Expert Systems with Applications最新文献

{"title":"Dynamic multi-objective optimization using historical evolutionary learning with global alignment local descriptor matching and collaborative guidance","authors":"Kaiquan Guan ,&nbsp;Haibin Ouyang ,&nbsp;Steven Li ,&nbsp;Gaige Wang ,&nbsp;Nagwan Abdel Samee ,&nbsp;Essam H. Houssein","doi":"10.1016/j.eswa.2025.128915","DOIUrl":"10.1016/j.eswa.2025.128915","url":null,"abstract":"<div><div>Dynamic multi-objective optimization problems involve conflicting objectives that evolve over time, necessitating algorithms capable of efficiently tracking the dynamic Pareto optimal set and preserving solution diversity. To address this, the paper proposes a framework for dynamic multi-objective optimization algorithms based on Historical Evolutionary Learning (EHEL). The framework employs four strategies: using global alignment and local descriptor matching to improve the accuracy of historical individual searches; adopting a multi-history experience collaborative guidance strategy to integrate historical information and enhance the reliability of evolutionary direction; introducing a dynamic quadratic correction strategy to revise less-potential solutions; and proposing a shrinking boundary strategy to preserve directional information and enhance boundary exploration capability. Experiments on the CEC 2018 benchmark test set show that EHEL exhibits superior optimization capabilities across various dynamic environments, significantly enhancing convergence diversity and solution quality compared to existing algorithms. This research provides a robust and adaptive solution strategy for dynamic multi-objective optimization by effectively integrating historical experience with adaptive mechanisms.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128915"},"PeriodicalIF":7.5,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633494","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Evolutionary multi-task robust architecture search for network intrusion detection","authors":"Yeming Yang ,&nbsp;Zhihao Liu ,&nbsp;Ka-Chun Wong ,&nbsp;Qiuzhen Lin ,&nbsp;Jianping Luo ,&nbsp;Jianqiang Li","doi":"10.1016/j.eswa.2025.128899","DOIUrl":"10.1016/j.eswa.2025.128899","url":null,"abstract":"<div><div>Network Intrusion Detection (NID) becomes a key technology for ensuring network security. Recent researchers have proposed various NID systems based on neural networks. However, these networks require expensive expert knowledge for manual design, which is tedious and time-consuming. Moreover, they easily suffer from adversarial attacks, which limits their application in safety-critical scenarios. To alleviate the above problems, this paper proposes an evolutionary multi-task robust architecture search method, called EMR-NID, which can automatically design robust architectures for NID systems. First, we design an architecture transfer update strategy that achieves information sharing and knowledge transfer between different tasks. Then, we develop an architecture performance correction strategy that enhances the efficiency of robust search and strengthens NID’s defense capability. Finally, our EMR-NID method is validated on three well-known NID datasets, i.e., NSL-KDD, UNSW-NB15, and Edge-IIoTset. The experimental results show that EMR-NID can outperform some state-of-the-art NID methods in terms of clean and robust accuracy under multiple scenarios.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128899"},"PeriodicalIF":7.5,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"General machine learning models for interpreting and predicting efficiency degradation in organic solar cells","authors":"David Valiente ,&nbsp;Fernando Rodríguez-Mas ,&nbsp;Juan V. Alegre-Requena ,&nbsp;David Dalmau ,&nbsp;María Flores ,&nbsp;Juan C. Ferrer","doi":"10.1016/j.eswa.2025.128890","DOIUrl":"10.1016/j.eswa.2025.128890","url":null,"abstract":"<div><div>Photovoltaic (PV) energy plays a key role in addressing the growing global energy demand. Organic solar cells (OSCs) represent a promising alternative to silicon-based PVs due to their low cost, lightweight, and sustainable production. Despite achieving power conversion efficiencies (PCEs) over 20 %, OSCs still face challenges in stability and efficiency. Recent advances in manufacturing, artificial intelligence and machine learning (ML) achieve optimized and screened OSCs for greater sustainability and commercial viability, thus potentially reducing costs while ensuring stable and long term performance. This work presents optimal ML models to represent the temporal degradation on the PCE of polymeric OSCs with structure ITO/PEDOT:PSS/P3HT:PCBM/Al. First, we generated a database with 166 entries with measurements of 5 OSCs, and up to 7 variables regarding the manufacturing and environmental conditions for more than 180 days. Then, we relied on a software framework that provides a conglomeration of automated ML protocols that execute sequentially against our database by simply command-line interface. This easily permits hyper-optimizing the ML models through exhaustive benchmarking so that optimal models are obtained. The accuracy for predicting PCE over time reaches values of the coefficient determination widely exceeding 0.90, whereas the root mean squared error, sum of squared error, and mean absolute error are significantly low. Additionally, we assessed the predictive ability of the models using an unseen OSC as an external set. For comparative purposes, classical Bayesian regression fitting are also presented, which only perform sufficiently for univariate cases of single OSCs.