Advanced Engineering Informatics最新文献_第8页

Crafting user-centric prompts for UI generations based on Kansei engineering and knowledge graph

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-05 DOI: 10.1016/j.aei.2025.103217

Xuejing Feng , Huifang Du , Jun Ma , Haofen Wang , Lijuan Zhou , Meng Wang

{"title":"Crafting user-centric prompts for UI generations based on Kansei engineering and knowledge graph","authors":"Xuejing Feng , Huifang Du , Jun Ma , Haofen Wang , Lijuan Zhou , Meng Wang","doi":"10.1016/j.aei.2025.103217","DOIUrl":"10.1016/j.aei.2025.103217","url":null,"abstract":"<div><div>Text-to-image (T2I) models are emerging as a powerful tool for designers to create user interface (UI) prototypes from natural language inputs (i.e., prompts). However, the discrepancy between designer inputs and model-preferred prompts makes it challenging for designers to consistently deliver effective results to end users. To bridge this gap, we introduce a novel hybrid method that assists designers in crafting user-centric prompts for T2I models, ensuring that the generated UIs align with end-user expectations. First, this method merges text mining and Kansei Engineering (KE) to analyze online user reviews and construct a Knowledge Graph (KG), mapping the intricate relationships between diverse affective requirements of users, design features, and corresponding text prompts for UI generation. Then, our approach automatically transforms designer inputs into model-preferred prompts through entity mention recognition and entity linking during the human-AI collaborative design process. Finally, we validate the proposed approach with a case study on automotive human–machine interface design. Experimental results demonstrate that our approach achieves high performance in perceived efficiency, satisfaction, and expectation disconfirmation. Overall, this study represents a step forward in integrating human and AI contributions in design and innovation within engineering disciplines, enabling AI to inspire, develop, and reinforce human creativity from a human factors perspective.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103217"},"PeriodicalIF":8.0,"publicationDate":"2025-03-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143552175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Predicting the performance status of aero-engines using a spatio-temporal decoupled digital twin modeling method

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-04 DOI: 10.1016/j.aei.2025.103218

Dasheng Xiao, Shuo Song, Hong Xiao, Zhanxue Wang

{"title":"Predicting the performance status of aero-engines using a spatio-temporal decoupled digital twin modeling method","authors":"Dasheng Xiao, Shuo Song, Hong Xiao, Zhanxue Wang","doi":"10.1016/j.aei.2025.103218","DOIUrl":"10.1016/j.aei.2025.103218","url":null,"abstract":"<div><div>The digital twin model for predicting engine performance enhances engine health management. Key indicators such as exhaust gas temperature (EGT) and thrust are essential for evaluating engine performance. This study focuses on extracting and integrating complex spatio-temporal features from multiple sensors to construct an effective prediction model. A data-driven modeling method that combines the physical structure of an engine while achieving spatio-temporal feature decoupled was proposed. This method is based on Long Short-Term Memory (LSTM) and a self-attention mechanism, and incorporates time-variant parameter derivatives into the model’s input using first-order backward differences. Case studies were conducted on the EGT and thrust predictions. The mean absolute relative error (<span><math><mrow><mi>M</mi><mi>A</mi><mi>R</mi><mi>E</mi></mrow></math></span>) was used to evaluate the accuracy of each test, whereas the average <span><math><mrow><mi>M</mi><mi>A</mi><mi>R</mi><mi>E</mi></mrow></math></span> (<span><math><msub><mrow><mi>μ</mi></mrow><mrow><mi>M</mi><mi>A</mi><mi>R</mi><mi>E</mi></mrow></msub></math></span>) across ten tests was used to assess the accuracy of each model. The results show that the spatio-temporal decoupled modeling method improves prediction accuracy and stability, achieving a minimum <span><math><msub><mrow><mi>μ</mi></mrow><mrow><mi>M</mi><mi>A</mi><mi>R</mi><mi>E</mi></mrow></msub></math></span> of 0.64% for the EGT and 0.277% for the normalized thrust. Furthermore, to test the method’s robustness against varying sampling frequencies during deployment, the sampling intervals of the test data were adjusted to simulate changes in sampling frequency. The results demonstrate that the proposed method exhibits excellent stability.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103218"},"PeriodicalIF":8.0,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143552929","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Production-logistics collaborative scheduling in dynamic flexible job shops using nested-hierarchical deep reinforcement learning

