This paper proposes a method for topology optimization of three-dimensional braced steel frames under static seismic loads, aiming to minimize structural volume while satisfying response constraints. The method consists of a proposed hierarchical graph-based machine learning model that sequentially embeds graph representations of structural elements to create a comprehensive lumped mass model suitable for seismic analysis and optimization of building frames. The proposed model is utilized as a reinforcement learning agent, a class of machine learning, to observe the current building frame configurations, modify the frame to improve its performance, and adjust the model parameters using the obtained reward function. Numerical results demonstrate that the proposed graph-based model can improve its performance better than benchmark of a graph neural network utilized in previous research. When applied to three cases of unseen large three-dimensional building frames, the trained agent with the proposed model outperforms the genetic algorithm and simulated annealing in the optimization task with 11–46% lower objective function while utilizing the same computational cost. The generality, performance, and computational efficiency of the agent indicate its applicability to applied to the optimization of three-dimensional frames.
{"title":"Hierarchical graph-based machine learning model for optimization of three-dimensional braced steel frame","authors":"Chi-tathon Kupwiwat, Kazuki Hayashi, Makoto Ohsaki","doi":"10.1016/j.engappai.2025.110356","DOIUrl":"10.1016/j.engappai.2025.110356","url":null,"abstract":"<div><div>This paper proposes a method for topology optimization of three-dimensional braced steel frames under static seismic loads, aiming to minimize structural volume while satisfying response constraints. The method consists of a proposed hierarchical graph-based machine learning model that sequentially embeds graph representations of structural elements to create a comprehensive lumped mass model suitable for seismic analysis and optimization of building frames. The proposed model is utilized as a reinforcement learning agent, a class of machine learning, to observe the current building frame configurations, modify the frame to improve its performance, and adjust the model parameters using the obtained reward function. Numerical results demonstrate that the proposed graph-based model can improve its performance better than benchmark of a graph neural network utilized in previous research. When applied to three cases of unseen large three-dimensional building frames, the trained agent with the proposed model outperforms the genetic algorithm and simulated annealing in the optimization task with 11–46% lower objective function while utilizing the same computational cost. The generality, performance, and computational efficiency of the agent indicate its applicability to applied to the optimization of three-dimensional frames.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"147 ","pages":"Article 110356"},"PeriodicalIF":7.5,"publicationDate":"2025-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143454943","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-21DOI: 10.1016/j.engappai.2025.110291
Subhankar Jana , Anjali Patel , Juthika Mahanta
This article introduces an improved definition of Pythagorean fuzzy quasi-coincidence between the Pythagorean fuzzy sets enhancing the theoretical foundation of previous approaches. We investigate the theoretical aspects of the introduced Pythagorean fuzzy quasi-coincidence and corresponding Pythagorean fuzzy quasi-coincident set. The Pythagorean fuzzy quasi-coincident set is proven to be a Pythagorean fuzzy t-norm and the corresponding t-conorm with respect to the standard negation operator has been obtained. The introduced concept can reveal interrelationships among Pythagorean fuzzy sets, overcoming limitations of the fuzzy version of quasi-coincidence in scenarios involving uncertainty and hesitancy. Additionally, it enables range divisions, a feature previously unattainable in fuzzy quasi-coincidence methods. A significant novelty lies in the development of a Pythagorean fuzzy generator, the first of its kind, to generate Pythagorean fuzzy data from conventional fuzzy or Intuitionistic fuzzy information. This generator, accompanied by a generator sequence, allows customizable non-membership values based on user requirements. These advancements facilitate applications such as identifying high-risk zones during pandemics or natural disasters. Building on this foundation, we present practical applications to identify high-risk areas within regions affected by outbreaks of pandemics. We use he quasi-coincidence property, and group the affected area into distinct categories, such as red and yellow zones. Further, the application of the proposed theories is also depicted in medical diagnosis and we show that the method can also be used to re-frame traditional multi-criteria decision-making processes.
