Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102903
Jie Wang , Haidong Shao , Yiming Xiao , Bin Liu
Currently, source free domain adaptation (SFDA) methods are employed to address the issue of inaccessible source domain data (SDD) in transfer learning. However, existing SFDA methods often suffer from overfitting to specific domains, leading to poor generalization ability in the target domain. To address these challenges, this paper proposes a novel SFDA method named SFDA-T for fault diagnosis. Specifically, a Transformer-CNN-based feature extractor is constructed, to mine the transferable feature knowledge of faults in the SDD. The approach reduces the overfitting of the model to domain-specific information and improves model’s generalization ability. In addition, the feature attention loss is designed to calculate attention weights of the sample features to increase the model’s attention to the crucial feature regions in the target domain. A source similarity guided exponential loss is developed to guide target samples based on the decision boundaries of the source domain, facilitating cluster alignment of target sample categories and expanding distances between different categories. Furthermore, a self-training pseudo-labeling constraint is employed to reduce the effect of incorrect label matching and further constrain the model. The results of the experiments on gearboxes and bearings indicate that the proposed method achieves high fault diagnosis accuracy while effectively decoupling from SDD.
{"title":"SFDA-T: A novel source-free domain adaptation method with strong generalization ability for fault diagnosis","authors":"Jie Wang , Haidong Shao , Yiming Xiao , Bin Liu","doi":"10.1016/j.aei.2024.102903","DOIUrl":"10.1016/j.aei.2024.102903","url":null,"abstract":"<div><div>Currently, source free domain adaptation (SFDA) methods are employed to address the issue of inaccessible source domain data (SDD) in transfer learning. However, existing SFDA methods often suffer from overfitting to specific domains, leading to poor generalization ability in the target domain. To address these challenges, this paper proposes a novel SFDA method named SFDA-T for fault diagnosis. Specifically, a Transformer-CNN-based feature extractor is constructed, to mine the transferable feature knowledge of faults in the SDD. The approach reduces the overfitting of the model to domain-specific information and improves model’s generalization ability. In addition, the feature attention loss is designed to calculate attention weights of the sample features to increase the model’s attention to the crucial feature regions in the target domain. A source similarity guided exponential loss is developed to guide target samples based on the decision boundaries of the source domain, facilitating cluster alignment of target sample categories and expanding distances between different categories. Furthermore, a self-training pseudo-labeling constraint is employed to reduce the effect of incorrect label matching and further constrain the model. The results of the experiments on gearboxes and bearings indicate that the proposed method achieves high fault diagnosis accuracy while effectively decoupling from SDD.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102903"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142532008","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}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102861
Yuanrong Zhang , Wei Guo , Zhixing Chang , Jian Ma , Zhonglin Fu , Lei Wang , Hongyu Shao
In recent years, an increasing number of studies have focused on user requirement modeling based on online review texts. However, traditional methods often overlook the integration of user requirement models with product design frameworks, failing to effectively transform dynamically changing user requirements into a basis for product attribute upgrades. This paper proposes a user requirement modeling and evolutionary analysis method based on review data, supporting the design upgrade of product attributes. This approach differs from traditional user requirement modeling and analysis methods in two main aspects: (1) integrating the designer’s product design framework into the classification and modeling of user requirements; (2) analyzing the dynamic changes in user requirements during product upgrades and formulating new product attribute upgrade strategies. Initially, the study extracts three categories of product attributes that designers are concerned about from the review data: function (F), structure (S), and parameters (P), and establishes a correlation model between these product attributes. Subsequently, using natural language processing technology to calculate sentiment scores for product attributes and employing the Multi-Layer Perceptron (MLP) model to analyze the impact of product attribute sentiment on user satisfaction, the study constructs the FSP-Kano model, achieving classification and modeling of user requirements for these three categories of product attributes. Finally, based on the dynamic changes in user requirements within the FSP-Kano model, strategies for upgrading next-generation products are formulated. Additionally, the study illustrates the proposed method with the example of BYD’s “Qin” series of new energy vehicles. Our research demonstrates that the proposed method can accurately and comprehensively extract user requirements and develop successful product attribute improvement strategies for the next generation of products.
