In recent years, extreme temperatures have gained significant attention in urban studies, leading to the search for various adaptation and mitigation measures. While many studies employ heat-related indicators to assess climate-related health impacts, a better understanding of the multi-dimensional nature of these indicators can enhance their integration into urban policies, planning and design. This research aims to examine various dimensions of urban heat stress in built environments, using a systematic review of scientific articles (n = 146) and consequently, establishing a framework for effectively stratifying examples of related indicators across different dimensions. The results showcase dimensions including demographic, economic, health, urban climate, social, urban morphology, and institutional. However, literature disproportionately emphasizes demographic, health and climate dimensions, while social, urban morphology and institutional ones receive comparatively less attention. On the other hand, the co-occurrence analysis reveals connections among these dimensions and their related indicators, underlining the need for a holistic understanding of heat stress impacts. Additionally, the spatial distribution of the selected papers brings attention to the lack of studies in the regions identified as most exposed according to the Koppen Climate Classification. Accordingly, we advocate for more multidimensional and context-specific studies that bridge existing gaps. This research provides valuable insights for policymakers, planners, urban designers and researchers on advancing the understanding of urban heat stress in built environments and its impact on urban healh.

Multi-class traffic mass evacuation optimization considering non-compliance behavior

Early and accurate diagnosis of Sjögren’s syndrome (SjD) remains a significant challenge due to the disease’s heterogeneous clinical presentation and the high dimensionality of transcriptomic data. Meta-heuristic optimizers are attractive for navigating such landscapes, yet existing algorithms tend either to over-explore or to converge prematurely. To address this, we propose DSMAL, a Differential-Evolution & Slime-Mould Algorithm with adaptive Refresh Local Search, as the first hybrid optimization framework tailored to SjD diagnostics. DSMAL partitions the population into two co-evolving sub-swarms: a Differential-Evolution (DE) cohort that drives broad exploration through differential mutation, and a Slime-Mould (SMA) cohort that intensifies exploitation via adaptive position updates. A novel Refresh Local Search (RLS) operator periodically re-diversifies both cohorts, mitigating stagnation without sacrificing convergence speed. For SjD diagnostics, DSMAL is integrated with an XGBoost classifier, forming the DSMAL-XGBoost model, which is trained and validated using gene expression profiles derived from three GEO datasets (GSE23117, GSE40611, and GSE84844). DSMAL optimizes XGBoost hyperparameters in a cross-validation loop to identify the most predictive feature set and classifier configuration. The final model is evaluated using an independent external test set (GSE7451) and demonstrates superior diagnostic performance, achieving 96.6% F1-score, 96.4% recall, 95.0% precision, and 97.6% AUC. Benchmark testing on the IEEE CEC 2021 suite further validates DSMAL’s theoretical strengths, outperforming ten state-of-the-art optimizers in both accuracy and computational efficiency. These findings underscore the potential of DSMAL-XGBoost as a robust tool for transcriptomic-based SjD diagnosis, with broader implications for complex autoimmune disease modeling.

Artificial bee colony and adaptive large neighborhood search based approaches for the star-ring network design problem

Integrated Gate-Commutated Thyristor (IGCT) serves as the core power electronic device in high-voltage and high-power renewable energy conversion systems. Aiming at the problems of slow convergence, easy to fall into local optima, and difficulty in balancing multi-objective performance in traditional IGCT gate drive design under power fluctuation conditions, this paper proposes an IGCT gate drive optimization method based on the Improved Grey Wolf Optimization (IGWO) algorithm. A multi-objective optimization model is established with switching loss reduction, voltage overshoot suppression, current oscillation attenuation and driving capability guarantee as objectives and gate resistance and driving voltage as optimization variables. The traditional grey wolf algorithm is improved by adaptive weight adjustment and dynamic search step strategies to balance global exploration and local exploitation. Simulation and experimental results show that, compared with the traditional Grey Wolf Algorithm (GWO) and Particle Swarm Optimization (PSO), the convergence speed of IGWO is increased by 40.4% and 51.0%, and the optimization accuracy is improved by 12.7% and 18.1%, respectively. Compared with the conventional empirical design, the optimized drive circuit reduces the switching loss by 31.8%, suppresses the voltage overshoot by 33.7%, decreases the current oscillation by 38.6%, and shortens the driving rise time by 39.3%. The proposed method realizes the automatic and precise tuning of IGCT gate drive parameters, effectively improves the switching performance and operation stability of IGCT under renewable energy fluctuation conditions, and provides a practical intelligent optimization scheme for the high-performance gate drive design of high-power IGCT devices.

