检索结果-内蒙古大学图书馆

An Adaptive Continual Learning Method for Nonstationary Industrial Time Series Prediction

IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS 2025年第2期21卷 1160-1169页

作者： Wu, Mengqing Zhou, Xiaofeng Li, Shuai Shi, Haibo Chinese Acad Sci Key Lab Networked Control Syst Shenyang 110016 Peoples R China Chinese Acad Sci Shenyang Inst Automat Shenyang 110016 Peoples R China Chinese Acad Sci Inst Robot & Intelligent Mfg Shenyang 110169 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China

Deep learning models have gained significant attention and application in recent years to improve the accuracy and efficiency of industrial time series prediction. However, the dynamic changes in industrial processes present a key challenge for data-driven models. Specifically, the performance of deployed models deteriorates over time and fails to adapt to new operating conditions. Currently, two common update methods exist: Retraining the model using historical and new operating data, which incurs high computation and storage costs, or incrementally fine-tuning the model solely using new data, which leads to catastrophic forgetting of learned patterns. To address these issues, this article proposes an adaptive continual learning method for nonstationary industrial time series prediction. Our approach tackles the problems by hint-based network parameter learning to retain the dark knowledge from previous tasks and avoid catastrophic forgetting of accumulated knowledge. In addition, we design a soft buffer to aid memory and learning of key patterns under the current operating condition. Lastly, a time-sensitive activation function is proposed to endow the neural network with time-evolving properties, thereby enhancing the model's generalization ability. Compared with other update methods and different continual learning methods, the superiority of our method is validated on solar power generation data and real data of grinding and grading process.

关键词： Continuing education Adaptation models Predictive models data models Training Time series analysis Computational modeling Production Optimization Service robots Adaptive continual learning catastrophic forgetting generalization nonstationary industrial process time series prediction

来源：评论

学校读者我要写书评

暂无评论

Novel abstractions and experimental validation for digital twin microgrid design: Lab scale studies and large scale proposals

引用

APPLIED ENERGY 2025年 377卷

作者： Sifat, Md. Mhamud Hussen Choudhury, Safwat Mukarrama Das, Sajal K. Pota, Hemanshu Yang, Fuwen Rajshahi Univ Engn & Technol Rajshahi 6204 Bangladesh Univ New South Wales Canberra ACT Australia Griffith Univ Gold Coast Qld Australia

Managing the current high penetration of renewable energy sources globally poses a significant challenge due to the distributed and diverse nature of generation components. To maximize real-time power generation, these resources need to perform optimally. Digital twin technology offers a comprehensive framework for managerial support by replicating grid features in a digital environment. This research creates a digital twin of the microgrid to optimize power generation, focusing on computational efficiency and self-healing control. The framework is tested in a laboratory microgrid, with modeling performed using a polynomial regression algorithm. Optimization is achieved through a gradient descent algorithm, and the self-healing model is implemented using a logistic regression algorithm. Real-time data extracted from the microgrid drives this process. The results can be utilized for predictive analysis before deploying a microgrid or to optimize generation in existing systems using the digital twin model. Even though the research focuses on a single microgrid unit, it introduces a framework proposal for extensively distributed microgrids integrating multiple renewable energy sources.

关键词： Digital twin microgrid control Energy management Self-healing Microgrid optimization

来源：评论

学校读者我要写书评

暂无评论

2024 International conference on Green Energy and Mechatronics Engineering, ICGEME 2024

2024 International Conference on Green Energy and Mechatroni...

引用

2024 International conference on Green Energy and Mechatronics Engineering, ICGEME 2024

The proceedings contain 24 papers. The topics discussed include: ultra-thick dry-process cathode design through formula optimization;through the anodic oxidation of sodium sulfite aqueous solution to achieve energy saving cathodic hydrogen production;an improved control method of district heating system based on waste heat utilization in data center;a new slope gravity energy storage system with multi parallel and continuous circulation tracks;research and application of the cooling system for water-based drilling fluid in hot dry rock drilling;small signal modeling and frequency oscillation analysis of multi-VSGS in microgrids;extraction of key nuclides in carbonate environment using methyltrioctylammonium carbonate;and production prediction of pumping wells based on multi-mode transfer learning.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Defect Detection in Industrial Soldering processes Using Machine Learning: A Critical Literature Review

