Swarm and Evolutionary Computation最新文献

{"title":"Multiobjective security-constrained optimal reactive dispatch with enhanced sinusoidal discretization","authors":"Ellen Cristina Ferreira ,&nbsp;Eduardo N. Asada ,&nbsp;Fillipe Matos de Vasconcelos ,&nbsp;Eduardo Werley S. Angelos","doi":"10.1016/j.swevo.2025.102193","DOIUrl":"10.1016/j.swevo.2025.102193","url":null,"abstract":"<div><div>Optimal Reactive Dispatch (ORD) with security constraints is a nonlinear optimization problem used to assess the impacts of predefined sets of contingencies in power systems. Its resolution helps operators identify secure and stable operating conditions. Additionally, the problem involves determining the optimal settings of control components to achieve an improved system configuration. When more complex modeling is considered – incorporating multiple objectives and discrete decision variables – most existing approaches struggle to find good feasible solutions, particularly as the number of contingencies increases. Our contribution to this multi-objective mixed-integer nonlinear programming (MINLP) problem is a solution method based on Evolutionary Particle Swarm Optimization (EPSO), enhanced with a sinusoidal discretization function. With classical PSO-based methods, the consideration of discrete variables usually implies adapting scalar operations to operate with discrete variables, which degrades the original PSO algorithm. However, the proposed application of the sinusoidal function allows for accurate modeling of discrete variables used to coordinate reactive power sources, while simultaneously minimizing active power losses and generator reactive power outputs (i.e., increasing reactive margins) without losing accuracy and convergence speed. We conducted exhaustive tests on the IEEE-118 and IEEE-300 bus systems. These included a sensitivity analysis on the most influential parameters to demonstrate the effectiveness and robustness of the proposed multi-objective model in delivering reliable trade-off solutions that satisfy both technical and safety requirements across different objectives. Comparisons with other known metaheuristic algorithms have also been done, which confirmed the good performance of the proposed method in terms of resolution time and solution quality.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102193"},"PeriodicalIF":8.5,"publicationDate":"2025-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145362344","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":"Integrating evacuation and relief redistribution into last-mile relief network design: A two-stage distributionally robust optimization approach","authors":"Peiyu Zhang,&nbsp;Zhenge Yang,&nbsp;Haorui Ge,&nbsp;Zichao Xu,&nbsp;Luzheng Bi","doi":"10.1016/j.swevo.2025.102192","DOIUrl":"10.1016/j.swevo.2025.102192","url":null,"abstract":"<div><div>Last-mile relief networks (LMRNs) play a critical role in disaster response, as they directly connect relief supplies to affected populations and determine the timeliness and effectiveness of emergency operations. Consequently, an efficient and reliable LMRN is essential for post-disaster management, ensuring that evacuation and relief distribution processes are seamlessly integrated and effectively executed. However, current approaches often neglect the varying severity of disasters, which is essential for tailoring evacuation networks, shelter planning, and relief distribution to real-world post-disaster needs. This paper investigates the design of a last-mile relief network that jointly integrates emergency evacuation and relief redistribution in post-disaster scenarios. The objective is to provide effective evacuation plans for severely affected areas while ensuring equitable relief allocation, even in cases where distribution centers are damaged. To capture the dual uncertainties of disaster environments and varying severity levels, we propose a two-stage distributionally robust optimization (DRO) model, where distinct ambiguity sets describe uncertain parameters. In the first stage, we determine the locations of candidate distribution centers (CDCs), local distribution centers (LDCs), and shelters. In the second stage, we optimize the evacuation plan and the allocation of relief supplies in response to the realized post-disaster conditions. To address the computational complexity of the two-stage DRO, we derive a tractable robust counterpart based on partial probability information and design a tailored solution algorithm to efficiently obtain optimal strategies. The proposed approach is validated using a real-world case study of the Nepal earthquake, demonstrating improvements in evacuation efficiency and relief distribution reliability, while sensitivity analyses further yield practical insights for disaster response planning.