{"title":"User requirement modeling and evolutionary analysis based on review data: Supporting the design upgrade of product attributes","authors":"Yuanrong Zhang , Wei Guo , Zhixing Chang , Jian Ma , Zhonglin Fu , Lei Wang , Hongyu Shao","doi":"10.1016/j.aei.2024.102861","DOIUrl":"10.1016/j.aei.2024.102861","url":null,"abstract":"<div><div>In recent years, an increasing number of studies have focused on user requirement modeling based on online review texts. However, traditional methods often overlook the integration of user requirement models with product design frameworks, failing to effectively transform dynamically changing user requirements into a basis for product attribute upgrades. This paper proposes a user requirement modeling and evolutionary analysis method based on review data, supporting the design upgrade of product attributes. This approach differs from traditional user requirement modeling and analysis methods in two main aspects: (1) integrating the designer’s product design framework into the classification and modeling of user requirements; (2) analyzing the dynamic changes in user requirements during product upgrades and formulating new product attribute upgrade strategies. Initially, the study extracts three categories of product attributes that designers are concerned about from the review data: function (F), structure (S), and parameters (P), and establishes a correlation model between these product attributes. Subsequently, using natural language processing technology to calculate sentiment scores for product attributes and employing the Multi-Layer Perceptron (MLP) model to analyze the impact of product attribute sentiment on user satisfaction, the study constructs the FSP-Kano model, achieving classification and modeling of user requirements for these three categories of product attributes. Finally, based on the dynamic changes in user requirements within the FSP-Kano model, strategies for upgrading next-generation products are formulated. Additionally, the study illustrates the proposed method with the example of BYD’s “Qin” series of new energy vehicles. Our research demonstrates that the proposed method can accurately and comprehensively extract user requirements and develop successful product attribute improvement strategies for the next generation of products.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102861"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142532238","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}
Active back-support exoskeleton has emerged as a potential solution for mitigating work-related musculoskeletal disorders within the construction industry. Nevertheless, research has unveiled unintended consequences associated with its usage, most notably increased cognitive load. Elevated cognitive load has been shown to deplete working memory, potentially impeding task performance and situational awareness. Despite the susceptibility of exoskeleton users to increased cognitive load, there has been limited empirical evaluation of this risk while performing construction tasks. This study evaluates the cognitive load associated with using an active back-support exoskeleton while performing construction tasks. An experiment was conducted to capture brain activity using an Electroencephalogram, both with and without the use of an active back-support exoskeleton. A construction framing task involving six subtasks was considered as a case study. The participants’ cognitive load was assessed for the tested conditions and subtasks through the alpha band of the Electroencephalogram signals. The study identified the most sensitive Electroencephalogram channels for evaluating cognitive load when using exoskeletons. Statistical tests, including a one-way repeated measure ANOVA, paired t-test, and Spearman Rank were conducted to make inferences about the collected data. The results revealed that using an active back-support exoskeleton while performing the carpentry framing task increased the cognitive load of the participants, as indicated by four out of five significant Electroencephalogram channels. Selected channels in the frontal and occipital lobes emerged as the most influential channels in assessing cognitive load. Additionally, the study explores the relationships among Electroencephalogram channels, revealing strong correlations between selected channels in the frontal lobe and between channels in the occipital and frontal lobes. These findings enhance understanding of how specific brain regions respond to the use of active back support exoskeletons during construction tasks. By identifying which brain regions are most affected, this study contributes to optimizing exoskeleton designs to better manage cognitive load, potentially improving both the ergonomic effectiveness and safety of these devices in construction environments.