引用

IEEE ACCESS 2025年 13卷 41533-41558页

作者： Eisentraut, Luca Hosch, Johanna Roytenberg, Maksym Benecke, Amelie Penava, Pascal Buettner, Ricardo Helmut Schmidt Univ Univ Fed Armed Forces Hamburg Chair Hybrid Intelligence D-22043 Hamburg Germany

As electrical devices take on more life-critical roles, such as in autonomous driving, ensuring the quality of solder joints during production becomes increasingly important. Recently, there has been a growing interest in using machine learning techniques for this purpose. However, current research lacks a comprehensive overview that categorizes and analyzes relevant studies based on their specific intervention points within the production process. This literature review aims to examine and evaluate research coverage along three dimensions: intervention points in the process, non-destructive testing methods, and machine learning techniques employed. For this review, 112 conference papers and journal articles published since 2010 were selected from three databases using the PRISMA methodology. These publications were classified into the three dimensions previously mentioned, summarized, and analyzed. Furthermore, the literature core is critically evaluated to identify research gaps and limitations. The analysis shows that most studies focus on solder joint control, with few addressing intervention points in solder paste and component placement. Visual imaging and neural networks are the dominant techniques for non-destructive testing and machine learning, respectively. Despite a variety of literature that uses high-performance neural networks, meeting industrial detection standards often requires tolerating high false alarm rates. The findings contribute to structuring existing research and identifying research needs, particularly in validating these systems and integrating data from various testing methods and intervention points.

关键词： Systematic literature review Soldering Defect detection Testing Quality control Visualization Production process control Inspection databases Critical literature review defect detection machine learning solder joint

来源：评论

学校读者我要写书评

暂无评论

Development of a flood impact assessment framework integrating crowdsourced data and geospatial information for data sparse urban regions

引用

INTERNATIONAL JOURNAL OF DISASTER RISK REDUCTION 2025年 116卷

作者： Ponukumati, Padmini Regonda, Satish Kumar Indian Inst Technol Hyderabad Dept Civil Engn Hyderabad India Indian Inst Technol Hyderabad Dept Climate Change Hyderabad India

High spatial heterogeneity of urban floods poses challenges in its modeling and assessment, and a flood database is a basic requirement but it is lacking for many cities. The proposed Twitter-based framework addresses the issues via developing a finer resolution flood database and a product. The framework has multiple components including data quality control, validation of flood database via newspapers and flood impact assessment. Three flood events differing in rainfall characteristics are selected to showcase the utility of the proposed framework for the city of Hyderabad, India. analysis of tweets highlighted the resourcefulness of video tweets and wide coverage of the study area in terms of flood reporting. Tweets exhibited an association with tweet time and rainfall aspects. Further, tweets based flood locations are found to be in agreement with newspaper based flooding instances. A novel flood impact score (FIS) is developed for each flood location using analytical hierarchy process based weights for five variables (Twitter based attributes, rainfall, elevation), and the use of FIS is demonstrated in identifying flood impact areas. These kinds of databases and products, with a scope to improve further, serve as a potential tool to cater flood preparedness and management, thereby making cities flood resilient.

关键词： Urban floods Crowd sourced data Social media platform Flood database Analytical hierarchy process Hyderabad

来源：评论

学校读者我要写书评

暂无评论

data-driven estimation of impedance of inverter-based resources for efficient stability evaluation

引用

ELECTRIC POWER SYSTEMS RESEARCH 2025年 244卷

作者： Mohammed, Nabil Zhou, Weihua Ramasubramanian, Deepak Bahrani, Behrooz Dutta, Sudipta Bello, Mobolaji Monash Univ Clayton Vic Australia Elect Power Res Inst Golden CO USA