Amid tightening emission rules and growing cold-chain demand, ports face complex multi-objective scheduling under dual uncertainties in vessel arrivals and operations. This work develops a multi-objective chance-constrained stochastic MILP model for joint berth, QC, and OPS scheduling. Heavy-tailed operational delays are managed via chance constraints, converting Weibull distributions to time buffers, while convex formulations allow piecewise cargo damage penalties to be computed linearly. A reinforcement learning-based adaptive large neighborhood search (RL-ALNS) algorithm is proposed to solve this NP-hard continuous-time problem, integrating a spatiotemporal decoder and an MDP-based selector to ensure microgrid limits and efficiency. Simulations demonstrate RL-ALNS’s superior Pareto convergence versus conventional heuristics. The model cuts the 95th-percentile tail risk by 46.59% and actual costs by 24.44% under mild delays, compared to deterministic scheduling. Overall, it quantifies the non-linear cost–emission–reliability trade-off, providing a robust tool for port decision-making.

To enhance fleet replenishment efficiency and ensure navigational safety, this paper investigates the Underway Replenishment Routing Problem (URRP), focusing on the dynamic characteristics of receiving ships. Mathematical models for replenishment ship travel time and formation vessel speed adjustment are formulated, explicitly considering navigational state transitions and formation stability (risk control). Consequently, a dynamic route optimization model is constructed to provide intelligent decision support for fleet commanders. An intelligent optimization algorithm, the Hybrid Genetic Algorithm with Adaptive Variable Neighborhood Search (HGA-AVNS), is proposed to solve this model. Computational results demonstrate that the proposed approach outperforms the traditional empirical replenishment strategy, validating its effectiveness in enhancing maritime safety and operational efficiency. Extensive sensitivity analyses further reveal that under the strict premise of maintaining formation stability, appropriately reducing the cruise speed can offset the increase in overall speed over ground (SOG) induced by following ocean currents, thereby preventing systematic time loss. Additionally, fine-tuning the execution timing of sudden tactical turning based on the replenishment ship’s real-time operational status can further maximize overall replenishment efficiency without compromising navigational safety.

Structured physical education (PE) programs are essential for fostering students' physical development, cognitive performance, and emotional well-being in academic settings. This study introduces a novel intelligent decision-making algorithm (IDA) for the dynamic optimization and real-time assessment of college PE programs. The proposed framework synergistically integrates a hybrid genetic algorithm (GA) that is viable global optimizer with pattern search (PS) that is an adaptive local search technique. The exploration in the search domain is performed with GA while PS exploitation in a minimal computational budget is performed by the PS. The system encodes program performance data into chromosome-like representations, enabling nuanced evaluation across academic progress, health indices, and affective-psychomotor domains. A tailored vector of weight factors refines the fitness function to reflect individual learning trajectories and institutional goals. Experimental results demonstrate that the model achieves a high prediction accuracy of 98%, with quantifiable improvements of 151.13% in holistic performance, 19.63% in educational metrics, and 26.7% in psychomotor development as compared with reported results. Comparative models, including Random Forest (RF), Adaptive Neuro-Fuzzy Inference System (ANFIS), and RF regression, that achieved accuracies of 88.21%, 94.49%, and 82%, respectively. The hybrid framework maintained a mean global fitness value of 5.3451 × 10⁻¹² with an average computational time of 1357.04s over 100 runs that is used to validate the reliability of the proposed framework. By enhancing efficiency and reducing computational complexity, this AI-driven evaluation model offers a scalable and intelligent approach for real-time optimization and policy refinement in higher education PE curriculum planning.

