Pub Date : 2024-10-01DOI: 10.1016/j.aei.2024.102952
Tianchi Ma , Yuguang Fu
Deep learning-based blade crack detection models work on the premise of a fixed data distribution, while the influx of new dataset for faults under blade crack propagation often leads to a catastrophic forgetting problem. Meanwhile, it is difficult for a single sensor to reflect the health status of the blade comprehensively under the limitation of installation location and coverage. To solve the above problems, a multi-sensor fused incremental detection model (MFIDM) for blade cracks with the cross-attention mechanism and the Dempster-Shafer evidence theory (DST) is proposed. Firstly, vibration signals of centrifugal fans are collected by multiple accelerometers deployed at different locations. Then, a two-branch feature fusion method based on the cross-attention mechanism is proposed to overcome the class imbalance due to the replay incremental learning method. After that, the fused features are fed into a Softmax classifier to complete the initial classification of blade status. Finally, a modified DST based on the cross-correlation energy is adopted for multi-sensor decision fusion to obtain the final blade crack detection results. The effectiveness of the proposed method is verified by two incremental blade crack datasets, and MFIDM achieves the better performance compared with other related incremental detection methods.
{"title":"A multi-sensor fused incremental detection model for blade crack with cross-attention mechanism and Dempster-Shafer evidence theory","authors":"Tianchi Ma , Yuguang Fu","doi":"10.1016/j.aei.2024.102952","DOIUrl":"10.1016/j.aei.2024.102952","url":null,"abstract":"<div><div>Deep learning-based blade crack detection models work on the premise of a fixed data distribution, while the influx of new dataset for faults under blade crack propagation often leads to a catastrophic forgetting problem. Meanwhile, it is difficult for a single sensor to reflect the health status of the blade comprehensively under the limitation of installation location and coverage. To solve the above problems, a multi-sensor fused incremental detection model (MFIDM) for blade cracks with the cross-attention mechanism and the Dempster-Shafer evidence theory (DST) is proposed. Firstly, vibration signals of centrifugal fans are collected by multiple accelerometers deployed at different locations. Then, a two-branch feature fusion method based on the cross-attention mechanism is proposed to overcome the class imbalance due to the replay incremental learning method. After that, the fused features are fed into a Softmax classifier to complete the initial classification of blade status. Finally, a modified DST based on the cross-correlation energy is adopted for multi-sensor decision fusion to obtain the final blade crack detection results. The effectiveness of the proposed method is verified by two incremental blade crack datasets, and MFIDM achieves the better performance compared with other related incremental detection methods.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102952"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142659052","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.2024.102927
Xiaoke Huang , Chunjie Yang , Yuyan Zhang , Siwei Lou , Liyuan Kong , Heng Zhou
Due to the widespread presence of knowledge in factories, integrating various types of knowledge to solve different tasks in industrial production processes, including prediction, diagnosis, and control tasks, is of great significance and challenging. Knowledge graph, as a method of knowledge representation, holds significant promise for addressing challenges within industrial contexts. However, current research on knowledge graph has a limitation in that task related knowledge graph focus on structure information, while ignoring semantic and logical information in knowledge. Additionally, the existing ontologies designed for industrial production lack adaptability to cater to the diverse needs of different industrial tasks. This paper proposes a multi-level knowledge graph defined by ontology to introduce semantics and further explore the methods for combining semantics to complete practical industrial tasks. To ensure the accurate sampling of heterogeneous nodes, four semantic templates are generated using if-then rule logic. Different kinds of neighbor nodes are defined through the if-then rule logic, leading to a accelerated generation of target subgraphs related to different tasks. In this way, the plant-wide distributed computing for fault diagnosis tasks can be easily realized. Furthermore, this paper introduces a framework for semantics extraction and graph embedding based on multi-information fusion. This framework integrates semantic information, structural information, and node attribute information within the graph to deliver a holistic feature representation for prediction and control tasks. We take blast furnace ironmaking process as an industrial case study and the experimental results demonstrate the crucial role of semantics in enhancing the knowledge expression capability of graphs. Based on the blast furnace simulation experiment platform, the proposed method achieves 92.76% accuracy in the blast furnace fault diagnosis task, and the diagnosis time is reduced by 58.44% compared with the traditional rule-based method. In the self-healing control task of the blast furnace, the proposed graph embedding method can achieve a complete control process in three types of blast furnace faults: blowing out, tuyere failure, and low stockline. The control effect can be comparable to manual operation.