{"title":"Pythagorean fuzzy quasi coincidence: Analysis and applications","authors":"Subhankar Jana , Anjali Patel , Juthika Mahanta","doi":"10.1016/j.engappai.2025.110291","DOIUrl":"10.1016/j.engappai.2025.110291","url":null,"abstract":"<div><div>This article introduces an improved definition of Pythagorean fuzzy quasi-coincidence between the Pythagorean fuzzy sets enhancing the theoretical foundation of previous approaches. We investigate the theoretical aspects of the introduced Pythagorean fuzzy quasi-coincidence and corresponding Pythagorean fuzzy quasi-coincident set. The Pythagorean fuzzy quasi-coincident set is proven to be a Pythagorean fuzzy t-norm and the corresponding t-conorm with respect to the standard negation operator has been obtained. The introduced concept can reveal interrelationships among Pythagorean fuzzy sets, overcoming limitations of the fuzzy version of quasi-coincidence in scenarios involving uncertainty and hesitancy. Additionally, it enables range divisions, a feature previously unattainable in fuzzy quasi-coincidence methods. A significant novelty lies in the development of a Pythagorean fuzzy generator, the first of its kind, to generate Pythagorean fuzzy data from conventional fuzzy or Intuitionistic fuzzy information. This generator, accompanied by a generator sequence, allows customizable non-membership values based on user requirements. These advancements facilitate applications such as identifying high-risk zones during pandemics or natural disasters. Building on this foundation, we present practical applications to identify high-risk areas within regions affected by outbreaks of pandemics. We use he quasi-coincidence property, and group the affected area into distinct categories, such as red and yellow zones. Further, the application of the proposed theories is also depicted in medical diagnosis and we show that the method can also be used to re-frame traditional multi-criteria decision-making processes.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"147 ","pages":"Article 110291"},"PeriodicalIF":7.5,"publicationDate":"2025-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143454947","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110210
Aldo Ribeiro de Carvalho , Romário Parreira Pita , Thaís Mayra de Oliveira , Guilherme Jorge Brigolini Silva , Julia Castro Mendes
The goal of this work is to present and analyze studies that link artificial intelligence (AI) and geopolymer composites through an exploratory literature review. The systematic search comprised articles in the Scopus database, from 2000 to 2023. The results from 48 articles show that AI has been applied mainly to predict compressive strength and slump. Comparing the AI techniques to their R2, no clear trend was observed, i.e., there was no overall best algorithm; and their performance was not directly related to database size. Still, Random Forest obtained superior results in articles that compared multiple techniques. Some of the main gaps identified are the lack of studies with geopolymers synthesized in acidic environments or with waste-based or nanomaterials; and that AI has not been largely applied to properties besides compressive strength (e.g., fire resistance, durability, porosity, and thermal properties). Regarding transparency, authors have seldom publicized their codes, impairing proper review of the models. In conclusion, if properly and ethically adopted, AI is a promising tool for the development of geopolymer composites, which can optimize time and employee resources; and many applications remain unexplored.
