Engineering Applications of Artificial Intelligence最新文献_第6页

Informer learning framework based on secondary decomposition for multi-step forecast of ultra-short term wind speed 基于二次分解的超短期风速多步骤预报信息学习框架

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-22 DOI: 10.1016/j.engappai.2024.109702

Zihao Jin, Xiaomengting Fu, Ling Xiang, Guopeng Zhu, Aijun Hu

Accurate and dependable wind speed prediction holds paramount importance in facilitating the dispatch and safe operation of power systems. Nonetheless, the inherent instability of wind speed makes wind speed prediction challenging. Consequently, a short-term wind speed prediction framework, amalgamating secondary decomposition (SD)-Informer, has been proposed in this paper. Initially, the variational mode decomposition (VMD) is applied to decompose the primary wind speed sequence. Through the VMD feature decomposition module, it effectively filters and eliminates superfluous noise from wind speed data. Subsequently, the complete ensemble empirical mode decomposition with adaptive noise technique is introduced for a secondary decomposition targeting the high-frequency components derived from the initial decomposition. To address the limitation of neural network models in capturing essential information from lengthy sequential data concurrently, a predictive model based on Informer is proposed as wind speed prediction module, thereby enhancing prediction accuracy. The validation of this hybrid model encompasses four distinct time ranges. Multiple models are scrutinized through comparative analysis to ascertain the superior performance of the proposed hybrid model. The root mean square error of the proposed method is reduced by 33.02%、25.46%、24.26%, and 23.12% compared to gate recurrent unit (GRU), vision Transformer (ViT), attention (AT)-ViT, and CNN-atteneion (CA)-Bi-directional long short-term memory (BiLSTM) respectively. The mean absolute error of the proposed method in the first quarter is 0.432, with model comparison values reduction of 36.19%、22.99%、20.44%, and17.71% respectively. The experimental results indicate that the proposed model exhibits a strong capability in capturing the long-term dependencies between the input and output sequences of wind speed. It can perform multi-step predictions while ensuring high prediction accuracy.

准确可靠的风速预测对于促进电力系统调度和安全运行至关重要。然而，风速固有的不稳定性使得风速预测极具挑战性。因此，本文提出了一种短期风速预测框架，将二次分解（SD）与信息提供者相结合。首先，应用变异模式分解（VMD）来分解一次风速序列。通过 VMD 特征分解模块，可有效过滤和消除风速数据中的多余噪声。随后，引入了具有自适应噪声技术的完整集合经验模态分解，针对初始分解得出的高频成分进行二次分解。针对神经网络模型无法同时从冗长的序列数据中捕捉重要信息的局限性，提出了一种基于 Informer 的预测模型作为风速预测模块，从而提高了预测精度。该混合模型的验证包括四个不同的时间范围。通过比较分析，对多个模型进行了仔细研究，以确定所提出的混合模型的卓越性能。与门递归单元（GRU）、视觉变换器（ViT）、注意力（AT）-ViT 和 CNN-注意力（CA）-双向长短期记忆（BiLSTM）相比，所提方法的均方根误差分别减少了 33.02%、25.46%、24.26% 和 23.12%。所提方法在第一季度的平均绝对误差为 0.432，模型比较值分别减少了 36.19%、22.99%、20.44% 和 17.71%。实验结果表明，所提出的模型在捕捉风速输入和输出序列之间的长期依赖关系方面表现出很强的能力。它可以进行多步预测，同时确保较高的预测精度。

{"title":"Informer learning framework based on secondary decomposition for multi-step forecast of ultra-short term wind speed","authors":"Zihao Jin, Xiaomengting Fu, Ling Xiang, Guopeng Zhu, Aijun Hu","doi":"10.1016/j.engappai.2024.109702","DOIUrl":"10.1016/j.engappai.2024.109702","url":null,"abstract":"<div><div>Accurate and dependable wind speed prediction holds paramount importance in facilitating the dispatch and safe operation of power systems. Nonetheless, the inherent instability of wind speed makes wind speed prediction challenging. Consequently, a short-term wind speed prediction framework, amalgamating secondary decomposition (SD)-Informer, has been proposed in this paper. Initially, the variational mode decomposition (VMD) is applied to decompose the primary wind speed sequence. Through the VMD feature decomposition module, it effectively filters and eliminates superfluous noise from wind speed data. Subsequently, the complete ensemble empirical mode decomposition with adaptive noise technique is introduced for a secondary decomposition targeting the high-frequency components derived from the initial decomposition. To address the limitation of neural network models in capturing essential information from lengthy sequential data concurrently, a predictive model based on Informer is proposed as wind speed prediction module, thereby enhancing prediction accuracy. The validation of this hybrid model encompasses four distinct time ranges. Multiple models are scrutinized through comparative analysis to ascertain the superior performance of the proposed hybrid model. The root mean square error of the proposed method is reduced by 33.02%、25.46%、24.26%, and 23.12% compared to gate recurrent unit (GRU), vision Transformer (ViT), attention (AT)-ViT, and CNN-atteneion (CA)-Bi-directional long short-term memory (BiLSTM) respectively. The mean absolute error of the proposed method in the first quarter is 0.432, with model comparison values reduction of 36.19%、22.99%、20.44%, and17.71% respectively. The experimental results indicate that the proposed model exhibits a strong capability in capturing the long-term dependencies between the input and output sequences of wind speed. It can perform multi-step predictions while ensuring high prediction accuracy.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109702"},"PeriodicalIF":7.5,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720679","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}