{"title":"Ontology guided multi-level knowledge graph construction and its applications in blast furnace ironmaking process","authors":"Xiaoke Huang , Chunjie Yang , Yuyan Zhang , Siwei Lou , Liyuan Kong , Heng Zhou","doi":"10.1016/j.aei.2024.102927","DOIUrl":"10.1016/j.aei.2024.102927","url":null,"abstract":"<div><div>Due to the widespread presence of knowledge in factories, integrating various types of knowledge to solve different tasks in industrial production processes, including prediction, diagnosis, and control tasks, is of great significance and challenging. Knowledge graph, as a method of knowledge representation, holds significant promise for addressing challenges within industrial contexts. However, current research on knowledge graph has a limitation in that task related knowledge graph focus on structure information, while ignoring semantic and logical information in knowledge. Additionally, the existing ontologies designed for industrial production lack adaptability to cater to the diverse needs of different industrial tasks. This paper proposes a multi-level knowledge graph defined by ontology to introduce semantics and further explore the methods for combining semantics to complete practical industrial tasks. To ensure the accurate sampling of heterogeneous nodes, four semantic templates are generated using if-then rule logic. Different kinds of neighbor nodes are defined through the if-then rule logic, leading to a accelerated generation of target subgraphs related to different tasks. In this way, the plant-wide distributed computing for fault diagnosis tasks can be easily realized. Furthermore, this paper introduces a framework for semantics extraction and graph embedding based on multi-information fusion. This framework integrates semantic information, structural information, and node attribute information within the graph to deliver a holistic feature representation for prediction and control tasks. We take blast furnace ironmaking process as an industrial case study and the experimental results demonstrate the crucial role of semantics in enhancing the knowledge expression capability of graphs. Based on the blast furnace simulation experiment platform, the proposed method achieves 92.76% accuracy in the blast furnace fault diagnosis task, and the diagnosis time is reduced by 58.44% compared with the traditional rule-based method. In the self-healing control task of the blast furnace, the proposed graph embedding method can achieve a complete control process in three types of blast furnace faults: blowing out, tuyere failure, and low stockline. The control effect can be comparable to manual operation.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102927"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142659113","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.2024.102916
Quan Lu, Wenju Ju, Linfei Yin
Accurate and rapid prediction of supercapacitors’ remaining useful life (RUL) and timely replacement of failing supercapacitors are of great importance to systemic stability and safety. To decrease the effects of the manual extraction of aging characteristics and fluctuations in the capacity data of supercapacitors for supercapacitor RUL prediction, a parallel GhostNet classification prediction method for supercapacitor RUL prediction is proposed. In this study, the mapping relationship between supercapacitor charging/discharging capacity data and RUL is established directly. In addition, the aging characteristics are learned from the raw observation data without relevant reserve knowledge. The supercapacitor RUL is quantified into 30 rank intervals and predicted by the parallel GhostNet classification method. The validation results based on 60 supercapacitors indicate that the prediction precision of the parallel GhostNet for supercapacitor RUL is 81.84 %, 21.28 % higher than that of a single GhostNet, 19.86 % higher than that of the Xeption model with the highest accuracy among other classical networks. Furthermore, introducing depth separable convolution, the prediction speed of the proposed parallel GhostNet model is 50576 s faster than that of the Xeption model.