主动式背部支撑外骨骼已成为减轻建筑行业与工作有关的肌肉骨骼疾病的潜在解决方案。然而，研究揭示了与使用外骨骼相关的意外后果，其中最明显的是认知负荷的增加。研究表明，认知负荷的增加会耗尽工作记忆，从而可能妨碍任务执行和态势感知。尽管外骨骼使用者容易受到认知负荷增加的影响，但在执行建筑任务时对这种风险的实证评估却很有限。本研究评估了在执行建筑任务时使用主动式背部支撑外骨骼所带来的认知负荷。在使用和不使用主动式背部支撑外骨骼的情况下，都进行了使用脑电图捕捉大脑活动的实验。案例研究考虑了一项涉及六个子任务的建筑框架任务。通过脑电图信号的阿尔法波段来评估参与者在测试条件和子任务中的认知负荷。研究确定了使用外骨骼时评估认知负荷最敏感的脑电图通道。为了对收集到的数据进行推断，还进行了统计测试，包括单向重复测量方差分析、配对 t 检验和斯皮尔曼等级检验。结果表明，在执行木工框架任务时使用主动式背部支撑外骨骼会增加参与者的认知负荷，五个重要脑电图通道中的四个都表明了这一点。在评估认知负荷时，额叶和枕叶的选定通道成为最有影响力的通道。此外，该研究还探讨了脑电图通道之间的关系，揭示了额叶选定通道之间以及枕叶和额叶通道之间的强相关性。这些发现加深了人们对特定脑区在建筑任务中如何对使用主动式背部支撑外骨骼做出反应的理解。通过确定哪些脑区受到的影响最大，这项研究有助于优化外骨骼设计以更好地管理认知负荷，从而有可能提高这些设备在建筑环境中的人体工学效果和安全性。
{"title":"Cognitive load assessment of active back-support exoskeletons in construction: A case study on construction framing","authors":"Abiola Akanmu , Akinwale Okunola , Houtan Jebelli , Ashtarout Ammar , Adedeji Afolabi","doi":"10.1016/j.aei.2024.102905","DOIUrl":"10.1016/j.aei.2024.102905","url":null,"abstract":"<div><div>Active back-support exoskeleton has emerged as a potential solution for mitigating work-related musculoskeletal disorders within the construction industry. Nevertheless, research has unveiled unintended consequences associated with its usage, most notably increased cognitive load. Elevated cognitive load has been shown to deplete working memory, potentially impeding task performance and situational awareness. Despite the susceptibility of exoskeleton users to increased cognitive load, there has been limited empirical evaluation of this risk while performing construction tasks. This study evaluates the cognitive load associated with using an active back-support exoskeleton while performing construction tasks. An experiment was conducted to capture brain activity using an Electroencephalogram, both with and without the use of an active back-support exoskeleton. A construction framing task involving six subtasks was considered as a case study. The participants’ cognitive load was assessed for the tested conditions and subtasks through the alpha band of the Electroencephalogram signals. The study identified the most sensitive Electroencephalogram channels for evaluating cognitive load when using exoskeletons. Statistical tests, including a one-way repeated measure ANOVA, paired <em>t</em>-test, and Spearman Rank were conducted to make inferences about the collected data. The results revealed that using an active back-support exoskeleton while performing the carpentry framing task increased the cognitive load of the participants, as indicated by four out of five significant Electroencephalogram channels. Selected channels in the frontal and occipital lobes emerged as the most influential channels in assessing cognitive load. Additionally, the study explores the relationships among Electroencephalogram channels, revealing strong correlations between selected channels in the frontal lobe and between channels in the occipital and frontal lobes. These findings enhance understanding of how specific brain regions respond to the use of active back support exoskeletons during construction tasks. By identifying which brain regions are most affected, this study contributes to optimizing exoskeleton designs to better manage cognitive load, potentially improving both the ergonomic effectiveness and safety of these devices in construction environments.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102905"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142532245","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}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102899
Cheng Zhou , Yuxiang Wang , Ke You , Rubin Wang
Automated monitoring of bulldozer operation cycles is crucial for efficient productivity assessment and precise construction management. Harsh earthwork environments and complex, variable operation processes present challenges for identifying these cycles. To address this issue, we developed a multiscale temporal feature fusion and dual attention mechanism-based temporal action detection model (FDA-AFSD) for the automatic identification of bulldozer operation cycles from in to vehicle vision. This model enhances long-term sequence modeling, key temporal information learning, and precise action boundary identification through its multiscale temporal feature fusion structure, dual attention mechanism module, and scalable granularity perception (SGP) layer. In tests for earth levelling and mine edge dumping operations, the average detection accuracy (mAP) for bulldozer operation cycles reached 90.9%. Furthermore, under various adverse weather conditions and diverse construction processes, the model maintained stable and excellent detection capabilities, demonstrating its feasibility and practical application value.