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102192"},"PeriodicalIF":8.5,"publicationDate":"2025-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145362343","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 relation learning-assisted algorithm with knee point transfer for expensive dynamic multi-objective optimization","authors":"Xinyu Xue ,&nbsp;Ziqi Cheng ,&nbsp;Xuefeng Chen ,&nbsp;Liang Feng","doi":"10.1016/j.swevo.2025.102196","DOIUrl":"10.1016/j.swevo.2025.102196","url":null,"abstract":"<div><div>Expensive dynamic multi-objective optimization problems (EDMOPs) involve multiple objective functions that change over time steps, and only a limited number of function evaluations are allowed at each time step. Existing methods typically treat EDMOPs as multiple independent static expensive multi-objective optimization problems and track the evolving pareto optimal set (POS) through Gaussian Process (GP)-assisted optimization. However, the dynamic nature of the environment results in a severe scarcity of training samples at each time step, which may impact the fitting accuracy of the GP model and prevent the accurate prediction of the POS. Taking this cue, in this paper, we propose a relation learning-assisted expensive dynamic multi-objective optimization algorithm. Unlike existing methods, the proposed approach simultaneously constructs relation models based on category criteria and fitness criteria. These two types of models work collaboratively in a two-stage filtering mechanism to precisely select the optimized population, enhancing the exploration of the search space while maintaining good fitting accuracy. Additionally, to accelerate the convergence of the optimization process, we predict the knee point set at the current time step using the center points and manifolds of knee points from historical time steps, effectively guiding the search direction of the population. To evaluate the performance of the proposed method, we conduct an extensive empirical study utilizing commonly used EXDMOP benchmarks and a real-world case study on gradient material machining. Experimental results demonstrate the effectiveness of the proposed method in solving EDMOPs.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102196"},"PeriodicalIF":8.5,"publicationDate":"2025-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145362539","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":"Evolving neural network for few-shot learning via training-free NAS","authors":"Haidong Kang ,&nbsp;Nan Jiang ,&nbsp;Jianming Zhao ,&nbsp;Shi Cheng ,&nbsp;Hongjiang Wang ,&nbsp;Lianbo Ma","doi":"10.1016/j.swevo.2025.102177","DOIUrl":"10.1016/j.swevo.2025.102177","url":null,"abstract":"<div><div>Efforts to improve the performance of Few-Shot Learning (FSL) have mainly centered around introducing more FSL approaches. However, the role of neural networks in FSL is less extensively analyzed. In this paper, we aim to bridge the gap between neural networks and FSL, and to propose a novel method from a training-free Neural Architecture Search (NAS) perspective to FSL. Specifically, we first conduct an in-depth analysis of Model Agnostic Meta Learning (MAML) based methods tailored to FSL, and find the main bottleneck of MAML-based methods for FSL that is attributed to the second-order of MAML. To address this issue, we introduce a new Theorem to ensure the first-order convergence of MAML. Then, we propose a novel Few-shot Neural Architecture Search (FNAS) framework to efficiently design neural architectures for FSL at initialization. FNAS introduces a training-free proxy by combining principles from Neural Tangent Kernel (NTK) and Fisher Information Matrix (FIM). This proxy effectively captures the expressivity of candidate architectures from the given search space in a training-free manner. To further enhance search efficiency, we integrate this proxy into an improved evolutionary algorithm to comprehensively explore the architecture space under minimal computational budgets. Experimental validation on mainstream benchmarks demonstrates that FNAS achieves state-of-the-art performance while being less costly in terms of computational budgets than its competitors.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102177"},"PeriodicalIF":8.5,"publicationDate":"2025-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145362345","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 comprehensive survey on the P system optimization algorithms’ variants and their applications","authors":"Shipin Yang,&nbsp;Songlu Wang,&nbsp;Wenhua Jiao,&nbsp;Xue Mei,&nbsp;Qing Zhang,&nbsp;Yinqiang Zhang,&nbsp;Lijuan Li","doi":"10.1016/j.swevo.2025.102191","DOIUrl":"10.1016/j.swevo.2025.102191","url":null,"abstract":"<div><div>Computation inspired by natural phenomena, known as bio-inspired algorithms, is one of the main research directions in natural computing. P system optimization algorithms (POAs), sometimes also called the membrane algorithm, are a branch of bio-inspired algorithms. In light of the fact that they have rigorous and sound theoretical development, as well as providing a parallel distributed framework, POAs have become an emerging class of distributed computing models inspired by the structure and function of biological cells. With P systems developing steadily and more of their variant algorithms being published, new membrane structures and intra-membrane rules continue to appear, boosting the flexibility of P systems. In this paper, we conduct a systematic review of POAs to clarify their development context, application scenarios, and future directions, with the specific work arranged as follows. Firstly, the concepts of the membrane computing model are introduced; secondly, the algorithmic structure and algorithmic procedure of POAs are generalized, followed by a summary and classification of the different POAs’ variants in the light of current literature works. Then, the application areas of POAs are categorized and summed up. Finally, the current issues of POAs and potential future directions of their development are discussed.