引用次数: 0

The nexus of intelligent transportation: A lightweight Bi-input fusion detection model for autonomous-rail rapid transit 智能交通的纽带：用于自主轨道快速交通的轻量级双输入融合检测模型

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-22 DOI: 10.1016/j.engappai.2024.109705

Hongjie Tang , Jirui Wang , Jiaoyi Wu , Yanni Zhao , Jiangfan Chen , Fujian Liang , Zutao Zhang

Autonomous-rail Rapid Transit (ART) possesses various advantages in intelligent transportation, but it does not effectively recognize road conditions caused solely by deploying single-modal cameras. In this paper, a lightweight fusion-based object detection neural network was designed with multi-modal sensors for the ART. Firstly, the Light Detection and Ranging (LiDAR) applied additional encoding and preprocessing to the point cloud. Secondly, a backbone and a detection head of the network structure were proposed through re-parameterization and pruning techniques. Furthermore, a fusion module was designed with a selective soft attention mechanism to fuse the extracted features. The proposed model was tested on the open autonomous driving dataset; it achieved a 7.38% improvement in mean average precision (mAP) compared to the original you only look once (YOLO) as well as other state-of-the-art (SOTA) models. Finally, practical experiments were conducted in the maintenance center of ART to simulate the operational scenarios and validate the feasibility of the proposed method in this study. By fully utilizing the information in different modalities and addressing the limitations of single-modal recognition, efforts were made to improve the robustness of road object detection for ART under different road conditions. Consequently, our method provides effective solutions which benefit intelligent transportation with advanced algorithms and strategies.

自主轨道捷运（ART）在智能交通领域具有多种优势，但仅靠部署单模态摄像头并不能有效识别路况。本文设计了一种基于多模态传感器融合的轻量级物体检测神经网络，用于自动轨道捷运。首先，光探测和测距（LiDAR）对点云进行了额外的编码和预处理。其次，通过重新参数化和剪枝技术，提出了网络结构的主干和检测头。此外，还设计了一个融合模块，采用选择性软关注机制来融合提取的特征。所提出的模型在开放的自动驾驶数据集上进行了测试；与原始的 "你只看一次"（YOLO）模型以及其他最先进的（SOTA）模型相比，该模型的平均精度（mAP）提高了 7.38%。最后，在 ART 维护中心进行了实际实验，模拟运行场景，验证了本研究提出的方法的可行性。通过充分利用不同模态的信息，解决单一模态识别的局限性，努力提高 ART 在不同路况下道路物体检测的鲁棒性。因此，我们的方法提供了有效的解决方案，通过先进的算法和策略造福于智能交通。

{"title":"The nexus of intelligent transportation: A lightweight Bi-input fusion detection model for autonomous-rail rapid transit","authors":"Hongjie Tang , Jirui Wang , Jiaoyi Wu , Yanni Zhao , Jiangfan Chen , Fujian Liang , Zutao Zhang","doi":"10.1016/j.engappai.2024.109705","DOIUrl":"10.1016/j.engappai.2024.109705","url":null,"abstract":"<div><div>Autonomous-rail Rapid Transit (ART) possesses various advantages in intelligent transportation, but it does not effectively recognize road conditions caused solely by deploying single-modal cameras. In this paper, a lightweight fusion-based object detection neural network was designed with multi-modal sensors for the ART. Firstly, the Light Detection and Ranging (LiDAR) applied additional encoding and preprocessing to the point cloud. Secondly, a backbone and a detection head of the network structure were proposed through re-parameterization and pruning techniques. Furthermore, a fusion module was designed with a selective soft attention mechanism to fuse the extracted features. The proposed model was tested on the open autonomous driving dataset; it achieved a 7.38% improvement in mean average precision (mAP) compared to the original you only look once (YOLO) as well as other state-of-the-art (SOTA) models. Finally, practical experiments were conducted in the maintenance center of ART to simulate the operational scenarios and validate the feasibility of the proposed method in this study. By fully utilizing the information in different modalities and addressing the limitations of single-modal recognition, efforts were made to improve the robustness of road object detection for ART under different road conditions. Consequently, our method provides effective solutions which benefit intelligent transportation with advanced algorithms and strategies.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109705"},"PeriodicalIF":7.5,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720676","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}

引用次数: 0

Multilingual entity alignment by abductive knowledge reasoning on multiple knowledge graphs 通过多知识图谱上的归纳知识推理进行多语言实体对齐

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-22 DOI: 10.1016/j.engappai.2024.109660

Muhammad Usman Akhtar , Jin Liu , Zhiwen Xie , Xiaohui Cui , Xiao Liu , Bo Huang

Objectives:

Entity alignment (EA) seeks to identify similar real-world objects in different multilingual knowledge graphs (KGs), also known as ontology alignment. EA assists in handling a wide range of language semantics and in building integrated knowledge bases. However, most mainstream studies have focused on structural information, paying little attention to insufficient contextual information and limited handling of complex relationships. This paper aims to address these limitations and improve EA performance and efficiency.