{"title":"Exploratory literature review and scientometric analysis of artificial intelligence applied to geopolymeric materials","authors":"Aldo Ribeiro de Carvalho , Romário Parreira Pita , Thaís Mayra de Oliveira , Guilherme Jorge Brigolini Silva , Julia Castro Mendes","doi":"10.1016/j.engappai.2025.110210","DOIUrl":"10.1016/j.engappai.2025.110210","url":null,"abstract":"<div><div>The goal of this work is to present and analyze studies that link artificial intelligence (AI) and geopolymer composites through an exploratory literature review. The systematic search comprised articles in the Scopus database, from 2000 to 2023. The results from 48 articles show that AI has been applied mainly to predict compressive strength and slump. Comparing the AI techniques to their R<sup>2</sup>, no clear trend was observed, i.e., there was no overall best algorithm; and their performance was not directly related to database size. Still, Random Forest obtained superior results in articles that compared multiple techniques. Some of the main gaps identified are the lack of studies with geopolymers synthesized in acidic environments or with waste-based or nanomaterials; and that AI has not been largely applied to properties besides compressive strength (e.g., fire resistance, durability, porosity, and thermal properties). Regarding transparency, authors have seldom publicized their codes, impairing proper review of the models. In conclusion, if properly and ethically adopted, AI is a promising tool for the development of geopolymer composites, which can optimize time and employee resources; and many applications remain unexplored.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"145 ","pages":"Article 110210"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143445248","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110331
Mei Hua , Ke Yan , Xin Li
Despite the great successes of supervised learning-based fault diagnosis techniques for heating, ventilation and air-conditioning (HVAC) systems, their applications are severely limited due to insufficient labeled data accompanied with massive unlabeled data. To address this drawback, a Transformer-based self-supervised representation learning model (TSSRL) is proposed in this study for HVAC fault diagnosis with limited labeled data. Specifically, a customized Transformer model is developed as the feature encoder by embedding a context-attention module on the self-attention module, which enables TSSRL to mine the contextual representations among input data. In addition, a joint data augmentation strategy is designed to improve the diversity of inputs, promoting the pretext tasks to learn more extensive representations from unlabeled data. Meanwhile, two cooperative pretext tasks, namely contrastive similarity matching and data reconstruction, are formulated to extract discriminative representations from unlabeled data. The diagnosis-beneficial representations learned from unlabeled data are used for downstream classification modeling tasks with limited labeled data. Experiments on two benchmark HVAC fault datasets demonstrate the superiority of the proposed TSSRL model over other state-of-the-art HVAC fault diagnosis methods.
{"title":"A Transformer-based self-supervised learning model for fault diagnosis of air-conditioning systems with limited labeled data","authors":"Mei Hua , Ke Yan , Xin Li","doi":"10.1016/j.engappai.2025.110331","DOIUrl":"10.1016/j.engappai.2025.110331","url":null,"abstract":"<div><div>Despite the great successes of supervised learning-based fault diagnosis techniques for heating, ventilation and air-conditioning (HVAC) systems, their applications are severely limited due to insufficient labeled data accompanied with massive unlabeled data. To address this drawback, a Transformer-based self-supervised representation learning model (TSSRL) is proposed in this study for HVAC fault diagnosis with limited labeled data. Specifically, a customized Transformer model is developed as the feature encoder by embedding a context-attention module on the self-attention module, which enables TSSRL to mine the contextual representations among input data. In addition, a joint data augmentation strategy is designed to improve the diversity of inputs, promoting the pretext tasks to learn more extensive representations from unlabeled data. Meanwhile, two cooperative pretext tasks, namely contrastive similarity matching and data reconstruction, are formulated to extract discriminative representations from unlabeled data. The diagnosis-beneficial representations learned from unlabeled data are used for downstream classification modeling tasks with limited labeled data. Experiments on two benchmark HVAC fault datasets demonstrate the superiority of the proposed TSSRL model over other state-of-the-art HVAC fault diagnosis methods.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"146 ","pages":"Article 110331"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143445100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110298
Jinrui Han , Zhen Chen , Di Zhou , Bing Hu , Tangbin Xia , Ershun Pan
As automated labeling on products in intelligent manufacturing grows in importance, detecting anomalies in the end-effectors used for industrial robots labeling is essential for maintaining production line stability and efficiency. Considering the distinct characteristics of specific movements in the labeling process, different motions, such as moving, labeling and rolling, contribute different effects to end-effector abnormalities. It's challenging to distinguish between normal and anomalies, instead of treating all motions as a homogeneous whole. Also, real-world industrial scenarios often lack the sufficient data on abnormal states, and resource constraints limit computation efficiency. In view of this, this paper aims to develop a task-specific anomaly detection solution tailored to the distinct motions of industrial robots labeling. To achieve this goal, an unsupervised, motion-based anomaly detection framework is proposed. The raw sensor signals from each motion are segmented and a group of encoder networks are employed to extract latent representations for each motion. Then, these motions are modeled as nodes in a graph, where a feature fusion module based on a Graph Attention Network (GAT) captures the interrelationships between them. A memory-augmented reconstruction module with multi-scale skip connections enhances the model's ability to detect anomalies. Finally, an anomaly detection module identifies abnormal states of the end-effector. Experimental validations are conducted on a dataset from real-world steel coil labeling task. The results show that the proposed framework can achieve an average performance of 98.24% with an inference time of 15 ms, also demonstrating the effectiveness of its structural design and key modules.