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102191"},"PeriodicalIF":8.5,"publicationDate":"2025-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145362346","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":"An improved dynamic multi-objective robust evolutionary algorithm and application based on PSTL","authors":"Zhongqiang Wu,&nbsp;Mingyang Liu","doi":"10.1016/j.swevo.2025.102195","DOIUrl":"10.1016/j.swevo.2025.102195","url":null,"abstract":"<div><div>A problem whose optimal solution evolves as environmental parameters change is known as a dynamic multi-objective optimization problem (DMOP). Commonly used approaches to DMOP are generally grouped into two categories: Dynamic Multi-Objective Evolutionary Algorithm (DMOEA) and the Dynamic Multi-Objective Robust Evolutionary Algorithm (DMOREA). DMOEA tracks the dynamic Pareto optimal solution through the dynamic response strategy, but it will lead to a high switching cost. DMOREA looks for robust solutions that is suitable in multiple environments, but the optimization effect is poor. To solve these problems, an improved dynamic multi-objective robust evolutionary algorithm based on preliminary search and transfer learning is proposed. Firstly, the preliminary search strategy is used to generate a high-quality target domain guiding population to avoid the occurrence of negative migration. Transfer learning is used to generate a well-distributed population and accelerate the convergence speed. Then, a switching strategy based on the severity of environmental change is proposed, which evaluates the applicability of DMOREA's robust solutions in future environments, switching between solutions generated by preliminary search and transfer learning or existing robust solutions. The proposed strategy improves the optimization effect of the algorithm while maintaining its robustness. The effectiveness of the proposed algorithm is verified by comparison with other algorithms.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102195"},"PeriodicalIF":8.5,"publicationDate":"2025-10-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145321205","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":"Advanced design for nonlinear photovoltaic system problems: A co-evolutionary framework based on a decomposition approach","authors":"Yujun Zhang ,&nbsp;Wenyin Gong ,&nbsp;Rui Zhong ,&nbsp;Huiling Chen ,&nbsp;Jun Yu ,&nbsp;Junbo Jacob Lian ,&nbsp;Juan Zhao ,&nbsp;Zhengming Gao","doi":"10.1016/j.swevo.2025.102179","DOIUrl":"10.1016/j.swevo.2025.102179","url":null,"abstract":"<div><div>Under complex outdoor environments, accurately estimating the unknown parameters of nonlinear photovoltaic (PV) systems remains a major challenge. Key parameters are often influenced by changing weather conditions such as temperature and irradiance. Although many approaches have been proposed, their reliability often drops when environments shift or computing resources are limited. To address these issues, this paper proposes a knowledge transfer-driven self-adaptive decomposition multi-problem cooperative co-evolutionary framework, named SaCEPV, for parameter estimation. SaCEPV is designed to solve a group of related problems simultaneously, where each problem corresponds to parameter estimation for PV modules under specific temperature and irradiance settings. First, the framework integrates a self-adaptive parameter method that dynamically controls the search behavior. Furthermore, knowledge transfer mechanism based on population dynamic diversity is introduced, which adaptively determines when and what to transfer among related problems by analyzing population evolution characteristics. This mechanism enables effective knowledge sharing across correlated problems. Moreover, to handle the complexity of nonlinear PV models, the framework incorporates a parameter pre-decomposition method that separates model components into linear and nonlinear subcomponents based on the nature of the unknown parameters. Then different estimation strategies are then applied to each component accordingly. To evaluate the effectiveness of SaCEPV, the first multi-problem test suite is constructed for PV parameter estimation, covering multiple PV models under various environmental conditions. Experimental results show that SaCEPV achieves superior accuracy and robustness across all problem instances, highlighting strong potential for real-world PV modeling in diverse scenarios.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102179"},"PeriodicalIF":8.5,"publicationDate":"2025-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145321195","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":"Multi-objective optimization based efficient resource scheduling scheme for infrastructure as a service cloud computing","authors":"Absa. S ,&nbsp;AS Radhamani ,&nbsp;Y. Mary Reeja","doi":"10.1016/j.swevo.2025.102168","DOIUrl":"10.1016/j.swevo.2025.102168","url":null,"abstract":"<div><div>Resource scheduling in Infrastructure as a Service (IaaS) cloud computing faces critical challenges such as inefficient task allocation, prolonged makespan, unbalanced resource utilization, and elevated operational costs due to dynamic workloads and complex multi-objective constraints. Traditional scheduling algorithms often struggle with scalability, real-time adaptability, and efficient provisioning. To overcome these issues, this research introduces a novel evolutionary Multi-Objective-based K-means clustering Hybrid White-Faced Success Capuchin (MOK-HWFSC) algorithm. This hybrid model combines K-means clustering for task grouping, Non-dominated Sorting Genetic Algorithm II (NSGA-II) for multi-objective trade-off optimization, and Hybrid White-Faced Success Capuchin optimization (HWFSC) for adaptive and heuristic-based task scheduling. The HWFSC component integrates white-faced capuchin optimization with success-based optimization to enhance convergence and search efficiency, thereby enabling balanced load distribution and improved scheduling accuracy. CloudSim serves as the simulation platform for evaluating the proposed model, providing a controlled and repeatable environment for performance testing. Experimental results demonstrate that MOK-HWFSC achieves superior performance, attaining resource utilization of 78 % at 300 tasks and 85 % at 600 tasks, outperforming benchmark models. Additionally, the model has significantly low computational overhead, with task scheduling processes completed in 20 ms for 300 tasks and 35 ms for 600 tasks, compared to 45 ms and 55 ms in existing methods. Overall, MOK-HWFSC enhances cloud resource scheduling by optimizing task distribution, minimizing makespan, improving energy efficiency, and ensuring scalable, cost-effective deployment in dynamic IaaS environments.