Inverter-based resources (IBRs) are key enabling technologies for integrating renewable energy sources and providing ancillary services in modern power systems. However, their dynamic behavior, defined by output impedance models, can pose a threat to power system stability. The primary challenge is that impedance models, typically derived at specific operating points, exhibit limited accuracy under varying conditions. Additionally, the lack of detailed vendor information on commercial IBR control structures complicates the accurate derivation of these models. To address these issues, this paper first investigates the effects of grid parameters and variations in IBR operating points on IBR's impedance model. Afterwards, a data-driven algorithm using Gaussian process regression (GPR) is then proposed to predict impedance models in the dq reference frame, achieving accurate results with a minimal dataset, thus reducing the cost and complexity of data collection for stability evaluation. The proposed approach is validated through case studies that compare predicted impedance models with analytical solutions for various IBR configurations and grid scenarios, including both grid-following and grid-forming inverters. Its superiority over artificial neural network (ANN)- based approaches is demonstrated using the same training dataset. The predicted impedance model is employed to evaluate IBR stability in the frequency domain, with findings validated through time-domain simulations using an electromagnetic transient (EMT) model when connected to grids of varying strengths. A promising application of the proposed GPR-based impedance modeling is its integration into IBR-based power system stability analysis and simulation tools, facilitating the study of emerging low-frequency oscillation phenomena.

关键词： Gaussian process regression Grid-connected inverters Grid impedance Impedance measurement Nyquist Operating point Stability System identification

来源：评论

学校读者我要写书评

暂无评论

ConcVAE: Conceptual Representation Learning

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2025年第4期36卷 7529-7541页

作者： Togo, Ren Nakagawa, Nao Ogawa, Takahiro Haseyama, Miki Hokkaido Univ Fac Informat Sci & Technol Sapporo 0600814 Japan Hokkaido Univ Grad Sch Informat Sci & Technol Sapporo 0600814 Japan

Disentangled representation learning aims at obtaining an independent latent representation without supervisory signals. However, the independence of a representation does not guarantee interpretability to match human intuition in the unsupervised settings. In this article, we introduce conceptual representation learning, an unsupervised strategy to learn a representation and its concepts. An antonym pair forms a concept, which determines the semantically meaningful axes in the latent space. Since the connection between signifying words and signified notions is arbitrary in natural languages, the verbalization of data features makes the representation make sense to humans. We thus construct Conceptual VAE (ConcVAE), a variational autoencoder (VAE)-based generative model with an explicit process in which the semantic representation of data is generated via trainable concepts. In visual data, ConcVAE utilizes natural language arbitrariness as an inductive bias of unsupervised learning by using a vision-language pretraining, which can tell an unsupervised model what makes sense to humans. Qualitative and quantitative evaluations show that the conceptual inductive bias in ConcVAE effectively disentangles the latent representation in a sense-making manner without supervision. Code is available at https://***/ganmodokix/concvae.

关键词： Representation learning Task analysis Natural languages Feature extraction Information science data models Computational modeling Deep learning disentanglement interpretable machine learning representation learning vision and language

来源：评论

学校读者我要写书评

暂无评论

Integrating Hybrid Physics-data Approaches for Enhanced Cutting Force modeling in Digital Twins of Helical End Mills 20

Integrating Hybrid Physics-Data Approaches for Enhanced Cutt...

引用

20th CIRP conference on modeling of Machining Operations in Mons, CIRP CMMO 2025

作者： Jing, Yuan Gong, Guanchen Hänel, Albrecht Ihlenfeldt, Steffen Fraunhofer Insitute for Machine Tools and Forming Technology IWU Chemnitz Germany Chair of Machine Tools Development and Adaptive Controls Institute Fo Mechatronic Engineering Technische Universität Dresden Dresden Germany

Industry 4.0 has significantly improved data efficiency by leveraging key technologies such as the Internet of Things and Machine Learning. Among these key technologies, digital twins stand out by offering a promising approach to intelligently utilize this data. In the virtual representation of a physical asset, data reflects the conditions of the physical entity, while models simulate and predict its behavior. In this paper, a hybrid cutting force model is proposed for digital twins of helical end mills, focusing on cutting force analysis during the utilization phase of the machining process. This model combines a fairly mature physical process modelling approach with a data-driven method, specifically a neural network trained on real process data, to address the limitations inherent in their respective applications. The physics-based model provides meaningful constraints on the neural network's training, ensuring reliable cutting force prediction, particularly in scenarios with limited process data availability. The cutter's profile, generated by the geometric model, and the cutter-workpiece engagement maps, derived from the virtual machining model, together serve as inputs for the hybrid cutting force model. © 2025 The Author(s).