{"title":"Parallel GhostNet classification prediction method for supercapacitor remaining useful life prediction","authors":"Quan Lu, Wenju Ju, Linfei Yin","doi":"10.1016/j.aei.2024.102916","DOIUrl":"10.1016/j.aei.2024.102916","url":null,"abstract":"<div><div>Accurate and rapid prediction of supercapacitors’ remaining useful life (RUL) and timely replacement of failing supercapacitors are of great importance to systemic stability and safety. To decrease the effects of the manual extraction of aging characteristics and fluctuations in the capacity data of supercapacitors for supercapacitor RUL prediction, a parallel GhostNet classification prediction method for supercapacitor RUL prediction is proposed. In this study, the mapping relationship between supercapacitor charging/discharging capacity data and RUL is established directly. In addition, the aging characteristics are learned from the raw observation data without relevant reserve knowledge. The supercapacitor RUL is quantified into 30 rank intervals and predicted by the parallel GhostNet classification method. The validation results based on 60 supercapacitors indicate that the prediction precision of the parallel GhostNet for supercapacitor RUL is 81.84 %, 21.28 % higher than that of a single GhostNet, 19.86 % higher than that of the Xeption model with the highest accuracy among other classical networks. Furthermore, introducing depth separable convolution, the prediction speed of the proposed parallel GhostNet model is 50576 s faster than that of the Xeption model.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102916"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142659192","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.2024.102831
Le Wu , Aiping Liu , Chang Li , Xun Chen
Electroencephalography (EEG) stands as one of the most vital noninvasive tools in neuroscience and clinical practice. Nevertheless, EEG data is highly susceptible to interference from various artifacts, which, in turn, can severely impact the subsequent analyses. As a result, the removal of these unwanted artifacts is of utmost importance. Recently, deep learning methods have demonstrated superior performance in artifact removal compared to traditional approaches. However, experts often invest substantial time and effort in identifying an efficient architecture, a process that is time-consuming and labor-intensive. In light of this challenge, this study introduces, for the first time, an artifact removal method based on neural network architecture search. This approach assigns probabilities to each potential operation within the network and optimizes the most suitable architecture based on the characteristics of the input data. Additionally, we expand the search space by incorporating large convolutional kernels, enabling the network to encompass a wider receptive field for the more effective capture of inherent EEG characteristics. The proposed method is evaluated on publicly available datasets, including electromyography (EMG), electrooculogram (EOG), electrocardiogram (ECG), and motion artifacts. Our results demonstrate that architectures incorporating convolution operations with varied kernel scales and shortcut connections are particularly effective for artifact removal. Notably, our method outperforms state-of-the-art techniques, achieving an average correlation coefficient (CC) exceeding 0.95, a relative root mean squared error (RRMSE) below 0.3, and a signal-to-noise ratio (SNR) above 12 dB. These findings underscore the potential of our method as a reliable and advanced technique for EEG denoising.
{"title":"Enhancing EEG artifact removal through neural architecture search with large kernels","authors":"Le Wu , Aiping Liu , Chang Li , Xun Chen","doi":"10.1016/j.aei.2024.102831","DOIUrl":"10.1016/j.aei.2024.102831","url":null,"abstract":"<div><div>Electroencephalography (EEG) stands as one of the most vital noninvasive tools in neuroscience and clinical practice. Nevertheless, EEG data is highly susceptible to interference from various artifacts, which, in turn, can severely impact the subsequent analyses. As a result, the removal of these unwanted artifacts is of utmost importance. Recently, deep learning methods have demonstrated superior performance in artifact removal compared to traditional approaches. However, experts often invest substantial time and effort in identifying an efficient architecture, a process that is time-consuming and labor-intensive. In light of this challenge, this study introduces, for the first time, an artifact removal method based on neural network architecture search. This approach assigns probabilities to each potential operation within the network and optimizes the most suitable architecture based on the characteristics of the input data. Additionally, we expand the search space by incorporating large convolutional kernels, enabling the network to encompass a wider receptive field for the more effective capture of inherent EEG characteristics. The proposed method is evaluated on publicly available datasets, including electromyography (EMG), electrooculogram (EOG), electrocardiogram (ECG), and motion artifacts. Our results demonstrate that architectures incorporating convolution operations with varied kernel scales and shortcut connections are particularly effective for artifact removal. Notably, our method outperforms state-of-the-art techniques, achieving an average correlation coefficient (CC) exceeding 0.95, a relative root mean squared error (RRMSE) below 0.3, and a signal-to-noise ratio (SNR) above 12 dB. These findings underscore the potential of our method as a reliable and advanced technique for EEG denoising.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102831"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142359377","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.2024.102847
Xiang Zhao , Sharul Azim Sharudin , Han-Lu Lv