Methods:

This paper investigates multilingual EA techniques and proposes a novel Abductive Knowledge Reasoning (AKR) model to address these issues. AKR can compute complex relationship semantics context by reasoning and enrich counterpart entity contextual information through centrality calculation, which helps connect distant entities in multilingual KGs.

Novelty:

The proposed AKR model introduces a new approach to EA by integrating centrality calculation and relational semantics reasoning. This method overcomes the limitations of existing EA techniques by effectively handling insufficient contextual information and complex relationships in multilingual KGs.

Findings:

AKR outperforms all state-of-the-art EA models across five datasets. AKR achieves

H i t @ 1

score of 79.4%, for entity alignment between Chinese-to-English knowledge graphs representing 19.9% improvement over the best-performing translation-based model, Neighborhood-Aware Attentional Representation Entity Alignment, and a 5.0% improvement over the best-performing graph neural network-based model, Relational Semantics Augmentation.

目标：实体对齐（EA）旨在识别不同多语言知识图谱（KG）中相似的现实世界对象，也称为本体对齐。实体对齐有助于处理各种语言语义和建立集成知识库。然而，大多数主流研究都侧重于结构信息，很少关注不充分的上下文信息和对复杂关系的有限处理。方法：本文研究了多语言 EA 技术，并提出了一种新颖的归纳知识推理（AKR）模型来解决这些问题。AKR可以通过推理计算复杂的关系语义上下文，并通过中心度计算丰富对应实体的上下文信息，从而帮助连接多语KG中的远距离实体。研究结果：在五个数据集上，AKR的表现优于所有最先进的EA模型。在中译英知识图谱之间的实体对齐方面，AKR获得了79.4%的Hit@1分数，比表现最好的基于翻译的模型 "邻域感知注意力表征实体对齐 "提高了19.9%，比表现最好的基于图神经网络的模型 "关系语义增强 "提高了5.0%。

{"title":"Multilingual entity alignment by abductive knowledge reasoning on multiple knowledge graphs","authors":"Muhammad Usman Akhtar , Jin Liu , Zhiwen Xie , Xiaohui Cui , Xiao Liu , Bo Huang","doi":"10.1016/j.engappai.2024.109660","DOIUrl":"10.1016/j.engappai.2024.109660","url":null,"abstract":"<div><h3>Objectives:</h3><div>Entity alignment (EA) seeks to identify similar real-world objects in different multilingual knowledge graphs (KGs), also known as ontology alignment. EA assists in handling a wide range of language semantics and in building integrated knowledge bases. However, most mainstream studies have focused on structural information, paying little attention to insufficient contextual information and limited handling of complex relationships. This paper aims to address these limitations and improve EA performance and efficiency.</div></div><div><h3>Methods:</h3><div>This paper investigates multilingual EA techniques and proposes a novel Abductive Knowledge Reasoning (AKR) model to address these issues. AKR can compute complex relationship semantics context by reasoning and enrich counterpart entity contextual information through centrality calculation, which helps connect distant entities in multilingual KGs.</div></div><div><h3>Novelty:</h3><div>The proposed AKR model introduces a new approach to EA by integrating centrality calculation and relational semantics reasoning. This method overcomes the limitations of existing EA techniques by effectively handling insufficient contextual information and complex relationships in multilingual KGs.</div></div><div><h3>Findings:</h3><div>AKR outperforms all state-of-the-art EA models across five datasets. AKR achieves <span><math><mrow><mi>H</mi><mi>i</mi><mi>t</mi><mi>@</mi><mn>1</mn></mrow></math></span> score of 79.4%, for entity alignment between Chinese-to-English knowledge graphs representing 19.9% improvement over the best-performing translation-based model, Neighborhood-Aware Attentional Representation Entity Alignment, and a 5.0% improvement over the best-performing graph neural network-based model, Relational Semantics Augmentation.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109660"},"PeriodicalIF":7.5,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720682","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}

引用次数: 0

Workload-based adaptive decision-making for edge server layout with deep reinforcement learning 利用深度强化学习为边缘服务器布局制定基于工作量的自适应决策

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-22 DOI: 10.1016/j.engappai.2024.109662

Shihua Li , Yanjie Zhou , Bing Zhou , Zongmin Wang

Mobile edge computing (MEC) is crucial in applications such as intelligent transportation, innovative healthcare, and smart cities. By deploying servers with computing and storage capabilities at the network edge, MEC enables low-latency services close to end users. However, the configuration of edge servers needs to meet the low-latency requirements and effectively balance the servers’ workloads. This paper proposes an adaptive layout and dynamic optimization method, modeling the edge server layout problem as a Markov decision process. It introduces a workload-based server placement rule that adjusts the locations of edge servers according to the load of base stations, enabling the learning of low-latency and load-balanced server layout strategies. Experimental validation on a real dataset from Shanghai Telecom shows that the proposed algorithm improves average latency performance by about 40% compared to existing technologies, and enhances workload balancing performance by about 17%.

