Engineering Applications of Artificial Intelligence最新文献

{"title":"A self-feedback knowledge elicitation approach for chemical reaction predictions","authors":"Pengfei Liu ,&nbsp;Jun Tao ,&nbsp;Zhixiang Ren","doi":"10.1016/j.engappai.2025.111112","DOIUrl":"10.1016/j.engappai.2025.111112","url":null,"abstract":"<div><div>The task of chemical reaction predictions (CRPs) plays a pivotal role in advancing drug discovery and material science. However, while artificial intelligence (AI) has significantly advanced in capturing chemical reaction patterns, its effectiveness is still constrained by the vast and uncertain chemical reaction space and challenges in achieving high reaction selectivity, particularly due to existing methods’ limitations in fully leveraging the inherent knowledge within the data. To address these challenges, we introduce a data-curated self-feedback knowledge elicitation approach. This method starts from iterative optimization of molecular representations and facilitates the extraction of knowledge on chemical reaction types (RTs). Then, we employ adaptive prompt learning to infuse the prior knowledge into the large language model (LLM). As a result, we achieve significant enhancements: a 14.2% increase in retrosynthesis prediction accuracy, a 74.2% rise in reagent prediction accuracy, and an expansion in the model’s capability for handling multi-task chemical reactions. This research offers a novel paradigm for knowledge elicitation in scientific research and showcases the untapped potential of LLMs in CRPs.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111112"},"PeriodicalIF":7.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144177704","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}

{"title":"Bearing life prediction via a novel temporal convolutional network with embedded uncertainty quantification","authors":"Yang Ge ,&nbsp;Guodong Sun ,&nbsp;Benlian Xu","doi":"10.1016/j.engappai.2025.111303","DOIUrl":"10.1016/j.engappai.2025.111303","url":null,"abstract":"<div><div>This paper presents a novel Modernized Temporal Convolutional Network (MTCN) framework with embedded uncertainty quantification for high-accuracy bearing remaining useful life (RUL) prediction. The proposed architecture introduces three key innovations: (1) A redesigned temporal convolution module leveraging grouped depthwise/pointwise convolutions to enable cross-dimensional feature interaction, capturing comprehensive degradation patterns from raw and derived signal dimensions; (2) A gated attention unit (GAU) that efficiently models temporal dependencies with 50 % lower computational cost than standard self-attention mechanisms; (3) A dual uncertainty quantification system combining epistemic uncertainty and aleatory uncertainty to generate reliable prediction intervals. Validated on two widely recognized public bearing degradation datasets, our framework achieves state-of-the-art performance, demonstrating 32.3 % and 36.5 % reductions in Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), respectively, compared to existing methods, while providing quantifiable confidence bounds critical for industrial decision-making. The integration of cross-dimensional learning, lightweight attention, and probabilistic modeling establishes a new paradigm for trustworthy predictive maintenance in rotating machinery.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111303"},"PeriodicalIF":7.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169187","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}

{"title":"A codebook-driven approach for low-light image enhancement","authors":"Xu Wu ,&nbsp;Xianxu Hou ,&nbsp;Zhihui Lai ,&nbsp;Jie Zhou ,&nbsp;Ya-nan Zhang ,&nbsp;Witold Pedrycz ,&nbsp;Linlin Shen","doi":"10.1016/j.engappai.2025.111115","DOIUrl":"10.1016/j.engappai.2025.111115","url":null,"abstract":"<div><div>Low-light image enhancement (LLIE) aims to improve low-illumination images. However, existing methods face two challenges: (1) uncertainty in restoration from diverse brightness degradations; (2) loss of texture and color information caused by noise suppression and light enhancement. In this paper, we propose a novel enhancement approach, CodeEnhance, by leveraging discrete codebook priors and image refinement to address these challenges. In particular, we reframe LLIE as learning an <strong>image-to-code</strong> mapping from low-light images to discrete codebook, which has been learned from high-quality images. To enhance this process, a Semantic Embedding Module (SEM) is introduced to integrate semantic information with low-level features, and a Codebook Shift (CS) mechanism, designed to adapt the pre-learned codebook to better suit the distinct characteristics of our low-light dataset. Additionally, we present an Interactive Feature Transformation (IFT) module to refine texture and color information during image reconstruction, allowing for interactive enhancement based on user preferences. Extensive experiments on both real-world and synthetic benchmarks demonstrate that the incorporation of prior knowledge and controllable information transfer significantly enhances LLIE performance in terms of quality and fidelity. The proposed CodeEnhance exhibits superior robustness to various degradations, including uneven illumination, noise, and color distortion. The code can be obtained from <span><span>https://github.com/csxuwu/CodeEnhance</span><svg><path></path></svg></span> or <span><span>https://www.scholat.com/laizhihui.cn</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111115"},"PeriodicalIF":7.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169204","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}