{"title":"Unsupervised motion-based anomaly detection with graph attention networks for industrial robots labeling","authors":"Jinrui Han , Zhen Chen , Di Zhou , Bing Hu , Tangbin Xia , Ershun Pan","doi":"10.1016/j.engappai.2025.110298","DOIUrl":"10.1016/j.engappai.2025.110298","url":null,"abstract":"<div><div>As automated labeling on products in intelligent manufacturing grows in importance, detecting anomalies in the end-effectors used for industrial robots labeling is essential for maintaining production line stability and efficiency. Considering the distinct characteristics of specific movements in the labeling process, different motions, such as moving, labeling and rolling, contribute different effects to end-effector abnormalities. It's challenging to distinguish between normal and anomalies, instead of treating all motions as a homogeneous whole. Also, real-world industrial scenarios often lack the sufficient data on abnormal states, and resource constraints limit computation efficiency. In view of this, this paper aims to develop a task-specific anomaly detection solution tailored to the distinct motions of industrial robots labeling. To achieve this goal, an unsupervised, motion-based anomaly detection framework is proposed. The raw sensor signals from each motion are segmented and a group of encoder networks are employed to extract latent representations for each motion. Then, these motions are modeled as nodes in a graph, where a feature fusion module based on a Graph Attention Network (GAT) captures the interrelationships between them. A memory-augmented reconstruction module with multi-scale skip connections enhances the model's ability to detect anomalies. Finally, an anomaly detection module identifies abnormal states of the end-effector. Experimental validations are conducted on a dataset from real-world steel coil labeling task. The results show that the proposed framework can achieve an average performance of 98.24% with an inference time of 15 ms, also demonstrating the effectiveness of its structural design and key modules.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"146 ","pages":"Article 110298"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143445096","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110234
Nabeela Anwar , Aqsa Saddiq , Muhammad Asif Zahoor Raja , Iftikhar Ahmad , Muhammad Shoaib , Adiqa Kausar Kiani
Computer viruses are significant from the perspective of reliable computer security for control infrastructure development that connotes the essential requirements for understanding virus spread and growth. This study investigates a delayed epidemiological computer virus model by applying artificial intelligence-inspired computing to analyze the behavior of a multi-delay differential system, incorporating the influence of the latent period on the dynamics of susceptible, infected, quarantined, and recovered computers using an adaptive autoregressive exogenous neural structure with Levenberg-Marquardt backpropagation. The reference data for executing networks is generated using the Adams numerical solver, incorporating parameters such as the computer infection rate, the saturation coefficient for calculating inhibitory effects, the immune loss rate of recovered computers, the number of infected computers in quarantine, the recovery rate, the mortality rate for each differential class, and delays associated with temporary immunity and the incubation period. The designed networks operate on generated synthetic data that are randomly distributed for testing, validation, and training samples to determine the approximate response of the nonlinear delay computer virus systems. The predictive solutions consistently aligned with the reference numerical outcomes indicating an error with negligible magnitude. The accuracy, stability, convergence, and efficiency of the designed intelligent networks are established on a thorough investigation by employing a variety of assessment metrics in terms of mean square error-based convergence curves, time series analysis of predicted outcomes, control parameter adaptation, error frequency distribution analysis, and exhaustive statistics on input-output and cross-correlations.