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128890"},"PeriodicalIF":7.5,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 learning-based co-evolution optimization framework for energy-aware distributed heterogeneous flexible flow shop lot-streaming scheduling problem","authors":"Fuqing Zhao,&nbsp;Fumin Yin,&nbsp;Jianlin Zhang,&nbsp;Tian Peng Xu","doi":"10.1016/j.eswa.2025.128986","DOIUrl":"10.1016/j.eswa.2025.128986","url":null,"abstract":"<div><div>The distributed heterogeneous flexible flow shop scheduling problem (DHFFSP) has been considered in the era of economic globalization. Meanwhile, in some actual production scenarios, some jobs are divided into multiple sub-lots to boost the efficiency of intelligent manufacturing systems. The complexity of the scheduling problem is increased by the inevitable multiple time constraints among the jobs. In addition, considering the energy consumption, the energy-aware distributed heterogeneous flexible flow shop lot-streaming scheduling problem (EADHFFLSP) with release times, sequence-dependent setup and transport times is studied in the context of green manufacturing, which conforms to the actual production scenario of aluminum industry in the non-ferrous metallurgical industry. A learning-based co-evolution optimization framework (LBCOF) is designed to address EADHFFLSP with the minimization objectives of the maximum completion time and total energy consumption. In LBCOF, the population is divided into a global population and a local population, which performs global search and local search operations, respectively. Three heuristic rules are devised to generate the initial population. In local search, eight single-factory knowledge-driven operators and ten multi-factory knowledge-driven operators are proposed to update local population. A learning-based selection mechanism with dueling double deep Q-network (Dueling DDQN) component is presented to pick the best local search operator for the local population. Two energy-saving strategies are developed to improve the local population. The experimental findings reveal that LBCOF exhibits superior performance compared to some state-of-the-art algorithms for addressing EADHFFLSP.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128986"},"PeriodicalIF":7.5,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633491","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 real-time speech interaction analytics framework for group activities using SNA and LLM techniques","authors":"Diego Monsalves ,&nbsp;Fabián Riquelme ,&nbsp;Hector Cornide-Reyes","doi":"10.1016/j.eswa.2025.128948","DOIUrl":"10.1016/j.eswa.2025.128948","url":null,"abstract":"<div><div>In the current digital era, analyzing the dynamics of interaction in groups presents challenges in fields such as education, the business sector, and healthcare. The lack of integrated tools that monitor and evaluate discursive and social interactions in real-time makes it difficult to understand the flow of collaboration, the formation of effective teams, or the monitoring of social cognitive processes. In this article, we present a framework designed to analyze speech interactions in group activities by combining Social Network Analysis (SNA) and Large Language Models (LLM). Naira enables the real-time capture, processing, and analysis of speech interaction data, providing tools to evaluate discursive effectiveness and collaborative dynamics. The framework’s components are detailed in its different stages, and application cases are explored in educational, business, and healthcare contexts. A proof of concept in an educational environment proves the versatility and potential of the proposal to improve the understanding and optimization of group processes. Integrating SNA and LLM offers a comprehensive perspective combining validated and interpretable techniques to analyze attribute and relational variables with advanced and current artificial intelligence techniques. The framework’s main innovation lies in its ability to fuse the quantitative structural analysis of SNA with the semantic and qualitative content analysis of LLMs, offering a novel perspective that overcomes the limitations of each technique in isolation.