{"title":"Chemically-Guided extreme gradient boosting models for predicting the elastic modulus of alkali-activated Concrete: Insights into base learner variants","authors":"Seyed Ali Eftekhar Afzali ,&nbsp;Emad Golafshani ,&nbsp;Tuan Ngo","doi":"10.1016/j.engappai.2025.111066","DOIUrl":"10.1016/j.engappai.2025.111066","url":null,"abstract":"<div><div>The widespread use of ordinary Portland cement (OPC) significantly contributes to carbon dioxide (CO₂) emissions, posing a major threat to the environment and accelerating climate change. Alkali-activated concrete (AAC) has emerged as a sustainable alternative to conventional OPC-based concrete, offering significant environmental benefits while providing desirable mechanical properties for structural applications. This study investigates the application of extreme gradient boosting regression (XGBR) algorithms to model the static modulus of elasticity (SME) of AAC, addressing the complexity of predicting SME by exploring the impact of different base learners on XGBR performance. Advancing beyond conventional approaches, this work integrates alternative base learners-namely artificial neural networks (ANN) and support vector regressors (SVR)- within the XGBR framework and systematically compares their performance with the widely adopted decision tree (DT) base learner. The models were developed and validated on a comprehensive database of AAC mixes, encompassing diverse material compositions and reactivities, mixture designs, and curing conditions. Results of Wilcoxon signed-rank test and the paired <em>t</em>-test showed that while the DT-based XGBR achieved the lowest error metrics during development, the novel SVR-based XGBR exhibited comparable generalisation performance on unseen data, outperforming the ANN-based XGBR. Interpretability analyses, including SHapley Additive exPlanations (SHAP) and permutation feature importance, identified compressive strength, total water content, and sodium oxide (Na₂O) content of the activator as key predictors of SME. Several interactions among factors influencing the system's chemistry and polymerisation reactions were revealed through partial dependence analysis to investigate the optimal chemical conditions. The findings underscore the efficacy of XGBR ensembles for modelling complex AAC properties, providing insights for optimising sustainable material design to meet the urgent need for environmentally responsible construction solutions.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111066"},"PeriodicalIF":7.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169190","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}

{"title":"Corrigendum to “Loss of coolant accident monitoring and pipe break diagnosis in pressurized water reactors using Bayesian-optimized long short-term memory models” [Eng. Appl. Artif. Intellig. 149 (2025) 110531]","authors":"Johndel Obra ,&nbsp;Shuichiro Miwa","doi":"10.1016/j.engappai.2025.111139","DOIUrl":"10.1016/j.engappai.2025.111139","url":null,"abstract":"","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111139"},"PeriodicalIF":7.5,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144177707","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}