The focus of consumer desire transitions from product functionality to emotional resonance in experience economy era, wherein emotional needs of users increasingly become a critical factor in product design. However, traditional approaches to product shape design often rely heavily on the designer’s intuition and experience, sometimes neglecting to incorporate emotional and humanistic elements into the product’s shape, thus resulting in inconsistencies in design results and quality. To address this challenge, this study introduces a novel method for emotionally driven product shape design that integrates Kansei engineering and the Whale Optimization Algorithm (WOA). This approach aims to fulfill consumer emotional demands related to product form and enhance overall user satisfaction. Firstly, the process utilized Python web crawlers to collect online product review texts and product images from e-commerce platforms. Next, Latent Dirichlet Allocation (LDA) and Analytic Hierarchy Process (AHP) were employed to extract representative emotional vocabularies, while representative samples were defined and deconstructed through clustering and morphological analysis. Then, semantic Differential (SD) questionnaires were distributed to collect consumer evaluations of product shape imagery, leading to the development of a user emotional prediction model for product shape. Then, WOA was introduced to optimize the performance of Back Propagation Neural Network (BPNN) and Support Vector Regression (SVR) models. Finally, Particle Swarm Optimization (PSO) and Seagull Optimization Algorithm (SOA) were employed to improve the prediction model, and the effect of these models was compared by the error method. This analysis explored the accuracy of these non-linear models in order to identify the optimal model for application in product form design cases. The scientific validity and effectiveness of this method were demonstrated utilizing whiskey bottle shape design as a case study. The results exhibit that under the WOA-BPNN model, the average error rates for four sets of perceptual words were 3.08%, 2.60%, 6.53%, and 5.70%, respectively. The WOA-based BPNN model outperformed other models in predictive ability, suggesting its utility as a valuable tool for designers during the front-end development stage of emotionally driven product form design.
{"title":"A novel product shape design method integrating Kansei engineering and whale optimization algorithm","authors":"Xiang Zhao , Sharul Azim Sharudin , Han-Lu Lv","doi":"10.1016/j.aei.2024.102847","DOIUrl":"10.1016/j.aei.2024.102847","url":null,"abstract":"<div><div>The focus of consumer desire transitions from product functionality to emotional resonance in experience economy era, wherein emotional needs of users increasingly become a critical factor in product design. However, traditional approaches to product shape design often rely heavily on the designer’s intuition and experience, sometimes neglecting to incorporate emotional and humanistic elements into the product’s shape, thus resulting in inconsistencies in design results and quality. To address this challenge, this study introduces a novel method for emotionally driven product shape design that integrates Kansei engineering and the Whale Optimization Algorithm (WOA). This approach aims to fulfill consumer emotional demands related to product form and enhance overall user satisfaction. Firstly, the process utilized Python web crawlers to collect online product review texts and product images from e-commerce platforms. Next, Latent Dirichlet Allocation (LDA) and Analytic Hierarchy Process (AHP) were employed to extract representative emotional vocabularies, while representative samples were defined and deconstructed through clustering and morphological analysis. Then, semantic Differential (SD) questionnaires were distributed to collect consumer evaluations of product shape imagery, leading to the development of a user emotional prediction model for product shape. Then, WOA was introduced to optimize the performance of Back Propagation Neural Network (BPNN) and Support Vector Regression (SVR) models. Finally, Particle Swarm Optimization (PSO) and Seagull Optimization Algorithm (SOA) were employed to improve the prediction model, and the effect of these models was compared by the error method. This analysis explored the accuracy of these non-linear models in order to identify the optimal model for application in product form design cases. The scientific validity and effectiveness of this method were demonstrated utilizing whiskey bottle shape design as a case study. The results exhibit that under the WOA-BPNN model, the average error rates for four sets of perceptual words were 3.08%, 2.60%, 6.53%, and 5.70%, respectively. The WOA-based BPNN model outperformed other models in predictive ability, suggesting its utility as a valuable tool for designers during the front-end development stage of emotionally driven product form design.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102847"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142417044","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.2024.102883
Genshen Liu , Peitang Wei , Xuesong Du , Siqi Liu , Li Luo , Rui Hu , Caichao Zhu , Jigui Zheng , Pengliang Zhou
The planetary roller screw mechanism (PRSM) faces an ever-increasing precision transmission demand in current advanced equipment. The relationship between machining errors and transmission accuracy remains elusive due to the over-simplified physical models and small-sample experimental datasets. This work proposes a physics-informed and data-driven hybrid strategy for PRSM transmission accuracy evaluation and tolerance optimization. In the physical model, a PRSM transmission accuracy model is developed to calculate transmission error that considers 16 machining errors in eccentric, nominal diameter, pitch, flank angle, and roller consistency. In the dataset establishment, thread profile measurements and dynamic leadscrew inspections are conducted for the machining error and transmission accuracy data acquisition. A data augmentation approach combining the physical model with the generative adversarial network is utilized to predict travel deviation, variations, and axial backlash and estimate machining error contribution with the small-sample experimental dataset. It is firstly found that the roller consistency of nominal diameter significantly affects PRSM travel variation V2π with a 17.3 % importance value. With the developed framework, the key tolerances for screw, roller, nut, and roller consistency are optimized toward a typical precision transmission requirement using the non-dominated sorting genetic algorithm. It also provides a tolerance grade recommendation table with PRSM transmission accuracy level in engineering practice.