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128948"},"PeriodicalIF":7.5,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144631522","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Scalar projective synchronization for uncertain T-S fuzzy systems with unified control fluctuation: Implementation to quadruple-tank process model","authors":"R. Elavarasi ,&nbsp;B. Adhira ,&nbsp;G. Nagamani ,&nbsp;Van Thanh Huynh","doi":"10.1016/j.eswa.2025.128878","DOIUrl":"10.1016/j.eswa.2025.128878","url":null,"abstract":"<div><div>This work intends to the study of the energy-based performances of the Takagi-Sugeno (T-S) fuzzy systems in the presence of parameter uncertainties and non-linear function. This type of fuzzy system refers to an extension of the T-S fuzzy systems, where non-linear terms are directly incorporated into the outcome of the fuzzy rules. The major motive of this study is to examine the scalar projective synchronization criterion of robust master and slave systems via unified control fluctuation, specifically, non-fragility control. In contrast to the current literature, both norm-bounded additive and multiplicative uncertainties are considered in the non-fragile robust state-feedback controller. The sufficient conditions for robust extended dissipative performance of the T-S fuzzy system are derived by utilizing a projective synchronization approach under state feedback controllers with non-fragility and by linearizing the integral terms obtained from the augmented Lyapunov-Krasovskii functional. Finally, numerical examples counting with fuzzified quadruple-tank process system model have been conducted through MATLAB software to demonstrate the importance and efficacy of the given theoretical results.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128878"},"PeriodicalIF":7.5,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633552","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 novel deep reinforcement learning framework based on digital twins for dynamic job shop scheduling problems","authors":"Wenquan Zhang,&nbsp;Zhaoxian Peng,&nbsp;Fei Zhao,&nbsp;Bo Feng,&nbsp;Xuesong Mei","doi":"10.1016/j.eswa.2025.128708","DOIUrl":"10.1016/j.eswa.2025.128708","url":null,"abstract":"<div><div>With the increasing diversity of product demands, the complexity of production scheduling planning has also been continuously escalating. Existing scheduling models exhibit significant deviations from actual production systems, making it challenging to directly apply scheduling algorithms to practical systems. High-fidelity digital twin (DT) models offer the capability to faithfully replicate production processes, providing effective means for training and validating scheduling algorithms. In this context, we propose a dynamic scheduling framework, DT-DRL, based on DT for deep reinforcement learning (DRL) applications in real production scheduling. Firstly, we employ DT technology to model actual production lines, effectively addressing the issue of model completeness. Secondly, we utilize the Double Deep Q-Network (DDQN) algorithm for offline training, followed by online decision-making, effectively addressing the challenge of real-time dynamic scheduling. Lastly, experimental training and validation are conducted using historical order and equipment data from the water heater inner tank welding production line. The experimental results demonstrate the robustness of our model.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128708"},"PeriodicalIF":7.5,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144631413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"One-step temporal subspace clustering with structured sparsity and indicator graph-Laplacian regularization","authors":"Wenyu Hu ,&nbsp;Huiying Huang ,&nbsp;Tinghua Wang","doi":"10.1016/j.eswa.2025.128889","DOIUrl":"10.1016/j.eswa.2025.128889","url":null,"abstract":"<div><div>Subspace clustering (SC) is a powerful technique for effectively segmenting data residing in multiple subspaces. However, traditional SC methods often fall short in accurately clustering temporal data due to their limited ability to capture temporal dependencies. These methods typically adopt a two-step framework: first, an affinity matrix is learned from the data; second, spectral clustering is performed using the affinity matrix to construct the indicator matrix for achieving the final segmentation. This disjointed process neglects the interdependence between the affinity matrix and the cluster assignments, and hence fails to fully exploit the temporal smoothness inherent in sequential data, where neighboring samples are usually similar. To address these limitations, we propose a novel one-step temporal subspace clustering method that integrates Structured Sparsity and Indicator graph-Laplacian regularization, termed SSIL. Our approach improves upon existing temporal SC techniques in two key aspects. First, we introduce a temporal Indicator Graph-Laplacian (IL) regularization directly on the indicator matrix, which promotes temporal smoothness and enhances alignment between the clustering result and ground truth. Second, we incorporate Structured Sparsity (SS) to jointly learn the affinity and indicator matrices within a unified optimization framework. We further develop an efficient optimization algorithm to alternatingly solve the affinity and indicator matrices. Extensive experiments on six benchmark datasets, particularly on motion capture data, demonstrate the effectiveness of our method and its superior performance compared to several state-of-the-art approaches.