移动边缘计算（MEC）在智能交通、创新医疗和智能城市等应用中至关重要。通过在网络边缘部署具有计算和存储功能的服务器，MEC 可在终端用户附近提供低延迟服务。然而，边缘服务器的配置需要满足低延迟要求，并有效平衡服务器的工作负载。本文提出了一种自适应布局和动态优化方法，将边缘服务器布局问题建模为马尔可夫决策过程。它引入了基于工作负载的服务器布局规则，可根据基站负载调整边缘服务器的位置，从而学习低延迟和负载平衡的服务器布局策略。在上海电信的真实数据集上进行的实验验证表明，与现有技术相比，该算法的平均延迟性能提高了约 40%，工作负载平衡性能提高了约 17%。

引用次数: 0

Anomaly detection in Smart-manufacturing era: A review 智能制造时代的异常检测：综述

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-21 DOI: 10.1016/j.engappai.2024.109578

Iñaki Elía, Miguel Pagola

Manufacturing downtime due to faults is costly and disruptive. With the increasing availability of real-time data in modern Smart Manufacturing (SM) environments, effective anomaly detection (AD) has become crucial but challenging due to diverse scenarios and methods. This paper aims to present a comprehensive review of state-of-the-art AD methods tailored for SM, to facilitate their implementation in real manufacturing environments while providing a foundation for future research. First, it introduces a structured SM classification framework highlighting recent and successful AD algorithms applied in real-world scenarios with a valuable repository of over 100 manufacturing datasets to support further research. Second, an extensive experimental evaluation of 16 AD algorithms across 29 SM datasets covering a broad and diverse spectrum of techniques, including supervised, unsupervised, and semi-supervised approaches, encompassing classic, deep, and ensemble methods. Finally, insights gained from these experiments are presented providing practical guidance on the most suitable methods for various manufacturing contexts, identifying key challenges and opportunities for future developments.

故障导致的生产停机代价高昂且具有破坏性。随着现代智能制造（SM）环境中实时数据的可用性不断提高，有效的异常检测（AD）变得至关重要，但由于应用场景和方法的多样性，这种检测方法具有挑战性。本文旨在全面回顾为智能制造量身定制的最先进的异常检测方法，以促进这些方法在实际制造环境中的应用，同时为未来研究奠定基础。首先，本文介绍了一个结构化的 SM 分类框架，重点介绍了最近成功应用于真实世界场景的 AD 算法，并提供了一个由 100 多个制造数据集组成的宝贵资料库，以支持进一步的研究。其次，在 29 个 SM 数据集上对 16 种 AD 算法进行了广泛的实验评估，涵盖了广泛多样的技术，包括有监督、无监督和半监督方法，涵盖了经典、深度和集合方法。最后，介绍了从这些实验中获得的启示，就最适合各种制造环境的方法提供了实用指导，并确定了未来发展的关键挑战和机遇。

{"title":"Anomaly detection in Smart-manufacturing era: A review","authors":"Iñaki Elía, Miguel Pagola","doi":"10.1016/j.engappai.2024.109578","DOIUrl":"10.1016/j.engappai.2024.109578","url":null,"abstract":"<div><div>Manufacturing downtime due to faults is costly and disruptive. With the increasing availability of real-time data in modern Smart Manufacturing (SM) environments, effective anomaly detection (AD) has become crucial but challenging due to diverse scenarios and methods. This paper aims to present a comprehensive review of state-of-the-art AD methods tailored for SM, to facilitate their implementation in real manufacturing environments while providing a foundation for future research. First, it introduces a structured SM classification framework highlighting recent and successful AD algorithms applied in real-world scenarios with a valuable repository of over 100 manufacturing datasets to support further research. Second, an extensive experimental evaluation of 16 AD algorithms across 29 SM datasets covering a broad and diverse spectrum of techniques, including supervised, unsupervised, and semi-supervised approaches, encompassing classic, deep, and ensemble methods. Finally, insights gained from these experiments are presented providing practical guidance on the most suitable methods for various manufacturing contexts, identifying key challenges and opportunities for future developments.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109578"},"PeriodicalIF":7.5,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720594","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}

引用次数: 0

Time-lagged relation graph neural network for multivariate time series forecasting 用于多元时间序列预测的时滞关系图神经网络

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-21 DOI: 10.1016/j.engappai.2024.109530

Xing Feng, Hongru Li, Yinghua Yang

Recently, Graph Neural Network-based approaches (GNNs) have been widely studied in Multivariate Time Series (MTS) prediction, which could extract information from the closely related variables for prediction. The variables contained in MTS data are lagged correlated, and the future trends of the lagging variables are guided by the leading variables. However, as the existing approaches only focus on delay-free relations, they cannot utilize the guidance information in leading variables to achieve accurate prediction. To address this issue, we propose a novel frame called the Time-Lagged Relation Graph Neural Network (TLGNN) including two key components: the time-lagged relation graph and the time-lagged relation graph learning. The time-lagged relation graph could explicitly model the time-delay relations among MTS variables by connecting variable nodes at lag intervals. The graph learning module could adaptively extract the time-delay relations among MTS variables. Based on the novel designed graph structure, the TLGNN could extract the guidance information from previous values of leading variables to generate more efficient feature representations for prediction. In experiments, the prediction accuracy is significantly improved due to the full exploration of the time-delay relations. Compared with existing methods, the TLGNN achieves the best results in both the single-step prediction and the multi-step prediction tasks.