{"title":"Novel machine intelligent expedition with adaptive autoregressive exogenous neural structure for nonlinear multi-delay differential systems in computer virus propagation","authors":"Nabeela Anwar , Aqsa Saddiq , Muhammad Asif Zahoor Raja , Iftikhar Ahmad , Muhammad Shoaib , Adiqa Kausar Kiani","doi":"10.1016/j.engappai.2025.110234","DOIUrl":"10.1016/j.engappai.2025.110234","url":null,"abstract":"<div><div>Computer viruses are significant from the perspective of reliable computer security for control infrastructure development that connotes the essential requirements for understanding virus spread and growth. This study investigates a delayed epidemiological computer virus model by applying artificial intelligence-inspired computing to analyze the behavior of a multi-delay differential system, incorporating the influence of the latent period on the dynamics of susceptible, infected, quarantined, and recovered computers using an adaptive autoregressive exogenous neural structure with Levenberg-Marquardt backpropagation. The reference data for executing networks is generated using the Adams numerical solver, incorporating parameters such as the computer infection rate, the saturation coefficient for calculating inhibitory effects, the immune loss rate of recovered computers, the number of infected computers in quarantine, the recovery rate, the mortality rate for each differential class, and delays associated with temporary immunity and the incubation period. The designed networks operate on generated synthetic data that are randomly distributed for testing, validation, and training samples to determine the approximate response of the nonlinear delay computer virus systems. The predictive solutions consistently aligned with the reference numerical outcomes indicating an error with negligible magnitude. The accuracy, stability, convergence, and efficiency of the designed intelligent networks are established on a thorough investigation by employing a variety of assessment metrics in terms of mean square error-based convergence curves, time series analysis of predicted outcomes, control parameter adaptation, error frequency distribution analysis, and exhaustive statistics on input-output and cross-correlations.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"146 ","pages":"Article 110234"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143445095","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110352
Yizhi Cao, Zijian Tian, Zhaoran Liu, Naizheng Jia, Xinggao Liu
In multi-agent reinforcement learning, establishing effective communication protocols is crucial for enhancing agent collaboration. However, traditional communication methods face challenges in scalability and efficiency as the number of agents increases, due to the expansion in the dimensions of observation and action spaces. This leads to heightened resource consumption and degrades performance in large multi-agent scenarios. To address these issues, we introduce a novel Attentional Multi-agent Twin Delayed Deep Deterministic Policy Gradient (AMATD3) algorithm that incorporates an attentional communication policy gradient approach. This approach selectively initiates communications through an attention unit that assesses the necessity of information exchange among agents, combined with a communication module that effectively integrates essential information. By implementing a double-Q function, AMATD3 further addresses issues of overestimation and suboptimal policy choices in existing methods, enhancing the algorithm's accuracy and reducing communication overheads. Specifically, our algorithm demonstrates superior performance in the StarCraft II environment by achieving higher cumulative rewards and enhancing task success rates compared to existing algorithms. For example, AMATD3 yields reward values of 16.908 and 6.858 for the 8m and 25m scenarios, respectively, which is more than double the reward achieved by other methods. This confirms the algorithm's enhanced efficiency and effectiveness in complex multi-agent settings, contributing to the ongoing development of scalable and efficient communication protocols in artificial intelligence.