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128889"},"PeriodicalIF":7.5,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144631412","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Assessment of reaching laws for linear and nonlinear sliding surfaces: Improving performance of pipe crack sealing manipulator","authors":"Santosh Kumar,&nbsp;S.K. Dwivedy","doi":"10.1016/j.eswa.2025.128935","DOIUrl":"10.1016/j.eswa.2025.128935","url":null,"abstract":"<div><div>This study investigates the control of a tree-type pipe crack sealing manipulator (PCSM) with five specialized branches, designed to seal cracks inside the concrete pipes. This PCSM has the ability to navigate in vertical and horizontal pipes for crack repair tasks. A CAD model of the pipe with target crack and PCSM has been modelled in SolidWorks, and conclusive data of the crack has been further exported in Simulink for precise control. Observation and simulation studies in SolidWorks reveal that only the fifth branch of the PCSM successfully and effectively executes crack repair over substantial lengths. Furthermore, the repair of the crack has been assured for the branch of the PCSM. However, the accurate trajectory tracking in pipe crack sealing tasks is challenging due to model uncertainties and disturbances in confined environments. Sliding mode control (SMC) with various reaching laws for linear and nonlinear sliding surfaces has been examined to counter these challenges. The research investigated the efficacy of nine SMC reaching laws. These reaching laws are based on recent advancements in SMC. The novelty of this work uniquely compares nine reaching laws on both linear and nonlinear surfaces, where only CPRL (constant plus proportional rate reaching law) has previously been explored in nonlinear cases. Simulation-based evaluation using a SolidWorks-modeled PCSM and trajectory control in Simulink, incorporating mass uncertainty (+5 %) and joint disturbances equal to 40 % of the mean torque, demonstrates that nonlinear sliding surfaces consistently outperform linear ones in terms of tracking accuracy and disturbance rejection. Among the nine tested reaching laws on nonlinear surfaces, FTSMRL (novel fast terminal sliding mode reaching law), TSMRL (terminal sliding mode reaching law), and NSMRL (new sliding mode reaching law) demonstrated significantly lower tracking errors compared to the others, with the rms error of these three laws are comparable. The control efforts of FTSMRL and TSMRL were moderately high and chattering is very high compared to NSMRL. Therefore, NSMRL is proposed as the most suitable reaching law in this study due to its balanced performance, minimal chattering and control effort, as well reliable tracking accuracy. Furthermore, the study also calculated the reaching times for each law and confirmed their asymptotic stability using the Lyapunov stability criterion, demonstrating the efficacy of SMC in enhancing PCSM control.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128935"},"PeriodicalIF":7.5,"publicationDate":"2025-07-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144613828","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"SDERIME: Enhanced RIME algorithm with Sobol sequences and differential evolution for heavy calcium carbonate powder particle size distribution soft sensor model optimization","authors":"Shuai Zou ,&nbsp;Maohui Peng ,&nbsp;Jing Yang ,&nbsp;Qing Feng ,&nbsp;Mingyuan Dou ,&nbsp;Fuchuan Huang ,&nbsp;Lin Chen","doi":"10.1016/j.eswa.2025.128993","DOIUrl":"10.1016/j.eswa.2025.128993","url":null,"abstract":"<div><div>The original RIME algorithm is regarded as an efficient <em>meta</em>-heuristic algorithm, it has limitations such as an imbalance between exploration and exploitation, local optimal sensitivity, and suboptimal convergence accuracy. To address these challenges, this paper proposes an enhanced RIME algorithm with Sobol sequences strategy and differential evolution (DE) strategy (SDERIME), which introduce the Sobol sequences strategy in the initialization stage of the RIME algorithm, the elite DE strategy in the hard-rime puncture mechanism, and combine the DE strategy after rime-searching process. In the CEC2017&amp;2022 benchmark functions and 6 engineering problems test, by comparing with 14 other algorithms, the experimental results and statistical analysis proved that SDERIME is effective and efficient in various optimization tasks. And the application of SDERIME in the particle size distribution soft-sensing model of the heavy calcium carbonate (HCC) vertical roller mill (VRM) system has improved the prediction accuracy. These findings indicate that SDERIME has wide applicability and can be used as an advanced optimization technology in a variety of practical applications.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128993"},"PeriodicalIF":7.5,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144605660","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}