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102168"},"PeriodicalIF":8.5,"publicationDate":"2025-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145321204","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":"Hierarchical heterogeneous ant colony optimization based weight generation for texture classification","authors":"Sreeja N․K","doi":"10.1016/j.swevo.2025.102183","DOIUrl":"10.1016/j.swevo.2025.102183","url":null,"abstract":"<div><div>Texture classification is an important problem in pattern recognition and computer vision. The goal of texture analysis is to represent texture in a model that is invariant to changes influenced by illumination, rotation and noise. Natural images exhibit structures that are highly complex and therefore texture analysis have turned a challenging problem. Moreover, textures in real-time environment contain textural information at varying scales. Local Binary Pattern (LBP) is an effective non-parametric texture operator that encodes the local structure around each pixel. This paper proposes a Similarity based Texture Classification for LBP (STC-LBP) algorithm for classification of texture images. STC-LBP algorithm classifies the query texture image based on the similarity of LBP features. To find an optimal weight that emphasizes the similarity between features, a Hierarchical Heterogeneous Ant Colony Optimization based Weight Generation (HHACOWG) algorithm is proposed. Experiments were performed on five benchmark texture datasets namely KTH-TIPS, Brodatz, CUReT, Outex_TC10 and Outex_TC12 datasets. Experiments reveal that the proposal achieves better classification accuracy when compared to the state-of-art methods. The scale and rotation invariance property of STC-LBP is significantly better compared to the existing texture descriptors. The proposal is also tolerant to noise and illumination changes. The results of experiments were validated using non-parametric statistical tests. The feature dimension of the proposal is significantly less compared to the existing descriptors for texture classification. The time taken for feature extraction is less compared to the existing methods indicating that the proposal is well suited for real-time applications.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102183"},"PeriodicalIF":8.5,"publicationDate":"2025-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145321202","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":"Negative hypervolume improvement assisted infill criterion for multi-objective efficient global optimization and its applications","authors":"Shengguan Xu ,&nbsp;Kaiyuan Yang ,&nbsp;Hongquan Chen ,&nbsp;Jianfeng Tan ,&nbsp;Yisheng Gao","doi":"10.1016/j.swevo.2025.102190","DOIUrl":"10.1016/j.swevo.2025.102190","url":null,"abstract":"<div><div>This study introduces a novel hypervolume enhancement approach derived from negative hypervolume improvement (NHVI) concepts, addressing inherent limitations in conventional strategies that generate extensive zero-gradient regions detrimental to late-stage optimization efficiency. In contrast to traditional methodologies that nullify dominated regions, our proposed strategy systematically calculates negative improvements within these domains. This critical modification transforms problematic zero-gradient plateaus into negatively inclined hypervolume regions that actively drive optimization momentum, effectively accelerating the entire multi-objective optimization process. The research implements this negative hypervolume improvement paradigm within multi-objective Efficient Global Optimization (EGO) frameworks. The method's efficacy is rigorously validated against a comprehensive suite of standard multi-objective benchmarks, challenging many-objective test cases, and an aerodynamic airfoil optimization case study. Across all numerical tests, the proposed algorithm demonstrated statistically significant superiority, and in the engineering application, comparative analysis reveals that the enhanced algorithm produces a 485% increase in Pareto solution density (from 7 to 41 solutions) while maintaining superior solution quality. These empirical results substantiate the strategy's effectiveness in addressing real-world engineering challenges, particularly demonstrating its capacity to improve optimization precision through systematic gradient management. The demonstrated performance enhancements highlight the methodology's practical significance for complex multi-objective engineering applications requiring both computational efficiency and solution quality.</div></div>","PeriodicalId":48682,"journal":{"name":"Swarm and Evolutionary Computation","volume":"99 ","pages":"Article 102190"},"PeriodicalIF":8.5,"publicationDate":"2025-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145321203","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}