{"title":"Reducing overestimation with attentional multi-agent twin delayed deep deterministic policy gradient","authors":"Yizhi Cao, Zijian Tian, Zhaoran Liu, Naizheng Jia, Xinggao Liu","doi":"10.1016/j.engappai.2025.110352","DOIUrl":"10.1016/j.engappai.2025.110352","url":null,"abstract":"<div><div>In multi-agent reinforcement learning, establishing effective communication protocols is crucial for enhancing agent collaboration. However, traditional communication methods face challenges in scalability and efficiency as the number of agents increases, due to the expansion in the dimensions of observation and action spaces. This leads to heightened resource consumption and degrades performance in large multi-agent scenarios. To address these issues, we introduce a novel Attentional Multi-agent Twin Delayed Deep Deterministic Policy Gradient (AMATD3) algorithm that incorporates an attentional communication policy gradient approach. This approach selectively initiates communications through an attention unit that assesses the necessity of information exchange among agents, combined with a communication module that effectively integrates essential information. By implementing a double-Q function, AMATD3 further addresses issues of overestimation and suboptimal policy choices in existing methods, enhancing the algorithm's accuracy and reducing communication overheads. Specifically, our algorithm demonstrates superior performance in the StarCraft II environment by achieving higher cumulative rewards and enhancing task success rates compared to existing algorithms. For example, AMATD3 yields reward values of 16.908 and 6.858 for the 8m and 25m scenarios, respectively, which is more than double the reward achieved by other methods. This confirms the algorithm's enhanced efficiency and effectiveness in complex multi-agent settings, contributing to the ongoing development of scalable and efficient communication protocols in artificial intelligence.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"146 ","pages":"Article 110352"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143453750","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110311
Yu Zhang , Ping Tu , Zhiyuan Zhao , Xuan-Yan Chen
Artificial intelligence (AI) has played a key role in advancing autonomous navigation for unmanned ships, where ship trajectory prediction is crucial for ensuring maritime safety. As the shipping industry grows and the number of ships increases, especially with autonomous ships operating in complex environments, collision risks have become a major concern. Accurate trajectory prediction, supported by advanced AI techniques, is crucial for the safe operation of these ships. While current models predict ship trajectories with high precision using Automatic Identification System (AIS) data, they often fail to incorporate prior knowledge of collision risks and struggle to model ship interactions that could lead to collisions. To overcome these limitations, the DGCN-Transformer (Dynamic Graph Convolution Network-Transformer) model is proposed. This model enhances the accuracy and reliability of ship trajectory predictions by incorporating collision risk modeling into the prediction framework. It uses the Quaternion Ship Domain (QSD) to model potential collision scenarios, integrating an advanced understanding of ships' spatial and kinematic properties. The model integrates QSD-based prior knowledge into an advanced Graph Convolutional Network (GCN) for spatial modeling, while the Transformer component captures and analyzes temporal features, overcoming the limitations of traditional Long Short-Term Memory (LSTM) networks. Experiments with AIS data from Tianjin, Caofeidian, and Chengshanjiao ports demonstrate that the DGCN-Transformer model outperforms state-of-the-art models, significantly improving trajectory prediction accuracy. Specifically, at Tianjin Port, the DGCN-Transformer model reduces Final Displacement Error (FDE) by 36.1%, Maximum Displacement Error (MDE) by 15.4%, and Average Displacement Error (ADE) by 50% compared to the best baseline model, highlighting the model's effectiveness in enhancing the safety of autonomous ship navigation.