{"title":"In-vehicle vision-based automatic identification of bulldozer operation cycles with temporal action detection","authors":"Cheng Zhou , Yuxiang Wang , Ke You , Rubin Wang","doi":"10.1016/j.aei.2024.102899","DOIUrl":"10.1016/j.aei.2024.102899","url":null,"abstract":"<div><div>Automated monitoring of bulldozer operation cycles is crucial for efficient productivity assessment and precise construction management. Harsh earthwork environments and complex, variable operation processes present challenges for identifying these cycles. To address this issue, we developed a multiscale temporal feature fusion and dual attention mechanism-based temporal action detection model (FDA-AFSD) for the automatic identification of bulldozer operation cycles from in to vehicle vision. This model enhances long-term sequence modeling, key temporal information learning, and precise action boundary identification through its multiscale temporal feature fusion structure, dual attention mechanism module, and scalable granularity perception (SGP) layer. In tests for earth levelling and mine edge dumping operations, the average detection accuracy (mAP) for bulldozer operation cycles reached 90.9%. Furthermore, under various adverse weather conditions and diverse construction processes, the model maintained stable and excellent detection capabilities, demonstrating its feasibility and practical application value.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102899"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142532246","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}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102921
Ruizhe Liu , Qiubing Ren , Mingchao Li , Xiaocui Ji , Ting Liu , Hao Liu
Precisely predicting concrete dam displacements is crucial for assessing their structural behavior during operation. Many studies have testified that ensemble methods are more accurate and applicable in practice than individual predictive models. Nevertheless, the common way handling massive monitoring data is still conventional, that is, training and testing them as a whole, neglecting the internal law and pattern difference within data, which probably limits advancements in predictive effect. To this end, the patterns of monitoring data are identified and classified before model establishment, and a similarity-aware ensemble method (SAEM) using temporal division and fully Bayesian learning is presented for dam displacement prediction. Specifically, the unsupervised fuzzy C-means approach and sparrow search algorithm are fused for similar pattern clustering of environmental factors, thus achieving temporal division in displacement responses. Fully considering the adaptability of model structure and parameters to various data patterns, a non-parametric fully Bayesian Gaussian process regression (FBGPR) model is proposed by augmenting the standard GPR with Markov chain Monte Carlo simulation and Bayesian evidence evaluation theory. Different data clusters are then fed into FBGPR in chronological order, and the final results are derived through a grouping ensemble scheme. Multiple sets of monitoring data collected from a real-world dam project are employed for method verification. The results show that our proposed SAEM has better prediction accuracy compared to homogeneous clustering-based ensemble methods and commonly used individual models. The superior performance in two additional cases also verifies the adaptability and generalization ability of our method, which provides a new modeling tool for structural health monitoring of concrete dams.
