sequence, which requires a lot of processing time. In this paper, to face this problem, an endto-end neural network model is proposed to reconstruct the background of every selected frame from an X-ray angiography sequence, to be finally subtracted from the original frame. The proposed approach benefits from the auxiliary information available in another frame of the same sequence, recorded before injecting the contrast agent. The main block of the proposed method is a convolutional autoencoder (CAE) specially optimized to reconstruct the angiography images. In the middle layer of the CAE, the feature codes of the input frame are generated. The guided inpainting approach is provided by replacing parts of this generated codes with the codes of another angiography frame, recorded before injecting the contrast agent. Moreover, an accessory refinement network is especially developed, to perform better compatibility between the replaced and the remained codes. It creates a refined version of its input concatenated codes. The proposed method, is compared with some of the known state of the art methods, like the PCP-RPCA, FPCP-RPCA, MAMR and OSTD, using the SSIM and Precision criteria, and showed its superiority over the compared approaches by obtaining SSIM = 0.97 and Precision = 0.98.

关键词： X-ray angiography Neural network model convolutional autoencoder Guided inpainting

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：