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-04 DOI: 10.1016/j.aei.2025.103195

Jiaxuan Shi , Fei Qiao , Juan Liu , Yumin Ma , Dongyuan Wang , Chen Ding

{"title":"Production-logistics collaborative scheduling in dynamic flexible job shops using nested-hierarchical deep reinforcement learning","authors":"Jiaxuan Shi , Fei Qiao , Juan Liu , Yumin Ma , Dongyuan Wang , Chen Ding","doi":"10.1016/j.aei.2025.103195","DOIUrl":"10.1016/j.aei.2025.103195","url":null,"abstract":"<div><div>Effective manufacturing in flexible job shops often requires collaboratively organizing production and logistics activities. This necessitates a thorough exploration of corresponding collaborative scheduling problem. However, extant studies remain relatively preliminary, not only neglecting the inevitable disturbances in real-world but also failing to satisfy the essential need for collaboration, that is, to simultaneously optimize both activities’ objectives. Therefore, this study proposes a novel production-logistics collaborative scheduling problem for dynamic flexible job shops, in which the common yet underappreciated disturbance of logistics equipment breakdowns is meticulously considered, and two typical objectives individually pursued by two activities are optimized simultaneously. To solve the proposed problem, a nested-hierarchical deep reinforcement learning method is developed. In this method, a new nested-hierarchical framework that rationally deploys multiple agents is designed to facilitate the required multi-objective optimization while ensuring the practicality of decision-making process. Based on this framework, appropriate state features, actions, and reward functions are devised for each agent, and a training mechanism based on multi-agent proximal policy optimization is proposed to train agents effectively. Experiments in an aviation component production shop are conducted to confirm the effectiveness of proposed method and problem.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103195"},"PeriodicalIF":8.0,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143552930","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi-scale 4D localized spatio-temporal graph convolutional networks for spatio-temporal sequences forecasting in aluminum electrolysis

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-03 DOI: 10.1016/j.aei.2025.103222

Yubo Sun , Xiaofang Chen , Weihua Gui , Lihui Cen , Yongfang Xie , Zhong Zou

{"title":"Multi-scale 4D localized spatio-temporal graph convolutional networks for spatio-temporal sequences forecasting in aluminum electrolysis","authors":"Yubo Sun , Xiaofang Chen , Weihua Gui , Lihui Cen , Yongfang Xie , Zhong Zou","doi":"10.1016/j.aei.2025.103222","DOIUrl":"10.1016/j.aei.2025.103222","url":null,"abstract":"<div><div>Spatio-temporal sequences forecasting fulfills a vital role in the intelligent advancement of aluminum electrolysis production process. The localized spatio-temporal correlations contained in spatio-temporal sequences, caused by the dynamicity of regional working conditions, have complex and diverse (multi-scale) characteristics. The existing deep learning-based prediction methods are difficult to capture the multi-scale localized spatio-temporal correlations, and the adverse effects of industrial noise on spatio-temporal correlation acquisition have not been considered. In this article, we propose the multi-scale 4D localized spatio-temporal graph convolutional networks (Ms-4D-LStGCN) to address the above issues. Concretely, we propose a data-driven accurate similarity search method and fuse the prior knowledge to construct the spatio-temporal graph. Then,a novel 4D localized spatio-temporal graph convolution module is proposed to capture the complex localized spatio-temporal correlations. Finally, the multi-scale 4D localized spatio-temporal graph convolution framework is designed to obtain the multi-scale and multi-depth localized spatio-temporal correlation features. Illustrative examples on 16 real-world industrial aluminum electrolysis datasets attest that our method has superior prediction performance compared with state-of-the-art methods.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103222"},"PeriodicalIF":8.0,"publicationDate":"2025-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143529440","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A physics-guided approach for accurate battery SOH estimation using RCMHCRE and BatteryPINN