ABSTRACT Accurate prediction of reservoir residual oil saturation (SOIL) is hindered by complex nonlinear spatio-temporal dynamics. We propose EAOO-PSO, a cooperative hybrid metaheuristic algorithm that balances exploration and exploitation through dynamic role-based population assignment. The framework integrates an adaptive PSO for global search with AOO’s Scroll and Eject operators for elite-driven exploitation and stagnation escape. Benchmarking on the CEC2022 and CEC2017 suites confirms EAOO-PSO’s superior robustness and computational efficiency in high-dimensional optimization compared to nine state-of-the-art algorithms. When applied to LSTM hyperparameter tuning for SOIL prediction, EAOO-PSO achieved exceptional accuracy (R2 = 0.9461, RMSE = 0.0035), significantly outperforming baseline methods and validating its practical efficacy in complex energy engineering tasks.

Searching for information, objects, or places in virtual reality (VR) is often a cumbersome process that breaks user immersion. Existing interaction techniques like pointing or voice commands are context-dependent, yet current systems fail to help users select the optimal strategy for a given situation. To address this challenge, we introduce an adaptive VR search framework that recommends search strategies based on situational factors. We first propose a taxonomy of five common VR search strategies and construct a factorial dataset linking task types, object properties, and environmental conditions to strategy choices. We then conduct a two-stage study. In Stage 1, we capture users' natural strategy selections in a free-search environment to create a ground-truth dataset. In Stage 2, we use this data to train and evaluate two adaptive systems, one driven by a machine learning (ML) model and another by a Large Language Model (LLM), against the free-search baseline. Our results show that both adaptive systems significantly reduce the number of attempts and perceived workload compared to free search. While the ML-based system achieved the fastest task completion times, both adaptive approaches were rated as significantly more usable. This work demonstrates that adaptive, context-aware systems can enhance search efficiency and user experience in VR, paving the way for more intelligent and immersive information-seeking interfaces.

Priming of Pop-out (PoP), when a target-defining feature repeats, accelerates visual search. While previous studies highlight the influence of display density-sparse versus dense arrays-on PoP, how display density interacts with memory encoding and retrieval stages remains unresolved. The present study disentangled the contributions of encoding (trial n-1) from retrieval (trial n) and tracked their influence on early orienting (pre-selective) versus late identification (post-selective) processes. Participants searched for a uniquely colored target under blocked and interleaved density regimes, with eye movements and manual responses recorded. By crossing sparse and dense displays across consecutive trials, four transition types were probed to determine where density exerts its effect. Color repetition reduced reaction times and first saccadic latency, and boosted first-saccade accuracy, but only when the current (i.e., the retrieval) display remained sparse. In contrast, dense displays showed fast responses, with the absence of PoP, due to saliency-driven guidance. Pre-selective eye-movement metrics showed robust PoP in sparse retrieval arrays, whereas dense retrieval arrays defaulted to saliency-driven guidance. Post-selective decision times were comparable across conditions, indicating that PoP drives only the early attentional guidance. These results indicate that while encoding reliably forms target templates, it is the retrieval context that switches feature-biasing mechanisms on or off, highlighting a dynamic interplay between memory and bottom-up salience in adaptive search.

The accidental release of chemicals poses significant risks to human life and property, requiring rapid and accurate source seeking. Chemicals released from an intermittent indoor source form narrow, dynamic, and discrete concentration patches, leaving robots with intermittent cues and posing challenges for source seeking. This study proposed an adaptive robot search algorithm that balances exploitation of historical data and exploration of unknown areas. The proposed exploitation strategy utilizes historical data to probabilistically estimate the source location, aiming to guide the robot upstream for faster source localization. The proposed exploration strategy directs the robot to unknown areas, aiming to gather more information by prioritizing points with the highest information gain based on frontier evaluation. A novel time-dependent factor reduces repeated visits to early regions while mitigating exploration bias in later search stages, enabling a dynamic balance between the two strategies when selecting navigation goals. The algorithm was tested in simulated environments with varying airflow speeds, source release cycles, source release duty cycles, and different scenarios. The results demonstrated reliable and excellent performance. The effectiveness of the proposed algorithm was further validated in real-world robot experiments.