最近，基于图神经网络的方法（GNN）在多变量时间序列（MTS）预测中得到了广泛研究，这种方法可以从密切相关的变量中提取信息进行预测。多变量时间序列数据中包含的变量是滞后相关的，滞后变量的未来趋势受领先变量的引导。然而，由于现有方法只关注无延迟关系，无法利用前导变量中的引导信息实现准确预测。为解决这一问题，我们提出了一种名为时滞关系图神经网络（TLGNN）的新框架，包括两个关键部分：时滞关系图和时滞关系图学习。时滞关系图可以通过连接时滞间隔的变量节点，明确地模拟 MTS 变量之间的时滞关系。图学习模块可以自适应地提取 MTS 变量之间的时滞关系。基于设计新颖的图结构，TLGNN 可以从前导变量的先前值中提取引导信息，生成更有效的预测特征表征。在实验中，由于充分挖掘了时延关系，预测精度得到了显著提高。与现有方法相比，TLGNN 在单步预测和多步预测任务中都取得了最佳结果。

{"title":"Time-lagged relation graph neural network for multivariate time series forecasting","authors":"Xing Feng, Hongru Li, Yinghua Yang","doi":"10.1016/j.engappai.2024.109530","DOIUrl":"10.1016/j.engappai.2024.109530","url":null,"abstract":"<div><div>Recently, Graph Neural Network-based approaches (GNNs) have been widely studied in Multivariate Time Series (MTS) prediction, which could extract information from the closely related variables for prediction. The variables contained in MTS data are lagged correlated, and the future trends of the lagging variables are guided by the leading variables. However, as the existing approaches only focus on delay-free relations, they cannot utilize the guidance information in leading variables to achieve accurate prediction. To address this issue, we propose a novel frame called the Time-Lagged Relation Graph Neural Network (TLGNN) including two key components: the time-lagged relation graph and the time-lagged relation graph learning. The time-lagged relation graph could explicitly model the time-delay relations among MTS variables by connecting variable nodes at lag intervals. The graph learning module could adaptively extract the time-delay relations among MTS variables. Based on the novel designed graph structure, the TLGNN could extract the guidance information from previous values of leading variables to generate more efficient feature representations for prediction. In experiments, the prediction accuracy is significantly improved due to the full exploration of the time-delay relations. Compared with existing methods, the TLGNN achieves the best results in both the single-step prediction and the multi-step prediction tasks.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109530"},"PeriodicalIF":7.5,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720762","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}

引用次数: 0

A framework for Interpretable deep learning in cross-subject detection of event-related potentials 事件相关电位跨主体检测中的可解释深度学习框架

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-21 DOI: 10.1016/j.engappai.2024.109642

Shayan Jalilpour , Gernot Müller-Putz

Event-related potential-based Brain-Computer Interfaces are becoming widely popular due to their ability to send commands quickly with high accuracy. However, the stationary characteristics of electroencephalographic signals, coupled with their low signal-to-noise ratio, lead to variations in amplitude, time period, and latency in the patterns of event-related potentials across different trials, sessions, days and subjects. Conventional feature extraction and machine learning algorithms are not designed to handle these differences, requiring the development of methods that can address these variations. Here, we propose a novel lightweight deep neural network for event-related potential classification, consisting of three modules. In this model, we have a spatio-temporal module that learns local features simultaneously across channels and time points. Following this, there's a component extractor module comprising depthwise convolutions, inspired by mixed depthwise convolutions, to capture the event-related potential characteristics with different temporal durations. Lastly, an advanced temporal layer addresses event-related potential shape and scale variations using deformable convolutions. We conducted experiments on event-related potential detection in a subject-independent scenario using one error-related negativity potential dataset and three perturbation-evoked potential datasets. Comparisons were made with established methods including two conventional machine learning algorithms and three well-known deep learning architectures, demonstrating that our model outperformed them in terms of classification accuracy and parameter efficiency. In our analysis, we aimed to understand the model's performance using gradient-weighted class activation mapping and t-distributed stochastic neighbor embedding. These methods facilitated the visualization and interpretation of our model's effectiveness, providing insights into its relationship with the neuroscientific characteristics of event-related potentials.