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-03 DOI: 10.1016/j.aei.2025.103211

Yaming Liu , Haolin Chen , Ligang Yao , Jiaxin Ding , Shiqiang Chen , Zhenya Wang

{"title":"A physics-guided approach for accurate battery SOH estimation using RCMHCRE and BatteryPINN","authors":"Yaming Liu , Haolin Chen , Ligang Yao , Jiaxin Ding , Shiqiang Chen , Zhenya Wang","doi":"10.1016/j.aei.2025.103211","DOIUrl":"10.1016/j.aei.2025.103211","url":null,"abstract":"<div><div>Accurate monitoring of a battery’s state of health (SOH) is crucial for ensuring reliable operation. Data-driven methods for SOH estimation often involve complex feature extraction strategies and models that are difficult to interpret, limiting their generalizability. To overcome these challenges, this paper presents a battery SOH estimation method based on the refined composite multiscale Hilbert cumulative residual entropy algorithm (RCMHCRE) and the battery physical information neural network (BatteryPINN). First, the proposed RCMHCRE algorithm is applied to automatically extract high-quality health features from the battery’s voltage and current data, serving as the feature engineering in this study. Second, the network structure of BatteryPINN is developed for SOH prediction, based on the mathematical theory of solid electrolyte interphase (SEI) membrane growth. The proposed strategy enables BatteryPINN to be constrained by the battery aging mechanism during training, thereby ensuring that the network adheres to the underlying physical laws during propagation. To validate the effectiveness of the proposed method, a four-month battery aging experiment is conducted, and a dataset is constructed. Experimental results from three datasets demonstrate that the proposed method offers significant advantages in health feature extraction and SOH estimation compared to other state-of-the-art battery SOH estimation methods, achieving prediction accuracies of less than 1% for both RMSE and MAPE metrics.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103211"},"PeriodicalIF":8.0,"publicationDate":"2025-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143529494","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A new performance evaluation model based on approximate belief rule base with local uncertainty

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-03 DOI: 10.1016/j.aei.2025.103225

Jie Wang , Pengyun Ning , Zhijie Zhou , Peng Zhang

{"title":"A new performance evaluation model based on approximate belief rule base with local uncertainty","authors":"Jie Wang , Pengyun Ning , Zhijie Zhou , Peng Zhang","doi":"10.1016/j.aei.2025.103225","DOIUrl":"10.1016/j.aei.2025.103225","url":null,"abstract":"<div><div>Performance evaluation is of vital significance in guaranteeing the reliable operation of complex systems. During the process of performance evaluation, the limitations of expert knowledge and insufficient observation data pose challenges in differentiating adjacent performance states of complex systems. As such, a new performance evaluation model based on the approximate belief rule base with local uncertainty (ABRB-LU) is proposed in this paper. Regarding the model inference, the local uncertainty is assigned to the predefined vague state, which can effectively address the difficulty of distinguishing adjacent performance states. Subsequently, the multiple belief rules incorporating local uncertainty are fused by employing the evidential reasoning (ER) rule, contributing to establishing the evaluation model based on ABRB-LU. Meanwhile, an optimization objective is set to improve the evaluation accuracy. Regarding the model analysis, starting from two belief rules, a rigorous mathematical derivation is carried out to obtain the sensitivity factor of the evaluation results concerning the local uncertainty. On this basis, the analysis process is extended to multiple belief rules, forming a generalized method for sensitivity analysis. This can provide a scientific basis for decision-makers to locate weak links. An engineering example of servo mechanism is carried out to verify the validity of the proposed model.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103225"},"PeriodicalIF":8.0,"publicationDate":"2025-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143529441","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Deep-learning-driven intelligent tool wear identification of high-precision machining with multi-scale CNN-BiLSTM-GCN

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-02 DOI: 10.1016/j.aei.2025.103234

Zhicheng Xu , Baolong Zhang , Louis Luo Fan , Edward Hengzhou Yan , Dongfang Li , Zejia Zhao , Wai Sze Yip , Suet To