Solving electric vehicle routing problem with heterogeneous drones and no-fly zones using enhanced adaptive large neighborhood search algorithm

The rapid growth of e-commerce has increased the complexity of supply chain management, particularly in urban logistics where efficiency and sustainability are critical concerns. In response, this study proposes a selection hyper-heuristic framework for time-dependent green logistics, incorporating key factors such as economic costs, carbon emissions, rider types, real-time traffic conditions, and time-window constraints. To address the complexities introduced by these real-world factors, we design a two-layer distribution model with crowdsourced delivery that covers the flow from city distribution centers to regional hubs and ultimately to end customers. The first layer involves location selection and the delivery process, while the second layer focuses on order allocation and last-mile delivery. For the location selection and order allocation problems, exact optimization models are developed to obtain high-quality solutions. In the delivery process, we integrate Deep Reinforcement Learning to replace the traditional adaptive layer of the Adaptive Large Neighborhood Search algorithm, enabling dynamic and intelligent adjustments during the search. Comparative analyses against existing and traditional methods across various benchmark instances demonstrate the superior efficiency and solution quality of the proposed framework. Simulation experiments based on a real-world road network in China validate the effectiveness of the proposed framework. In addition, the well-trained model can be directly applied to various scenarios, highlighting its strong generalization capability.

A multi-agent deep reinforcement learning driven adaptive construction and search algorithm for time-varying agile Earth observation satellite scheduling

Hot rolling production scheduling (HRPS) is an essential process in modern steel manufacturing. Its effectiveness is influenced by three primary challenges: selecting suitable slabs to maximize efficiency and maintain product quality, adhering to increasingly stringent carbon tax regulations, and managing uncertainties in processing times stemming from fluctuations in rolling speed. This article presents a novel robust HRPS problem under carbon tax regulation (RHRPSP-CTR) that considers slab selection, carbon tax regulation, and uncertain processing times simultaneously for the first time. Based on a budgeted uncertainty set, we develop a robust counterpart model that utilizes a classical dualization scheme and dynamic programming recursive equations to address the challenges associated with evaluating the worst case cost of carbon emissions (CEs) and determining the worst case completion time for each slab caused by processing time uncertainty, respectively. Recognizing the characteristics of slab selection and the high computational complexity in the large-scale RHRPSP-CTR, we propose an adaptive large neighborhood search algorithm incorporating two enhancement strategies: a slab selection rule and a max-min weight update mechanism. Extensive computational experiments demonstrate that the proposed method yields optimal solutions for small-scale problem instances and high-quality, robust solutions for large-scale instances within a relatively short computation time. Moreover, the results indicate that compared with deterministic HRPS schemes, robust HRPS schemes lead to only a slight increase in cost. Notably, a higher carbon tax does not necessarily lead to lower CEs, and a larger uncertainty budget coefficient or uncertainty range does not always result in higher CEs.

In this paper, we investigate energy-aware production scheduling under Time-of-Use (TOU) tariffs, specifically for industries where inventory storage incurs significant energy costs. We develop a mixed-integer programming model that minimises total costs, including production energy costs, storage energy costs, and late-penalty costs. Given the model's complexity, we propose two algorithms: Multicost-Based Adaptive Genetic Algorithm (MBAGA) and Multicost-Based Adaptive Large Neighborhood Search (MBALNS). These algorithms integrate cost-based weighting mechanisms to efficiently balance trade-offs between different cost components. Numerical experiments demonstrate that, given a computational time constraint, our algorithms significantly outperform the exact method in efficiency and solution quality. Compared with three benchmark algorithms for scheduling under TOU tariffs, both MBAGA and MBALNS consistently achieve lower total costs and faster convergence. Sensitivity analyses with respect to key model and algorithmic parameters confirm the robustness of the proposed approaches. Our results highlight the importance of considering energy consumption during storage in scheduling models and offer a scalable and adaptable framework for optimising late-penalty cost and energy cost during production and storage.