基于事件相关电位的脑机接口能够快速、准确地发送指令，因此广受欢迎。然而，脑电信号的静态特性加上其信噪比低，导致不同试验、会话、日期和受试者的事件相关电位模式在振幅、时间段和延迟方面存在差异。传统的特征提取和机器学习算法无法处理这些差异，因此需要开发能处理这些差异的方法。在此，我们提出了一种用于事件相关电位分类的新型轻量级深度神经网络，由三个模块组成。在这个模型中，我们有一个时空模块，可以跨信道和时间点同时学习局部特征。随后是一个由深度卷积组成的成分提取模块，其灵感来自于混合深度卷积，用于捕捉不同时间长度的事件相关电位特征。最后，高级时间层利用可变形卷积处理事件相关电位的形状和尺度变化。我们使用一个错误相关负性电位数据集和三个扰动诱发电位数据集，在不依赖受试者的情况下进行了事件相关电位检测实验。实验结果表明，我们的模型在分类准确性和参数效率方面都优于它们。在分析中，我们使用梯度加权类激活映射和 t 分布随机邻域嵌入来了解模型的性能。这些方法促进了模型效果的可视化和解释，使我们能够深入了解模型与事件相关电位的神经科学特征之间的关系。

{"title":"A framework for Interpretable deep learning in cross-subject detection of event-related potentials","authors":"Shayan Jalilpour , Gernot Müller-Putz","doi":"10.1016/j.engappai.2024.109642","DOIUrl":"10.1016/j.engappai.2024.109642","url":null,"abstract":"<div><div>Event-related potential-based Brain-Computer Interfaces are becoming widely popular due to their ability to send commands quickly with high accuracy. However, the stationary characteristics of electroencephalographic signals, coupled with their low signal-to-noise ratio, lead to variations in amplitude, time period, and latency in the patterns of event-related potentials across different trials, sessions, days and subjects. Conventional feature extraction and machine learning algorithms are not designed to handle these differences, requiring the development of methods that can address these variations. Here, we propose a novel lightweight deep neural network for event-related potential classification, consisting of three modules. In this model, we have a spatio-temporal module that learns local features simultaneously across channels and time points. Following this, there's a component extractor module comprising depthwise convolutions, inspired by mixed depthwise convolutions, to capture the event-related potential characteristics with different temporal durations. Lastly, an advanced temporal layer addresses event-related potential shape and scale variations using deformable convolutions. We conducted experiments on event-related potential detection in a subject-independent scenario using one error-related negativity potential dataset and three perturbation-evoked potential datasets. Comparisons were made with established methods including two conventional machine learning algorithms and three well-known deep learning architectures, demonstrating that our model outperformed them in terms of classification accuracy and parameter efficiency. In our analysis, we aimed to understand the model's performance using gradient-weighted class activation mapping and t-distributed stochastic neighbor embedding. These methods facilitated the visualization and interpretation of our model's effectiveness, providing insights into its relationship with the neuroscientific characteristics of event-related potentials.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109642"},"PeriodicalIF":7.5,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720894","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Decision-making systems improvement based on explainable artificial intelligence approaches for predictive maintenance 基于可解释人工智能方法的决策系统改进，用于预测性维护

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-21 DOI: 10.1016/j.engappai.2024.109601

Lala Rajaoarisoa , Raubertin Randrianandraina , Grzegorz J. Nalepa , João Gama

To maintain the performance of the latest generation of onshore and offshore wind turbine systems, a new methodology must be proposed to enhance the maintenance policy. In this context, this paper introduces an approach to designing a decision support tool that combines predictive capabilities with anomaly explanations for effective IoT predictive maintenance tasks. Essentially, the paper proposes an approach that integrates a predictive maintenance model with an explicative decision-making system. The key challenge is to detect anomalies and provide plausible explanations, enabling human operators to determine the necessary actions swiftly. To achieve this, the proposed approach identifies a minimal set of relevant features required to generate rules that explain the root causes of issues in the physical system. It estimates that certain features, such as the active power generator, blade pitch angle, and the average water temperature of the voltage circuit protection in the generator’s sub-components, are particularly critical to monitor. Additionally, the approach simplifies the computation of an efficient predictive maintenance model. Compared to other deep learning models, the identified model provides up to 80% accuracy in anomaly detection and up to 96% for predicting the remaining useful life of the system under study. These performance metrics and indicators values are essential for enhancing the decision-making process. Moreover, the proposed decision support tool elucidates the onset of degradation and its dynamic evolution based on expert knowledge and data gathered through Internet of Things (IoT) technology and inspection reports. Thus, the developed approach should aid maintenance managers in making accurate decisions regarding inspection, replacement, and repair tasks. The methodology is demonstrated using a wind farm dataset provided by Energias De Portugal.