{"title":"Deep-learning-driven intelligent tool wear identification of high-precision machining with multi-scale CNN-BiLSTM-GCN","authors":"Zhicheng Xu , Baolong Zhang , Louis Luo Fan , Edward Hengzhou Yan , Dongfang Li , Zejia Zhao , Wai Sze Yip , Suet To","doi":"10.1016/j.aei.2025.103234","DOIUrl":"10.1016/j.aei.2025.103234","url":null,"abstract":"<div><div>In high-precision machining, the inevitable tool wear will significantly affect the surface quality. Traditional tool wear modeling is complicated due to the complex mathematical reasoning process and the identification of numerous unknown parameters linked to the wear mechanism. Data-driven modeling for tool wear prediction can avoid the above issues but suffers from high-cost and low-efficiency because of amounts of expensive tool consumption and time-consumed experiments for generating the training dataset. To fill this gap, this study proposed an innovative approach to accurately identifying tool wear in high-precision machining effortlessly. Firstly an unsupervised Modified Toeplitz Inverse Covariance Clustering (MTICC) algorithm was first proposed to objectively categorize tool wear phase from multi-channel time-series data to break through traditional manual-experience-based division, whose effectiveness was validated by the well-designed experiments. Then, a hybrid deep learning model with a multi-scale CNN-BiLSTM-GCN and cross-attention structures was developed to deeply extract spatial–temporal features from multi-channel signals by first considering the interdependencies of the sensor network for higher accuracy. After hyperparameters optimization, features importance analysis was conducted to identify the most important features, which are “X-force”, “Y-force” and “Phase1 Active Power”, and the hyperparameters importance quantitatively analyze the contribution of CNN, BiLSTM, and GCN modules, respectively. Through the comparative studies, the proposed multi-scale CNN-BiLSTM-GCN model performed better with a weighted average F1 score of 0.987 than other models. The proposed model was finally employed on the intelligent IoT platform and successfully achieved the real-time identification of tool wear in the HPM process.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103234"},"PeriodicalIF":8.0,"publicationDate":"2025-03-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143527434","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhanced deep learning framework for accurate near-failure RUL prediction of bearings in varying operating conditions

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-03-01 DOI: 10.1016/j.aei.2025.103231

Anil Kumar , Chander Parkash , Pradeep Kundu , Hesheng Tang , Jiawei Xiang

{"title":"Enhanced deep learning framework for accurate near-failure RUL prediction of bearings in varying operating conditions","authors":"Anil Kumar , Chander Parkash , Pradeep Kundu , Hesheng Tang , Jiawei Xiang","doi":"10.1016/j.aei.2025.103231","DOIUrl":"10.1016/j.aei.2025.103231","url":null,"abstract":"<div><div>Accurate prediction of remaining useful life (RUL) is crucial near the end of a component’s life to optimize maintenance resources and maximize life utilization. However, as a component nears failure, the uncertainty in its performance increases, which further complicates RUL prediction. This paper presents a framework for RUL prediction of bearings under varying conditions, with a focus on improving accuracy as failure approaches, where prediction reliability is essential. The framework employs a multi-head dynamic Graph Convolutional Network (GCN) with graph feature alignment using Maximum Mean Discrepancy (MMD). The multi-head graph attention mechanism selectively weights neighboring nodes, capturing diverse relationships and improving the model’s resilience to noise and structural variations. A novel Weighted-Huber loss function is introduced, applying a 100-fold penalty when the normalized RUL drops below 0.25, prioritizing accuracy in the critical end-of-life stage and addressing limitations in previous methods. Experimental results demonstrate the effect of the number of nearest neighbours (K) on GCN performance, with K = 5 yielding the best results. Additionally, the impact of the loss function on GCN performance is examined, as well as the performance comparison of the model with and without MMD domain adaptation. Finally, a comparison of the dynamic GCN with the standard GCN is included. The proposed method achieves lower mean squared error (MSE) and Huber loss compared to conventional approaches, validating the framework’s effectiveness under varied conditions and its superior performance in critical end-of-life prediction.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103231"},"PeriodicalIF":8.0,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143520998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Knowledge-enhanced spatial–temporal multi-frequency fusion transformer for plant-wide industrial process monitoring