{"title":"Physics-informed and data-driven hybrid method for transmission accuracy design optimization of planetary roller screw mechanism","authors":"Genshen Liu , Peitang Wei , Xuesong Du , Siqi Liu , Li Luo , Rui Hu , Caichao Zhu , Jigui Zheng , Pengliang Zhou","doi":"10.1016/j.aei.2024.102883","DOIUrl":"10.1016/j.aei.2024.102883","url":null,"abstract":"<div><div>The planetary roller screw mechanism (PRSM) faces an ever-increasing precision transmission demand in current advanced equipment. The relationship between machining errors and transmission accuracy remains elusive due to the over-simplified physical models and small-sample experimental datasets. This work proposes a physics-informed and data-driven hybrid strategy for PRSM transmission accuracy evaluation and tolerance optimization. In the physical model, a PRSM transmission accuracy model is developed to calculate transmission error that considers 16 machining errors in eccentric, nominal diameter, pitch, flank angle, and roller consistency. In the dataset establishment, thread profile measurements and dynamic leadscrew inspections are conducted for the machining error and transmission accuracy data acquisition. A data augmentation approach combining the physical model with the generative adversarial network is utilized to predict travel deviation, variations, and axial backlash and estimate machining error contribution with the small-sample experimental dataset. It is firstly found that the roller consistency of nominal diameter significantly affects PRSM travel variation <em>V</em><sub>2π</sub> with a 17.3 % importance value. With the developed framework, the key tolerances for screw, roller, nut, and roller consistency are optimized toward a typical precision transmission requirement using the non-dominated sorting genetic algorithm. It also provides a tolerance grade recommendation table with PRSM transmission accuracy level in engineering practice.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102883"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529670","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}
To provide technical support for the initial design of products, we propose an innovative text2shape-based technology for intelligent computational design, which can map engineering semantics to functional/structural/Kansei feature spaces to generate product shapes. New energy vehicles were selected as the application object of this technology, as there are many creative ideas for the outer contour design of new energy vehicles. Firstly, a dataset with 2900 + samples was built based on feature engineering (FE) and Kansei engineering (KE). Each sample contains the car’s outer contour shape’s functional, structural, and Kansei features. Secondly, we proposed an improved conditional Wasserstein generative adversarial network (CWGAN) model suitable to the dataset. The generator’s loss in the model is designed to evaluate the authenticity of the generated results, while the discriminator’s loss assesses the conditional matching of these results. Finally, in case studies, the trained CWGAN was compared with the conditional variational auto-encoder (C-VAE), diffusion, Wasserstein generative adversarial network with gradient penalty (WGAN-GP) and style generative adversarial network (StyleGAN) models, demonstrating superior performance.