{"title":"A similarity-aware ensemble method for displacement prediction of concrete dams based on temporal division and fully Bayesian learning","authors":"Ruizhe Liu , Qiubing Ren , Mingchao Li , Xiaocui Ji , Ting Liu , Hao Liu","doi":"10.1016/j.aei.2024.102921","DOIUrl":"10.1016/j.aei.2024.102921","url":null,"abstract":"<div><div>Precisely predicting concrete dam displacements is crucial for assessing their structural behavior during operation. Many studies have testified that ensemble methods are more accurate and applicable in practice than individual predictive models. Nevertheless, the common way handling massive monitoring data is still conventional, that is, training and testing them as a whole, neglecting the internal law and pattern difference within data, which probably limits advancements in predictive effect. To this end, the patterns of monitoring data are identified and classified before model establishment, and a similarity-aware ensemble method (SAEM) using temporal division and fully Bayesian learning is presented for dam displacement prediction. Specifically, the unsupervised fuzzy C-means approach and sparrow search algorithm are fused for similar pattern clustering of environmental factors, thus achieving temporal division in displacement responses. Fully considering the adaptability of model structure and parameters to various data patterns, a non-parametric fully Bayesian Gaussian process regression (FBGPR) model is proposed by augmenting the standard GPR with Markov chain Monte Carlo simulation and Bayesian evidence evaluation theory. Different data clusters are then fed into FBGPR in chronological order, and the final results are derived through a grouping ensemble scheme. Multiple sets of monitoring data collected from a real-world dam project are employed for method verification. The results show that our proposed SAEM has better prediction accuracy compared to homogeneous clustering-based ensemble methods and commonly used individual models. The superior performance in two additional cases also verifies the adaptability and generalization ability of our method, which provides a new modeling tool for structural health monitoring of concrete dams.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102921"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142572469","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}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102909
L. Magadán , C. Ruiz-Cárcel , J.C. Granda , F.J. Suárez , A. Starr
Rotating machinery plays an essential role in various industrial processes such as manufacturing, power generation, and transportation. These machines, which include turbines, pumps, motors, compressors, and many others, are the heartbeats of numerous industries. The seamless operation of these machines is critical for the efficiency and productivity of these sectors. However, over time, these machines degrade and can suffer faults. One of the most critical components are bearings, which can suffer different types of faults. This paper presents a novel approach for bearing fault classification and diagnosis under limited data. A Monotonic Smoothed Stacked AutoEncoder (MS2AE) is used to infer a smoothed monotonic health index from raw bearing acceleration data. The MS2AE is trained using only healthy data, so this approach can also be used with recently comisioned equipment that has not failed yet. Then, using the evolution of the health index, a first faulty point is computed, so two stages are identified in the lifespan of the rotating machinery: healthy and faulty. Correlation matrices are computed to show the relationship of the health index with time-domain and frequency-domain features in order to provide explainability and validate the health index construction process. When the health index is classified as faulty, Dynamic Time Warping is applied between healthy samples and faulty samples to extract differences. Finally, based on a 1/3-binary tree 3 level kurtogram, these differences are filtered using a bandpass filter and converted to the frequency domain, where characteristic harmonics are used to identify the type of bearing fault. The explainability provided in the health index construction process makes the system useful in certain industries where black-box AI models cannot be trusted due to strict regulations. The classification and diagnosis system achieves robustness in fault classification under different working conditions by utilizing multiple bearing fault datsets. Its ability to be trained using only healthy data and the interpretability offered, makes it suitable for recently installed rotating machinery in real industrial facilities, without requiring qualified staff.
{"title":"Explainable and interpretable bearing fault classification and diagnosis under limited data","authors":"L. Magadán , C. Ruiz-Cárcel , J.C. Granda , F.J. Suárez , A. Starr","doi":"10.1016/j.aei.2024.102909","DOIUrl":"10.1016/j.aei.2024.102909","url":null,"abstract":"<div><div>Rotating machinery plays an essential role in various industrial processes such as manufacturing, power generation, and transportation. These machines, which include turbines, pumps, motors, compressors, and many others, are the heartbeats of numerous industries. The seamless operation of these machines is critical for the efficiency and productivity of these sectors. However, over time, these machines degrade and can suffer faults. One of the most critical components are bearings, which can suffer different types of faults. This paper presents a novel approach for bearing fault classification and diagnosis under limited data. A Monotonic Smoothed Stacked AutoEncoder (MS2AE) is used to infer a smoothed monotonic health index from raw bearing acceleration data. The MS2AE is trained using only healthy data, so this approach can also be used with recently comisioned equipment that has not failed yet. Then, using the evolution of the health index, a first faulty point is computed, so two stages are identified in the lifespan of the rotating machinery: healthy and faulty. Correlation matrices are computed to show the relationship of the health index with time-domain and frequency-domain features in order to provide explainability and validate the health index construction process. When the health index is classified as faulty, Dynamic Time Warping is applied between healthy samples and faulty samples to extract differences. Finally, based on a 1/3-binary tree 3 level kurtogram, these differences are filtered using a bandpass filter and converted to the frequency domain, where characteristic harmonics are used to identify the type of bearing fault. The explainability provided in the health index construction process makes the system useful in certain industries where black-box AI models cannot be trusted due to strict regulations. The classification and diagnosis system achieves robustness in fault classification under different working conditions by utilizing multiple bearing fault datsets. Its ability to be trained using only healthy data and the interpretability offered, makes it suitable for recently installed rotating machinery in real industrial facilities, without requiring qualified staff.