Bone remodeling is a highly regulated, hierarchical process critical for maintaining structural integrity and mineral homeostasis. At the nano-scale, the osteocytes orchestrate mechanosensing, signaling, and nutrient transport across the mineralized matrix utilizing their extensive network of cell dendrites. The lacunar-canalicular network (OLCN) houses the cellular components within the matrix. How this network integrates across bone regions formed during different remodeling cycles remains unresolved. How the cellular network is connected across interfaces between different remodeling regions or cement lines is the focus of this exploration: is the network integration merely stochastical occurrences or result of a cued, directed formation process? Using synchrotron-based nano computed tomography (nano-CT), we analyze human bone samples of 35 different patients with sub-micron resolution to characterize canalicular structures around cement lines. The results show the network's ability and affinity to integrate, and the strong influence of local tissue conditions on the degree of integration. We novelly include the structural analysis of canalicular network architecture to interpret underlying formation processes. Besides 'cross-generational' canalicular connections, we identify previously overlooked canalicular loops in newly formed bone near cement lines and interpret these as morphological indicators of a directed, adaptive search for reconnection. The study suggests a mechanism combining random outgrowth and directed progression influenced by local cues. We propose a 'cross-generational' OLCN: a deliberately integrated network that enhances tissue connectivity, functional resilience, and osteocyte survival across temporal remodeling stages. These findings advance the understanding of bone network complexity and introduce canalicular looping as a nano-structural signature of directed formation in bone network architecture.

Classical information retrieval methods face increasing difficulty in handling large-scale, high-dimensional datasets due to the rapid growth of digital content. As feature dimensionality increases, traditional retrieval techniques suffer from high computational complexity, increased noise sensitivity, and reduced retrieval efficiency. This study introduces a new method based on the principles of quantum mechanics for unsupervised feature selection (UFS) known as Adaptive Optical Search for Unsupervised Feature Selection (AOSUFS). This is aimed at exploring high-dimensional data for information retrieval in the absence of labeled data. The new approach is based on a multi-layer search space and a criterion using the mean absolute difference to obtain the optimal feature subsets. AOSUFS is evaluated using the Reuters dataset comprising 12,152 bag-of-words features and is compared with several optimisation algorithms, including Genetic Algorithm, Harmony Search, Particle Swarm Optimisation, Simulated Annealing, and Krill Herd. The results of the experiments show that AOSUFS cuts the dimensionality by 51.4%, leaving only 5,904 features in the feature space. The proposed method achieves the highest mean average precision of 0.251. This is 9 percent higher than the baseline that does not use feature selection. The Mean Average Recall drops to 0.1384. This shows a 73 percent drop. Krill Herd got second place with a MAP of 0.2499. The unfiltered Harmony Search variant got the lowest score. This work presents the first application of adaptive optical search to unsupervised information retrieval, demonstrating improved retrieval effectiveness, reduced computational requirements, and efficient dimensionality reduction for large, sparse datasets.

Artificial intelligence (AI) is rife with optimization problems, from automating feature engineering and hyperparameter tuning to training intricate neural networks. Finding a balance between exploration and exploitation remains a significant challenge, despite the widespread usage of metaheuristic algorithms to tackle these complex black-box problems. We suggest the River Erosion and Deposition Algorithm (REDA), a unique search technique for numerical and engineering optimization, as a solution to this problem. In order to simulate a dynamic equilibrium state, the algorithm incorporates an adaptive search weight that alternates cyclically between local exploitation and global exploration. While a randomized Boolean operator preserves population diversity, its position-updating process incorporates a stochastic recombination of current population members with an elite memory set to direct the search. We used 19 constrained engineering optimization problems and 29 unconstrained CEC2017 benchmark test functions in a systematic validation process to assess REDA's performance. According to experimental results, REDA performs noticeably better than 13 cutting-edge comparator algorithms. The higher performance of REDA, especially in low-dimensional areas, is confirmed by statistical analyses based on the Friedman test and the Wilcoxon signed-rank test. Furthermore, when utilized to detect parameters in a solar system, REDA showed good accuracy and stability. Collectively, these tests verify that the proposed method effectively balances exploration and exploitation in difficult solution domains.