为了保持最新一代陆上和海上风力涡轮机系统的性能，必须提出一种新的方法来加强维护政策。在此背景下，本文介绍了一种设计决策支持工具的方法，该工具将预测能力与异常解释相结合，可有效执行物联网预测性维护任务。从本质上讲，本文提出了一种将预测性维护模型与解释性决策系统相结合的方法。关键的挑战在于检测异常并提供合理的解释，使人类操作员能够迅速确定必要的行动。为实现这一目标，所提出的方法确定了生成规则所需的最小相关特征集，以解释物理系统问题的根本原因。据估计，某些特征（如有功发电机、叶片俯仰角和发电机子组件中电压电路保护的平均水温）对监控尤为重要。此外，该方法还简化了高效预测性维护模型的计算。与其他深度学习模型相比，所确定的模型在异常检测方面的准确率高达 80%，在预测所研究系统的剩余使用寿命方面的准确率高达 96%。这些性能指标和指标值对加强决策过程至关重要。此外，基于专家知识和通过物联网（IoT）技术和检测报告收集的数据，拟议的决策支持工具还能阐明退化的开始及其动态演变。因此，所开发的方法可帮助维护管理人员就检查、更换和维修任务做出准确决策。该方法使用葡萄牙能源公司（Energias De Portugal）提供的风电场数据集进行了演示。

{"title":"Decision-making systems improvement based on explainable artificial intelligence approaches for predictive maintenance","authors":"Lala Rajaoarisoa , Raubertin Randrianandraina , Grzegorz J. Nalepa , João Gama","doi":"10.1016/j.engappai.2024.109601","DOIUrl":"10.1016/j.engappai.2024.109601","url":null,"abstract":"<div><div>To maintain the performance of the latest generation of onshore and offshore wind turbine systems, a new methodology must be proposed to enhance the maintenance policy. In this context, this paper introduces an approach to designing a decision support tool that combines predictive capabilities with anomaly explanations for effective IoT predictive maintenance tasks. Essentially, the paper proposes an approach that integrates a predictive maintenance model with an explicative decision-making system. The key challenge is to detect anomalies and provide plausible explanations, enabling human operators to determine the necessary actions swiftly. To achieve this, the proposed approach identifies a minimal set of relevant features required to generate rules that explain the root causes of issues in the physical system. It estimates that certain features, such as the active power generator, blade pitch angle, and the average water temperature of the voltage circuit protection in the generator’s sub-components, are particularly critical to monitor. Additionally, the approach simplifies the computation of an efficient predictive maintenance model. Compared to other deep learning models, the identified model provides up to 80% accuracy in anomaly detection and up to 96% for predicting the remaining useful life of the system under study. These performance metrics and indicators values are essential for enhancing the decision-making process. Moreover, the proposed decision support tool elucidates the onset of degradation and its dynamic evolution based on expert knowledge and data gathered through Internet of Things (IoT) technology and inspection reports. Thus, the developed approach should aid maintenance managers in making accurate decisions regarding inspection, replacement, and repair tasks. The methodology is demonstrated using a wind farm dataset provided by Energias De Portugal.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109601"},"PeriodicalIF":7.5,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720678","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}

引用次数: 0

Towards salient object detection via parallel dual-decoder network 通过并行双解码器网络实现突出物体检测

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-21 DOI: 10.1016/j.engappai.2024.109638

Chaojun Cen , Fei Li , Zhenbo Li , Yun Wang

Salient object detection, an important preprocessing step in computer vision, segments the most prominent objects in an image. However, existing research in this field utilizes transformer-based methods to capture global context information, failing to effectively obtain local spatial features. To solve this issue, we propose a parallel dual-decoder network, which consists of a novel semantic decoder and a modified salient decoder. Specifically, the proposed semantic decoder is designed to learn the local spatial details, and the salient decoder utilizes the learnable queries to establish global saliency dependencies among objects. Moreover, the two decoders establish correlations between saliency and multi-scale semantic representations through cross-attention interaction, significantly enhancing the performance of salient object detection. In other words, we obtain global context information in the decoder to prevent discriminative features from being diluted during information propagation. Extensive experiments on 15 benchmark datasets demonstrate that our model significantly outperforms other comparison methods and shows promising potential for real-world applications such as challenging optical remote sensing, underwater, low-light, and other open scenarios. In addition, our method shows excellent performance in other downstream tasks such as camouflaged object detection, transparent object detection, shadow detection, and semantic segmentation.

突出物体检测是计算机视觉中一个重要的预处理步骤，用于分割图像中最突出的物体。然而，该领域的现有研究利用基于变换器的方法来捕捉全局上下文信息，无法有效获取局部空间特征。为了解决这个问题，我们提出了一种并行双解码器网络，它由一个新颖的语义解码器和一个改进的突出解码器组成。具体来说，所提出的语义解码器旨在学习局部空间细节，而显著性解码器则利用可学习的查询来建立物体之间的全局显著性依赖关系。此外，这两个解码器还通过交叉注意力交互建立了突出度与多尺度语义表征之间的相关性，从而显著提高了突出物体检测的性能。换句话说，我们在解码器中获得了全局上下文信息，从而避免了在信息传播过程中区分性特征被稀释。在 15 个基准数据集上进行的广泛实验表明，我们的模型明显优于其他比较方法，并在现实世界的应用中展现出巨大的潜力，如具有挑战性的光学遥感、水下、低照度和其他开放场景。此外，我们的方法在伪装物体检测、透明物体检测、阴影检测和语义分割等其他下游任务中也表现出色。