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102909"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142586518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102872
Lixiang Zhang , Yan Yan , Chen Yang , Yaoguang Hu
Achieving mass personalization presents significant challenges in performance and adaptability when solving dynamic flexible job-shop scheduling problems (DFJSP). Previous studies have struggled to achieve high performance in variable contexts. To tackle this challenge, this paper introduces a novel scheduling strategy founded on heterogeneous multi-agent reinforcement learning. This strategy facilitates centralized optimization and decentralized decision-making through collaboration among job and machine agents while employing historical experiences to support data-driven learning. The DFJSP with transportation time is initially formulated as heterogeneous multi-agent partial observation Markov Decision Processes. This formulation outlines the interactions between decision-making agents and the environment, incorporating a reward-shaping mechanism aimed at organizing job and machine agents to minimize the weighted tardiness of dynamic jobs. Then, we develop a dueling double deep Q-network algorithm incorporating the reward-shaping mechanism to ascertain the optimal strategies for machine allocation and job sequencing in DFJSP. This approach addresses the sparse reward issue and accelerates the learning process. Finally, the efficiency of the proposed method is verified and validated through numerical experiments, which demonstrate its superiority in reducing the weighted tardiness of dynamic jobs when compared to state-of-the-art baselines. The proposed method exhibits remarkable adaptability in encountering new scenarios, underscoring the benefits of adopting a heterogeneous multi-agent reinforcement learning-based scheduling approach in navigating dynamic and flexible challenges.
{"title":"Dynamic flexible job-shop scheduling by multi-agent reinforcement learning with reward-shaping","authors":"Lixiang Zhang , Yan Yan , Chen Yang , Yaoguang Hu","doi":"10.1016/j.aei.2024.102872","DOIUrl":"10.1016/j.aei.2024.102872","url":null,"abstract":"<div><div>Achieving mass personalization presents significant challenges in performance and adaptability when solving dynamic flexible job-shop scheduling problems (DFJSP). Previous studies have struggled to achieve high performance in variable contexts. To tackle this challenge, this paper introduces a novel scheduling strategy founded on heterogeneous multi-agent reinforcement learning. This strategy facilitates centralized optimization and decentralized decision-making through collaboration among job and machine agents while employing historical experiences to support data-driven learning. The DFJSP with transportation time is initially formulated as heterogeneous multi-agent partial observation Markov Decision Processes. This formulation outlines the interactions between decision-making agents and the environment, incorporating a reward-shaping mechanism aimed at organizing job and machine agents to minimize the weighted tardiness of dynamic jobs. Then, we develop a dueling double deep Q-network algorithm incorporating the reward-shaping mechanism to ascertain the optimal strategies for machine allocation and job sequencing in DFJSP. This approach addresses the sparse reward issue and accelerates the learning process. Finally, the efficiency of the proposed method is verified and validated through numerical experiments, which demonstrate its superiority in reducing the weighted tardiness of dynamic jobs when compared to state-of-the-art baselines. The proposed method exhibits remarkable adaptability in encountering new scenarios, underscoring the benefits of adopting a heterogeneous multi-agent reinforcement learning-based scheduling approach in navigating dynamic and flexible challenges.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102872"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142446798","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}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102932
Ankit Das , Debraj Ghosh , Shing-Fung Lau , Pavitra Srivastava , Aniruddha Ghosh , Chien-Fang Ding
Additive manufacturing (AM) is a versatile, primary manufacturing method widely employed in aerospace, medical, and automotive industries. This environmentally friendly process involves complex phenomena, necessitating comprehensive monitoring for process insights. This review examines AM process monitoring systems, including optical cameras, thermography, and radiography. These technologies generate substantial data, enabling soft computing and machine learning applications for efficiency enhancement and process optimization. Focusing on laser-based AM, the review discusses existing monitoring methods, their limitations, and potential solutions. It explores intelligent AM systems and in-situ X-ray synchrotron techniques, highlighting the transformative potential of efficient process monitoring. The review briefly introduces AM classification, outlines current monitoring methods and their constraints, and proposes smart laser-based AM systems with an overview of applicable machine learning techniques. Finally, it presents plausible solutions to identified limitations and discusses future prospects, emphasizing the revolutionary impact of effective process monitoring on laser AM processes.