关键词： Machining centers

来源：评论

学校读者我要写书评

暂无评论

Enhancing trusted computing operation efficiency based on parallel file verification techniques 4

Enhancing trusted computing operation efficiency based on pa...

引用

4th International conference on Network Communication and Information Security, ICNCIS 2024

作者： Zhang, Zhilong Pan, Fengyan Hou, Haibo Li, Yuan Shanghai Jiaotong University Shanghai China Beijing Chenguang Rongxin Technology Co. Beijing China

ISBN: (数字)9781510688261

ISBN: (纸本)9781510688254

Trusted Computing technology represents a significant element of cyber security systems, serving to guarantee the integrity and accessibility of data and systems. The incorporation of Trusted Computing introduces a series of security detection and verification mechanisms to the host system, consequently impacting the performance of host system startup and application operation. In this paper, following an analysis of the verification mechanisms of Trusted Computing, a parallel verification mechanism is proposed. The objective is to facilitate the parallel operation of the host system program and the TPM-based file verification process, thereby enhancing the operational efficiency of the trusted computing system. Through the configurable and micro-intervention design scheme, flexible expansion and configuration within the existing Trusted Computing technology framework can be achieved. © 2025 SPIE.

关键词： Computing systems Design Computer security process control Systems modeling Mathematical optimization Operating systems

来源：评论

学校读者我要写书评

暂无评论

Quantifying, predicting, and mitigating nitrous oxide emissions in a full-scale partial nitritation/anammox reactor treating reject water

引用

WATER RESEARCH 2025年 278卷 123200页

作者： Flores-Alsina, Xavier Vangsgaard, Anna Katrine Uri-Carreno, Nerea Nielsen, Per H. Gernaey, Krist V. Tech Univ Denmark PROSYS Res Ctr Dept Chem & Biochem Engn Soltofts Plads Bldg 228A DK-2800 Lyngby Denmark EnviDan A S Fuglebaekvej 1A DK-2770 Kastrup Denmark Vandcenter Syd A S Vandvaerksvej 7 DK-5000 Odense Denmark N118 Consulting Skt Jorgens Gade 44K DK-5000 ODENSE Denmark

In this paper, a set of mathematical tools are developed and assembled to quantify, predict and virtually assess N2O emission mitigation strategies in partial nitritation (PN) / anammox (ANX) granular based reactors. The proposed approach is constructed upon a set of data pre-treatment methods, process simulation models, control tools (and algorithms) and key performance indicators to analyze, reproduce, and forecast the behavior of multiple operational variables within aerobic granular sludge systems. All these elements are tested on two full-scale data sets (#D1, #D2) collected over a period of four months (Sept-Dec 2023). Results show that data pretreatment is essential for noise reduction, filling data gaps, and ensuring smooth process simulations. The model accurately predicts (normalized RMSE< 1) multiple N oxidation states (NHx, NO2-, NO3-, N2O) and dissolved oxygen (DO), demonstrating its capability to describe bacterial behavior within the studied system. Special emphasis is placed on weak acid-base chemistry where pH is reliably reproduced, and it can be used for control purposes. Both biological and physico-chemical aspects are predicted at different time scales (months, days, minutes). While nitritation mainly occurred in the bulk, biofilm distribution showed inactive inner granule parts and increasing biomass (mostly ANX) towards the surface, with distinct organic concentrations. Gradients for multiple soluble compounds could also be reflected. Nitrifier denitrification (ND) is identified as the main N2O production pathway. The model revealed that the system was suffering from low ANX activity leading to NO2- accumulation. This in combination with low DO levels resulted in an unusually high emission factor (EF). The validation data set also yielded satisfactory results (normalized RMSE< 1). The scenario analysis revealed that modification of the operational parameters could improve the ANX activity and lead to N2O emission rates that are in line with wh

关键词： Climate change mitigation data science Digital twins GHG emissions modeling Net zero

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：