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-02-27 DOI: 10.1016/j.aei.2025.103213

Jinyu Chen , Jianxiang Jin , Lei Zhang , Chengguang Wang , Wenjun Huang

{"title":"Knowledge-enhanced spatial–temporal multi-frequency fusion transformer for plant-wide industrial process monitoring","authors":"Jinyu Chen , Jianxiang Jin , Lei Zhang , Chengguang Wang , Wenjun Huang","doi":"10.1016/j.aei.2025.103213","DOIUrl":"10.1016/j.aei.2025.103213","url":null,"abstract":"<div><div>Plant-level industrial process monitoring is critical for ensuring production safety and efficiency. Existing state monitoring models predominantly focus on temporal modeling, neglecting the significance of underlying process structural knowledge and frequency domain information. To address these limitations, we propose a novel knowledge-enhanced spatial–temporal multi-frequency fusion transformer (KSTFformer) for comprehensive plant-wide process monitoring. Unlike existing data-driven methods, KSTFformer integrates prior knowledge with process data to enhance both modeling performance and interpretability. Initially, we decompose the entire process into multiple operating units based on equipment connection relationships, constructing variable correlations as a directed graph. We then design a graph convolution-based structural learning module that captures spatial–temporal dependencies using graph convolutional networks, leveraging the Gumbel-softmax sampling method to learn variable interconnections. Furthermore, a novel multi-band frequency domain feature extractor was innovatively proposed, which refines frequency domain sequences into multiple local sub-sequences and extracts frequency domain characteristics of variables through channel-independent networks. Finally, the model integrates spatial–temporal and frequency domain features through attention mechanism, enhancing anomaly detection capabilities. The proposed method’s effectiveness is validated through two real-world industrial process case studies.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103213"},"PeriodicalIF":8.0,"publicationDate":"2025-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143509140","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Universal federated domain adaptation for gearbox fault diagnosis: A robust framework for credible pseudo-label generation

IF 8 1区工程技术

Advanced Engineering Informatics Pub Date : 2025-02-27 DOI: 10.1016/j.aei.2025.103233

Xinyu Ren , Suixin Wang , Wanli Zhao , Xiangxing Kong , Manyi Fan , Haidong Shao , Ke Zhao

{"title":"Universal federated domain adaptation for gearbox fault diagnosis: A robust framework for credible pseudo-label generation","authors":"Xinyu Ren , Suixin Wang , Wanli Zhao , Xiangxing Kong , Manyi Fan , Haidong Shao , Ke Zhao","doi":"10.1016/j.aei.2025.103233","DOIUrl":"10.1016/j.aei.2025.103233","url":null,"abstract":"<div><div>Intelligent fault diagnosis methods based on transfer learning have been widely applied in the field of industrial equipment, effectively addressing the fundamental assumption of fault type matching between source and target domains. However, existing approaches face two critical limitations in practical applications: (1) Most current methods are typically tailored to specific domain adaptation scenarios, making it challenging to achieve satisfactory diagnostic performance under diverse real-world operating conditions. (2) These methods fail to adequately address data privacy concerns, as they often require all users to share local data for centralized model training—a practice that is infeasible in industrial settings. To overcome these challenges, this study proposes a universal federated domain adaptive fault diagnosis method. First, to mitigate the impact of categories absent in the target domain on diagnostic performance, multiple candidate pseudo-labels are generated by predicting the outputs of target domain samples using the source model. Next, a reliable pseudo-label generation mechanism based on a Gaussian Mixture Model (GMM) is developed to effectively eliminate ambiguities and noise in the pseudo-labels. Finally, a voting strategy is introduced, leveraging the consensus knowledge of all source and target models to partition target domain samples into shared and private categories. This enhances the global model’s ability to understand and classify target domain data. Experimental results on two gearbox datasets demonstrate that the proposed method excels in identifying unknown fault patterns while preserving data privacy. These findings highlight its superior performance and broad applicability in complex industrial scenarios.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"65 ","pages":"Article 103233"},"PeriodicalIF":8.0,"publicationDate":"2025-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143509319","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0