快速成型制造（AM）是一种多功能的初级制造方法，广泛应用于航空航天、医疗和汽车行业。这种环境友好型工艺涉及复杂的现象，需要进行全面监控以深入了解工艺。本综述探讨了 AM 工艺监控系统，包括光学相机、热成像和射线照相术。这些技术可生成大量数据，从而实现软计算和机器学习应用，以提高效率和优化工艺。本综述以基于激光的 AM 为重点，讨论了现有的监控方法、其局限性以及潜在的解决方案。它探讨了智能 AM 系统和原位 X 射线同步加速器技术，强调了高效流程监控的变革潜力。综述简要介绍了 AM 分类，概述了当前的监控方法及其限制因素，并提出了基于激光的智能 AM 系统，同时概述了适用的机器学习技术。最后，文章针对已发现的局限性提出了可行的解决方案，并讨论了未来前景，强调了有效过程监控对激光 AM 过程的革命性影响。
{"title":"A critical review of process monitoring for laser-based additive manufacturing","authors":"Ankit Das , Debraj Ghosh , Shing-Fung Lau , Pavitra Srivastava , Aniruddha Ghosh , Chien-Fang Ding","doi":"10.1016/j.aei.2024.102932","DOIUrl":"10.1016/j.aei.2024.102932","url":null,"abstract":"<div><div>Additive manufacturing (AM) is a versatile, primary manufacturing method widely employed in aerospace, medical, and automotive industries. This environmentally friendly process involves complex phenomena, necessitating comprehensive monitoring for process insights. This review examines AM process monitoring systems, including optical cameras, thermography, and radiography. These technologies generate substantial data, enabling soft computing and machine learning applications for efficiency enhancement and process optimization. Focusing on laser-based AM, the review discusses existing monitoring methods, their limitations, and potential solutions. It explores intelligent AM systems and in-situ X-ray synchrotron techniques, highlighting the transformative potential of efficient process monitoring. The review briefly introduces AM classification, outlines current monitoring methods and their constraints, and proposes smart laser-based AM systems with an overview of applicable machine learning techniques. Finally, it presents plausible solutions to identified limitations and discusses future prospects, emphasizing the revolutionary impact of effective process monitoring on laser AM processes.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102932"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142659053","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}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102912
Zhenyu Liu , Haowen Zheng , Hui Liu , Weiqiang Jia , Jianrong Tan
Real-time data may undergo distribution drift due to changes in operating conditions and other factors, which can affect the classification accuracy of online fault diagnosis models and potentially lead to serious consequences. Therefore, understanding the classification accuracy of the model on real-time data holds substantial significance. However, the absence of labels in real-time data presents a challenge for evaluating classification accuracy. Furthermore, the real-time nature of fault diagnosis necessitates a swift and straightforward evaluation method. For the above reasons, this paper proposes a method for evaluating the classification accuracy of a model on real-time data, which is done in the absence of labels for the real-time data. The proposed label-free evaluation method transforms the model’s output into a scalar that measures the degree of matching between the classification probabilities, termed the average free energy. It then establishes a mapping between the average free energy and the classification accuracy using an auxiliary dataset. Finally, it predicts the model’s classification accuracy on the real-time data through this mapping and the average free energy of the real-time data. The proposed method is experimentally evaluated on public datasets, demonstrating its superiority in various aspects.