{"title":"Incorporating prior knowledge of collision risk into deep learning networks for ship trajectory prediction in the maritime Internet of Things industry","authors":"Yu Zhang , Ping Tu , Zhiyuan Zhao , Xuan-Yan Chen","doi":"10.1016/j.engappai.2025.110311","DOIUrl":"10.1016/j.engappai.2025.110311","url":null,"abstract":"<div><div>Artificial intelligence (AI) has played a key role in advancing autonomous navigation for unmanned ships, where ship trajectory prediction is crucial for ensuring maritime safety. As the shipping industry grows and the number of ships increases, especially with autonomous ships operating in complex environments, collision risks have become a major concern. Accurate trajectory prediction, supported by advanced AI techniques, is crucial for the safe operation of these ships. While current models predict ship trajectories with high precision using Automatic Identification System (AIS) data, they often fail to incorporate prior knowledge of collision risks and struggle to model ship interactions that could lead to collisions. To overcome these limitations, the DGCN-Transformer (Dynamic Graph Convolution Network-Transformer) model is proposed. This model enhances the accuracy and reliability of ship trajectory predictions by incorporating collision risk modeling into the prediction framework. It uses the Quaternion Ship Domain (QSD) to model potential collision scenarios, integrating an advanced understanding of ships' spatial and kinematic properties. The model integrates QSD-based prior knowledge into an advanced Graph Convolutional Network (GCN) for spatial modeling, while the Transformer component captures and analyzes temporal features, overcoming the limitations of traditional Long Short-Term Memory (LSTM) networks. Experiments with AIS data from Tianjin, Caofeidian, and Chengshanjiao ports demonstrate that the DGCN-Transformer model outperforms state-of-the-art models, significantly improving trajectory prediction accuracy. Specifically, at Tianjin Port, the DGCN-Transformer model reduces Final Displacement Error (FDE) by 36.1%, Maximum Displacement Error (MDE) by 15.4%, and Average Displacement Error (ADE) by 50% compared to the best baseline model, highlighting the model's effectiveness in enhancing the safety of autonomous ship navigation.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"146 ","pages":"Article 110311"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143453230","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110235
Mary Selvan , K. Ramar
Background
Character Segmentation plays a crucial role in analyzing the intricate script inscribed on these historical documents. Due to the unique script style and the physical characteristics of the palm leaves, character segmentation is considered to be challenging. Aim: An efficient character segmentation approach from palm-leaf manuscripts is developed in this work.
Method
ology: Initially, the images of Tamil palm leaf manuscripts are gathered manually and pre-processed with optimal binary thresholding and morphological operation. Then, the pre-processed images are utilized for Line segmentation by the Projection Profile method. The line-segmented images are fed into the feature extraction process. These extracted features are subjected to character segmentation by developing an Ensemble Machine Learning Structure (EMLS). EMLS is motivated by Bayesian Learning (BL), Support Vector Machine (SVM), and Artificial Neural Network (ANN). Moreover, parameter optimization is performed using the Hybridization of Sandpiper with Fireworks Algorithm (HSFA). Therefore, the developed method is suitable for real-time applications like mobile document scanning and educational-based applications.
Result
The experimental analysis is made to declare the efficiency of the developed approach, and given the accuracy to be 95.53%.
Conclusion
The developed model provides the character-segmented outcome and offers a promising tool for palm-leaf character recognition.
{"title":"Development of optimized ensemble machine learning-based character segmentation framework for ancient Tamil palm leaf manuscripts","authors":"Mary Selvan , K. Ramar","doi":"10.1016/j.engappai.2025.110235","DOIUrl":"10.1016/j.engappai.2025.110235","url":null,"abstract":"<div><h3>Background</h3><div>Character Segmentation plays a crucial role in analyzing the intricate script inscribed on these historical documents. Due to the unique script style and the physical characteristics of the palm leaves, character segmentation is considered to be challenging. <strong>Aim</strong>: An efficient character segmentation approach from palm-leaf manuscripts is developed in this work.</div></div><div><h3>Method</h3><div>ology: Initially, the images of Tamil palm leaf manuscripts are gathered manually and pre-processed with optimal binary thresholding and morphological operation. Then, the pre-processed images are utilized for Line segmentation by the Projection Profile method. The line-segmented images are fed into the feature extraction process. These extracted features are subjected to character segmentation by developing an Ensemble Machine Learning Structure (EMLS). EMLS is motivated by Bayesian Learning (BL), Support Vector Machine (SVM), and Artificial Neural Network (ANN). Moreover, parameter optimization is performed using the Hybridization of Sandpiper with Fireworks Algorithm (HSFA). Therefore, the developed method is suitable for real-time applications like mobile document scanning and educational-based applications.