{"title":"Text2shape: Intelligent computational design of car outer contour shapes based on improved conditional Wasserstein generative adversarial network","authors":"Tianshuo Zang, Maolin Yang, Yuhao Liu, Pingyu Jiang","doi":"10.1016/j.aei.2024.102892","DOIUrl":"10.1016/j.aei.2024.102892","url":null,"abstract":"<div><div>To provide technical support for the initial design of products, we propose an innovative text2shape-based technology for intelligent computational design, which can map engineering semantics to functional/structural/Kansei feature spaces to generate product shapes. New energy vehicles were selected as the application object of this technology, as there are many creative ideas for the outer contour design of new energy vehicles. Firstly, a dataset with 2900 + samples was built based on feature engineering (FE) and Kansei engineering (KE). Each sample contains the car’s outer contour shape’s functional, structural, and Kansei features. Secondly, we proposed an improved conditional Wasserstein generative adversarial network (CWGAN) model suitable to the dataset. The generator’s loss in the model is designed to evaluate the authenticity of the generated results, while the discriminator’s loss assesses the conditional matching of these results. Finally, in case studies, the trained CWGAN was compared with the conditional variational auto-encoder (C-VAE), diffusion, Wasserstein generative adversarial network with gradient penalty (WGAN-GP) and style generative adversarial network (StyleGAN) models, demonstrating superior performance.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102892"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142532089","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.2024.102893
Yidan Qiao , Haotian Li , Dengkai Chen , Hang Zhao , Lin Ma , Yao Wang
The complexity of human cognition is increased by the multiple and interactive information clusters brought about by the application of advanced intelligent information technologies. Especially in systems operating in extreme regions far from society, human errors are more pronounced than ever before. Prolonged social isolation, extreme weightless or overweight environments, stressful atmospheres, and lack of situational awareness are all added potential elements contributing to human risk. Although the development of human reliability analysis methods and their variants continues to mature, accurately predicting the potential risk of dynamic human behavior from sparse and discrete events remains a great challenge. We focus on deep learning computational architectures that are similar to the cognitive processes and mechanisms of the brain, and build neural networks that match the perceptual activation and memory cycling of the cognitive features of the brain. This study focuses on investigating the ability of the joint SNN-ITLSTM network to predict human error behavior and the clusters of performance shaping factors that effectively characterize the far-social nature. Combining the bionic properties of SNN and the temporal update mechanism of LSTM in the form of hierarchical events constitutes a computationally efficient network architecture. Our results show that the joint model proposed in this study has the performance to strengthen temporal influences and characterize cognitive features of the brain.
{"title":"Human risk recognition and prediction in manned submersible diving tasks driven by deep learning models","authors":"Yidan Qiao , Haotian Li , Dengkai Chen , Hang Zhao , Lin Ma , Yao Wang","doi":"10.1016/j.aei.2024.102893","DOIUrl":"10.1016/j.aei.2024.102893","url":null,"abstract":"<div><div>The complexity of human cognition is increased by the multiple and interactive information clusters brought about by the application of advanced intelligent information technologies. Especially in systems operating in extreme regions far from society, human errors are more pronounced than ever before. Prolonged social isolation, extreme weightless or overweight environments, stressful atmospheres, and lack of situational awareness are all added potential elements contributing to human risk. Although the development of human reliability analysis methods and their variants continues to mature, accurately predicting the potential risk of dynamic human behavior from sparse and discrete events remains a great challenge. We focus on deep learning computational architectures that are similar to the cognitive processes and mechanisms of the brain, and build neural networks that match the perceptual activation and memory cycling of the cognitive features of the brain. This study focuses on investigating the ability of the joint SNN-ITLSTM network to predict human error behavior and the clusters of performance shaping factors that effectively characterize the far-social nature. Combining the bionic properties of SNN and the temporal update mechanism of LSTM in the form of hierarchical events constitutes a computationally efficient network architecture. Our results show that the joint model proposed in this study has the performance to strengthen temporal influences and characterize cognitive features of the brain.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102893"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142532092","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.2024.102884
David F. Bucher , Jens J. Hunhevicz , Ranjith K. Soman , Pieter Pauwels , Daniel M. Hall
The field of construction informatics is fragmented and lacks clarity in understanding the interconnection of different data management strategies. This makes it challenging to address industry-specific data management issues. Using a critical interpretive synthesis, this study reviews and integrates both present and emerging data management approaches in construction informatics. The review is meant to be comprehensive, encompassing technologies and concepts such as Open Schema, Information Container, Common Data Environments, Linked Data, as well as cutting-edge Web3 technologies such as blockchain and decentralized data protocols. The different approaches are identified and classified into five categories and mapped into a two-dimensional framework that considers data storage and data processing modes. The systematic categorization provides a simple, but comprehensive understanding of data management strategies in construction informatics. Moreover, the framework allows to identify the state of the art and trends of data management approaches, providing guidance for future research perspectives, especially in the intersection with Web3 technologies.