{"title":"Label-free evaluation for performance of fault diagnosis model on unknown distribution dataset","authors":"Zhenyu Liu , Haowen Zheng , Hui Liu , Weiqiang Jia , Jianrong Tan","doi":"10.1016/j.aei.2024.102912","DOIUrl":"10.1016/j.aei.2024.102912","url":null,"abstract":"<div><div>Real-time data may undergo distribution drift due to changes in operating conditions and other factors, which can affect the classification accuracy of online fault diagnosis models and potentially lead to serious consequences. Therefore, understanding the classification accuracy of the model on real-time data holds substantial significance. However, the absence of labels in real-time data presents a challenge for evaluating classification accuracy. Furthermore, the real-time nature of fault diagnosis necessitates a swift and straightforward evaluation method. For the above reasons, this paper proposes a method for evaluating the classification accuracy of a model on real-time data, which is done in the absence of labels for the real-time data. The proposed label-free evaluation method transforms the model’s output into a scalar that measures the degree of matching between the classification probabilities, termed the average free energy. It then establishes a mapping between the average free energy and the classification accuracy using an auxiliary dataset. Finally, it predicts the model’s classification accuracy on the real-time data through this mapping and the average free energy of the real-time data. The proposed method is experimentally evaluated on public datasets, demonstrating its superiority in various aspects.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102912"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142659111","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}
Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102914
Meiyu Cui , Ranran Gao , Libiao Peng , Xifeng Li , Dongjie Bi , Yongle Xie
In the field of mechanical equipment maintenance, accurately estimating the remaining useful life (RUL) of rolling bearings is crucial for ensuring reliable equipment operation. However, prevalent deep learning methods face challenges such as limited sample sizes, and “black-box” mechanisms. To enhance the accuracy and interpretability of rolling bearing RUL prediction, a novel fractional-derivative kernel mean -power error filtering algorithm (FrKMPE) is introduced. A comprehensive analysis of convergence for this method in terms of both mean error and mean square error criteria is provided. By combining the memory properties of fractional-derivative with the adaptability of kernel method, it can effectively capture features of non-stationary signals and sensitively monitor changes of rolling bearing health states (HSs). The effectiveness of the FrKMPE is validated through its application to the prediction of RUL using the IEEE PHM 2012 challenge dataset and the XJTU-SY dataset. Experimental results demonstrate that the proposed FrKMPE outperforms existing kernel adaptive filtering and deep learning methods in rolling bearing RUL prediction. The proposed method has advantages in dealing with complex nonlinear data and improving prediction accuracy, and provides a new perspective and solution for rolling bearing RUL prediction.
{"title":"A fractional-derivative kernel learning strategy for predicting residual life of rolling bearings","authors":"Meiyu Cui , Ranran Gao , Libiao Peng , Xifeng Li , Dongjie Bi , Yongle Xie","doi":"10.1016/j.aei.2024.102914","DOIUrl":"10.1016/j.aei.2024.102914","url":null,"abstract":"<div><div>In the field of mechanical equipment maintenance, accurately estimating the remaining useful life (RUL) of rolling bearings is crucial for ensuring reliable equipment operation. However, prevalent deep learning methods face challenges such as limited sample sizes, and “black-box” mechanisms. To enhance the accuracy and interpretability of rolling bearing RUL prediction, a novel fractional-derivative kernel mean <span><math><mi>p</mi></math></span>-power error filtering algorithm (FrKMPE) is introduced. A comprehensive analysis of convergence for this method in terms of both mean error and mean square error criteria is provided. By combining the memory properties of fractional-derivative with the adaptability of kernel method, it can effectively capture features of non-stationary signals and sensitively monitor changes of rolling bearing health states (HSs). The effectiveness of the FrKMPE is validated through its application to the prediction of RUL using the IEEE PHM 2012 challenge dataset and the XJTU-SY dataset. Experimental results demonstrate that the proposed FrKMPE outperforms existing kernel adaptive filtering and deep learning methods in rolling bearing RUL prediction. The proposed method has advantages in dealing with complex nonlinear data and improving prediction accuracy, and provides a new perspective and solution for rolling bearing RUL prediction.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102914"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142659194","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号