</div></div><div><h3>Result</h3><div>The experimental analysis is made to declare the efficiency of the developed approach, and given the accuracy to be 95.53%.</div></div><div><h3>Conclusion</h3><div>The developed model provides the character-segmented outcome and offers a promising tool for palm-leaf character recognition.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"146 ","pages":"Article 110235"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143445094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1016/j.engappai.2025.110308
Yichang Shao , Yueru Xu , Zhirui Ye , Yuhan Zhang , Weijie Chen , Nirajan Shiwakoti , Xiaomeng Shi
In recent years, Artificial Intelligence (AI) has significantly enhanced road safety, with Explainable Artificial Intelligence (XAI) providing essential transparency and trust. Our research utilizes AI to improve Advanced Driving Assistance Systems (ADAS) by investigating the gap in Forward Collision Warning (FCW): the impact of previous negative warnings (false and nuisance warnings) on drivers’ response times to subsequent accurate FCWs. By integrating XAI methods, we offer insights into the factors affecting driver behavior and system trust. Utilizing extensive dataset that encompasses various driving scenarios and driver behaviors, we constructed a gradient-boosting machine model to forecast driver response times. To explain the underlying mechanics of the model, the Shapley Additive Explanations (SHAP) framework was employed, enabling a comprehensive interpretation of feature importance and inter-feature interactions. Key findings reveal that increased speeds heighten driver responsiveness due to amplified alertness, whereas slower speeds lead to delayed reactions. The influence of previous negative warnings, significantly extends response times to accurate warnings. Additionally, older drivers require longer response times. The relationship between the driving period and previous warning judgment profoundly affects subsequent driver responsiveness, indicating trust dynamics with FCW systems. By using interpretable machine learning, we provide insights into ADAS functionality, suggesting pathways for FCW responsiveness and contributing to the field of XAI applications. In the validation experiment, our approach improved driver response times, reducing the average time from 2.1 s to 1.6 s. The proportion of ignored warnings decreased from 12% to 6%, and the driver acceptance rate increased from 59% to 71%.
{"title":"Understanding the impacts of negative advanced driving assistance system warnings on hazardous materials truck drivers’ responses using interpretable machine learning","authors":"Yichang Shao , Yueru Xu , Zhirui Ye , Yuhan Zhang , Weijie Chen , Nirajan Shiwakoti , Xiaomeng Shi","doi":"10.1016/j.engappai.2025.110308","DOIUrl":"10.1016/j.engappai.2025.110308","url":null,"abstract":"<div><div>In recent years, Artificial Intelligence (AI) has significantly enhanced road safety, with Explainable Artificial Intelligence (XAI) providing essential transparency and trust. Our research utilizes AI to improve Advanced Driving Assistance Systems (ADAS) by investigating the gap in Forward Collision Warning (FCW): the impact of previous negative warnings (false and nuisance warnings) on drivers’ response times to subsequent accurate FCWs. By integrating XAI methods, we offer insights into the factors affecting driver behavior and system trust. Utilizing extensive dataset that encompasses various driving scenarios and driver behaviors, we constructed a gradient-boosting machine model to forecast driver response times. To explain the underlying mechanics of the model, the Shapley Additive Explanations (SHAP) framework was employed, enabling a comprehensive interpretation of feature importance and inter-feature interactions. Key findings reveal that increased speeds heighten driver responsiveness due to amplified alertness, whereas slower speeds lead to delayed reactions. The influence of previous negative warnings, significantly extends response times to accurate warnings. Additionally, older drivers require longer response times. The relationship between the driving period and previous warning judgment profoundly affects subsequent driver responsiveness, indicating trust dynamics with FCW systems. By using interpretable machine learning, we provide insights into ADAS functionality, suggesting pathways for FCW responsiveness and contributing to the field of XAI applications. In the validation experiment, our approach improved driver response times, reducing the average time from 2.1 s to 1.6 s. The proportion of ignored warnings decreased from 12% to 6%, and the driver acceptance rate increased from 59% to 71%.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"146 ","pages":"Article 110308"},"PeriodicalIF":7.5,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143445097","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","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