{"title":"A problem-specific knowledge-based multi-objective algorithm for sustainable scheduling of distributed heterogeneous welding permutation flow shop","authors":"Jianguo Duan ,&nbsp;Zixuan Liu ,&nbsp;Mengting Wang ,&nbsp;Yulin Du ,&nbsp;Mengpei Yang","doi":"10.1016/j.engappai.2025.111197","DOIUrl":"10.1016/j.engappai.2025.111197","url":null,"abstract":"<div><div>The distributed heterogeneous welding permutation flow shop scheduling problem (DHWPFSP) involves the diversity, heterogeneity, and coordination of welding processes, making it more complex than traditional distributed flow shop scheduling problems. Additionally, welding is a high-energy-consuming process, and reducing energy consumption to achieve sustainable green manufacturing has always been a primary goal in the industry. Currently, research on this problem is minimal, and existing studies often overlook the impact of transportation on production efficiency and energy consumption. This paper proposes an efficient and energy-saving scheduling model, the goal is to minimize the maximum completion time and total energy consumption. By combining the features of the multi-objective evolutionary algorithm based on decomposition (MOEA/D) algorithm and the non-dominated sorting genetic algorithm-II (NSGA-II) algorithm, we designed the MONS-II algorithm (MOea/d and NSga II). Initial solutions are generated using the distributed neighborhood exchange heuristic (DNEH) method and a random generation strategy, with improvements made to the encoding method. The partially matched crossover (PMX) and precedence operation crossover (POX) strategies are applied, along with local neighborhood search, and enhancements to external archive management and adaptive adjustment strategies. Experimental results demonstrate that the MONS-II algorithm performs excellently in terms of both total time and energy consumption, providing more uniform and reasonable solutions. Using a crane manufacturing enterprise as an example, the effectiveness of the model and algorithm in the distributed welding shop scheduling problem is verified, providing theoretical support for the sustainable development of enterprises.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111197"},"PeriodicalIF":7.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169114","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}

{"title":"Cloud classification and cloud cover estimation using hybrid deep Kronecker network- ResNeXt","authors":"Gujanatti Rudrappa ,&nbsp;Nataraj Vijapur","doi":"10.1016/j.engappai.2025.111254","DOIUrl":"10.1016/j.engappai.2025.111254","url":null,"abstract":"<div><div>Clouds have a huge effect on the weather, climate, and balancing energy on the Earth's surface. Accurate cloud classification is essential as different clouds have various radiation effects and estimating the cloud cover can assist in accurate weather prediction. In this paper, a Hybrid Deep Kronecker Network- Residual Networks with Aggregated Transformations is devised for cloud classification and cloud cover estimation. At first, the input image is attained from the dataset, and a Kalman filter is employed to preprocess the image. Later, cloud segmentation is performed through Swin-Unet for deriving the cloud cover. Thereafter, features, such as Entropy with Median Binary Pattern, and statistical features, like skewness, mean, standard deviation, and variance are mined. Subsequently, the generated features and input image are fed to the established model for cloud classification. The proposed model is formulated by merging Residual Networks with Aggregated Transformations and Deep Kronecker Network. Lastly, the classified output and preprocessed output are fed to the cloud cover estimation module, where cloud cover is estimated using the proposed model. Furthermore, the proposed model is examined using metrics, such as True Positive Rate, accuracy, and True Negative Rate. From the experiment, the proposed model verified the accuracy of 0.915. Also, the True Negative Rate of 0.923, and True Positive Rate of 0.896 are obtained higher.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111254"},"PeriodicalIF":7.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169184","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}

{"title":"Artificial intelligence-driven models for predicting mechanical properties of low-emission microwave-cured geopolymer mortar","authors":"Faidhalrahman Khaleel ,&nbsp;Haitham Abdulmohsin Afan ,&nbsp;Alaa H. AbdUlameer ,&nbsp;Abdulrahman S. Abdullah ,&nbsp;Gökhan Kaplan ,&nbsp;Cengiz Duran Atiş","doi":"10.1016/j.engappai.2025.111291","DOIUrl":"10.1016/j.engappai.2025.111291","url":null,"abstract":"<div><div>This study reveals the unprecedented potential of artificial intelligence (AI) models in accurately predicting the mechanical properties of microwave-cured geopolymer mortars, thereby addressing a critical gap in the integration of AI and materials science. Furthermore, applying advanced algorithmic structure and machine learning is an unexplored area in extant literature. Four parameters: conventional curing period (time- NH), conventional curing temperature (temp-NH), microwave power (W), and microwave curing period (time-MW) are considered to generate a comprehensive dataset to predict compressive strength (CS) and flexural strength (FS). Four AI models have been adopted and rigorously compared: deep learning neural network (DL-NN), probabilistic neural network (PNN), radial basis function neural network (RBF-NN), and support vector machine (SVM). The performance was evaluated using various statistical matrices and visualization graphs. The findings showed that the DL-NN model performs exceptionally well in predicting compressive and flexural strengths, with correlation coefficient (R) values of 0.966 and 0.931, mean absolute error (MAE) of 3.544 MPa and 0.990 MPa, and root mean square error (RMSE) of 5.856 MPa and 1.442 MPa, respectively. These results demonstrate the model's ability to handle the complex, non-linear relationships inherent in the data. Meanwhile, the PNN model ranked second, with R values of 0.930 and 0.833, MAE values of 5.151 MPa and 1.566 MPa, and RMSE values of 7.947 MPa and 2.089 MPa, respectively. Furthermore, the carbon dioxide (CO₂) emissions and embodied energy were investigated. Finally, a sensitivity analysis was conducted to assess the relative importance of each parameter on the mechanical properties.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111291"},"PeriodicalIF":7.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169186","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}