{"title":"Towards salient object detection via parallel dual-decoder network","authors":"Chaojun Cen , Fei Li , Zhenbo Li , Yun Wang","doi":"10.1016/j.engappai.2024.109638","DOIUrl":"10.1016/j.engappai.2024.109638","url":null,"abstract":"<div><div>Salient object detection, an important preprocessing step in computer vision, segments the most prominent objects in an image. However, existing research in this field utilizes transformer-based methods to capture global context information, failing to effectively obtain local spatial features. To solve this issue, we propose a parallel dual-decoder network, which consists of a novel semantic decoder and a modified salient decoder. Specifically, the proposed semantic decoder is designed to learn the local spatial details, and the salient decoder utilizes the learnable queries to establish global saliency dependencies among objects. Moreover, the two decoders establish correlations between saliency and multi-scale semantic representations through cross-attention interaction, significantly enhancing the performance of salient object detection. In other words, we obtain global context information in the decoder to prevent discriminative features from being diluted during information propagation. Extensive experiments on <em>15</em> benchmark datasets demonstrate that our model significantly outperforms other comparison methods and shows promising potential for real-world applications such as challenging optical remote sensing, underwater, low-light, and other open scenarios. In addition, our method shows excellent performance in other downstream tasks such as camouflaged object detection, transparent object detection, shadow detection, and semantic segmentation.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109638"},"PeriodicalIF":7.5,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720773","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}

引用次数: 0

A velocity adaptive steering control strategy of autonomous vehicle based on double deep Q-learning network with varied agents 基于不同代理的双深度 Q-learning 网络的自动驾驶汽车速度自适应转向控制策略

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-11-21 DOI: 10.1016/j.engappai.2024.109655

Xinyou Lin, Jiawang Huang, Biao Zhang, Binhao Zhou, Zhiyong Chen

Autonomous vehicle steering control is sensitive to the vehicle driving speed and traditional model-based approaches are limited by the accuracy of the control model in various driving speed scenarios. To address these challenges, this study proposes a model-free control strategy based on deep reinforcement learning (DRL). In this strategy, the improved double deep Q-learning network (DDQN) with varied agents is employed for steering control to minimize the tracking errors across varying speeds. According to the kinematic characteristics of the vehicle, a dynamic action space is applied to enhance the tracking capability at high speeds. Furthermore, to ensure the output of the agent is more stable, a velocity adaptive reward function is designed by incorporating an action penalty factor. The performance of the proposed strategy is evaluated through simulation and experimental comparisons with other existing algorithms at a double-lane change maneuver. The results demonstrate that the DDQN-based strategy can effectively adapt to various vehicle speeds and perform the tracking task more accurately and stably. Finally, the feasibility of this strategy is verified using an actual prototype vehicle.

自动驾驶汽车转向控制对车辆行驶速度非常敏感，而传统的基于模型的方法受限于各种行驶速度场景下控制模型的准确性。为了应对这些挑战，本研究提出了一种基于深度强化学习（DRL）的无模型控制策略。在该策略中，改进的双深度 Q-learning 网络（DDQN）与不同的代理被用于转向控制，以最小化不同速度下的跟踪误差。根据车辆的运动特性，采用动态动作空间来增强高速行驶时的跟踪能力。此外，为了确保代理的输出更加稳定，还设计了一个速度自适应奖励函数，其中包含一个动作惩罚因子。在双车道变道演习中，通过模拟和与其他现有算法的实验比较，对所提策略的性能进行了评估。结果表明，基于 DDQN 的策略能有效适应各种车速，并能更准确、更稳定地完成跟踪任务。最后，使用实际原型车辆验证了该策略的可行性。

{"title":"A velocity adaptive steering control strategy of autonomous vehicle based on double deep Q-learning network with varied agents","authors":"Xinyou Lin, Jiawang Huang, Biao Zhang, Binhao Zhou, Zhiyong Chen","doi":"10.1016/j.engappai.2024.109655","DOIUrl":"10.1016/j.engappai.2024.109655","url":null,"abstract":"<div><div>Autonomous vehicle steering control is sensitive to the vehicle driving speed and traditional model-based approaches are limited by the accuracy of the control model in various driving speed scenarios. To address these challenges, this study proposes a model-free control strategy based on deep reinforcement learning (DRL). In this strategy, the improved double deep Q-learning network (DDQN) with varied agents is employed for steering control to minimize the tracking errors across varying speeds. According to the kinematic characteristics of the vehicle, a dynamic action space is applied to enhance the tracking capability at high speeds. Furthermore, to ensure the output of the agent is more stable, a velocity adaptive reward function is designed by incorporating an action penalty factor. The performance of the proposed strategy is evaluated through simulation and experimental comparisons with other existing algorithms at a double-lane change maneuver. The results demonstrate that the DDQN-based strategy can effectively adapt to various vehicle speeds and perform the tracking task more accurately and stably. Finally, the feasibility of this strategy is verified using an actual prototype vehicle.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109655"},"PeriodicalIF":7.5,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720886","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}

引用次数: 0