{"title":"From BIM to Web3: A critical interpretive synthesis of present and emerging data management approaches in construction informatics","authors":"David F. Bucher , Jens J. Hunhevicz , Ranjith K. Soman , Pieter Pauwels , Daniel M. Hall","doi":"10.1016/j.aei.2024.102884","DOIUrl":"10.1016/j.aei.2024.102884","url":null,"abstract":"<div><div>The field of construction informatics is fragmented and lacks clarity in understanding the interconnection of different data management strategies. This makes it challenging to address industry-specific data management issues. Using a critical interpretive synthesis, this study reviews and integrates both present and emerging data management approaches in construction informatics. The review is meant to be comprehensive, encompassing technologies and concepts such as Open Schema, Information Container, Common Data Environments, Linked Data, as well as cutting-edge Web3 technologies such as blockchain and decentralized data protocols. The different approaches are identified and classified into five categories and mapped into a two-dimensional framework that considers data storage and data processing modes. The systematic categorization provides a simple, but comprehensive understanding of data management strategies in construction informatics. Moreover, the framework allows to identify the state of the art and trends of data management approaches, providing guidance for future research perspectives, especially in the intersection with Web3 technologies.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102884"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142552951","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.2024.102902
Zesen Wang, Yonggang Li, Chunhua Yang, Hongqiu Zhu, Can Zhou
Process monitoring is a key technology in the field of industrial production and manufacturing, where machine learning algorithms play a crucial role. However, the cost of data collection in industrial settings is very high, which seriously limits the performance improvement of monitoring models. To address this issue, a graph-based active semi-supervised learning (GASSL) strategy is proposed, which can derive reliable monitoring models with limited labeling costs. Specifically, first, a robust unsupervised active learning (RUAL) method is proposed, which incorporates data reconstruction, low-rank representation, and manifold learning into a unified framework to select the most representative samples for labeling, avoiding the poor performance of model-based active learning algorithms under the condition of limited initial sample size. Second, to maximize the use of the remaining unlabeled samples after labeling, pseudo-labels are assigned to the unlabeled samples through label propagation, thereby further expanding the sample set. At the same time, active learning selects the most valuable samples as the labeled node set of the graph model, strengthening the performance of label propagation. Experimental results on three datasets related to water quality monitoring, including public dataset, simulation dataset, and real total nitrogen detection dataset, extensively demonstrate the effectiveness of the proposed method.
{"title":"Graph-based active semi-supervised learning: Case study in water quality monitoring","authors":"Zesen Wang, Yonggang Li, Chunhua Yang, Hongqiu Zhu, Can Zhou","doi":"10.1016/j.aei.2024.102902","DOIUrl":"10.1016/j.aei.2024.102902","url":null,"abstract":"<div><div>Process monitoring is a key technology in the field of industrial production and manufacturing, where machine learning algorithms play a crucial role. However, the cost of data collection in industrial settings is very high, which seriously limits the performance improvement of monitoring models. To address this issue, a graph-based active semi-supervised learning (GASSL) strategy is proposed, which can derive reliable monitoring models with limited labeling costs. Specifically, first, a robust unsupervised active learning (RUAL) method is proposed, which incorporates data reconstruction, low-rank representation, and manifold learning into a unified framework to select the most representative samples for labeling, avoiding the poor performance of model-based active learning algorithms under the condition of limited initial sample size. Second, to maximize the use of the remaining unlabeled samples after labeling, pseudo-labels are assigned to the unlabeled samples through label propagation, thereby further expanding the sample set. At the same time, active learning selects the most valuable samples as the labeled node set of the graph model, strengthening the performance of label propagation. Experimental results on three datasets related to water quality monitoring, including public dataset, simulation dataset, and real total nitrogen detection dataset, extensively demonstrate the effectiveness of the proposed method.</div></div>","PeriodicalId":50941,"journal":{"name":"Advanced Engineering Informatics","volume":"62 ","pages":"Article 102902"},"PeriodicalIF":8.0,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142552956","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