{"title":"Underwater target pose Recognition: A deep learning approach based on sonar signals","authors":"Jikai Yang,&nbsp;Ziyan Gu,&nbsp;Peijun Li,&nbsp;Zihan Li,&nbsp;Wei Li","doi":"10.1016/j.engappai.2025.111309","DOIUrl":"10.1016/j.engappai.2025.111309","url":null,"abstract":"<div><div>Underwater target pose recognition has become a significant research focus in ocean exploration, resource investigation, and military applications. Traditional methods based on physical models and rule-based matching struggle with noise interference and dynamic underwater conditions. In this study, we propose an artificial intelligence-based approach, employing a multi-task deep learning model to enhance underwater target pose estimation. A synthetic sonar frequency response dataset was generated by simulating the backscattering characteristics of ellipsoidal targets under various incident angles. A multi-layer neural network was designed to simultaneously perform ellipsoid ratio classification and incidence angle estimation, utilizing a shared feature extraction framework for joint classification and regression learning. Experimental results demonstrate that the proposed model achieves a classification accuracy of 100 % under standard conditions and a mean absolute error (MAE) of 0.0595° in angle estimation. Even under significant noise interference (10 % noise added), the model maintains a classification accuracy of 99.5 % and an MAE of 0.3805°. In extreme conditions with high noise and strong signal attenuation, the model achieves 99 % classification accuracy and an MAE of 0.4328°, demonstrating its robustness and adaptability to complex underwater environments. These findings demonstrate that deep learning serves as a robust alternative to traditional physics-based modeling, significantly enhancing the precision and reliability of underwater target recognition. Future research will integrate real-world sonar data and explore advanced AI architectures such as convolutional neural networks (CNNs) and Transformers for enhanced feature extraction and generalization.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111309"},"PeriodicalIF":7.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169191","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}

{"title":"Multisource-domain regression transfer learning framework for predicting student academic performance considering balanced similarity","authors":"Li Wang ,&nbsp;Lucong Zhang ,&nbsp;Haotian Wu ,&nbsp;Teng Zhang ,&nbsp;Ke Qiu ,&nbsp;Tianyu Chen ,&nbsp;Hongwu Qin","doi":"10.1016/j.engappai.2025.111202","DOIUrl":"10.1016/j.engappai.2025.111202","url":null,"abstract":"<div><div>The increasing integration of information technology and artificial intelligence has extensively implemented computer-aided intelligent education systems in higher education. A critical task within these systems is student performance prediction, which forecasts future academic outcomes by analyzing data such as historical grades, learning behaviors, and classroom participation. This enables early intervention and personalized teaching based on scientific evidence. However, most existing methods rely on traditional machine learning techniques, which can hardly address issues such as domain distribution discrepancies and data imbalance effectively. To overcome these challenges, we propose a multisource-domain transfer learning regression framework that integrates domain selection, hybrid feature extraction, and dynamic joint distribution adaptation techniques. Specifically, the framework first selects appropriate source domains on the basis of preset thresholds via cross-validation. Thereafter, a hybrid feature extractor is used to derive (i) common features from the target and selected source domains and (ii) domain-specific features from the target domain. Finally, a dynamic adaptive factor is introduced to balance differences between the marginal and conditional distributions. Experimental results indicate that the proposed framework significantly reduces the root mean square error with an average prediction improvement of 21.05 %, compared with baseline methods and other advanced approaches.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111202"},"PeriodicalIF":7.5,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169112","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}