Intelligent optimization remains a critical challenge in modern engineering design and planning, primarily due to the multimodal, non-linear, and tightly constrained nature of real-world decision spaces. In such environments, balancing global exploration with local exploitation to avoid premature convergence is a well-known difficulty. This study proposes the Elite-Inspired Sociocultural Algorithm (EISA), an optimization approach inspired by sociocultural mechanisms in human societies. EISA integrates intra-group collaboration, inter-group cooperation, and a dual-phase inter-group competition mechanism within a hierarchical structure to maintain a dynamic search balance. The proposed method is evaluated using 23 classical test functions alongside the CEC2019, CEC2020, and CEC2022 benchmark suites. Quantitatively, EISA achieved a leading average Friedman rank (2.22) across the classical functions, demonstrating competitive accuracy and robustness compared to nine advanced algorithms. Furthermore, its practical effectiveness is verified through four real-world engineering design cases and a three-dimensional unmanned aerial vehicle (UAV) path planning problem. Under complex and constrained UAV scenarios, EISA consistently generated feasible safe paths and achieved lower comprehensive path costs (e.g., 5.28 × 103) than the evaluated comparative planners. These results indicate that EISA serves as a reliable and efficient intelligent optimization framework for tackling challenging engineering applications.
The hydrazine oxidation reaction (HzOR) has emerged as a kinetically superior substitute for the oxygen evolution reaction (OER) in energy-efficient hydrogen (H2 ) generation. However, engineering bifunctional catalysts that retain high activity for both H2 evolution reaction (HER) and HzOR under high-currentdensity operation continues to pose a substantial hurdle. Herein, an effective Pt-doping strategy serves as an effective means to manipulate its d-band center, inducing favorable electronic modifications that optimize intermediate binding and simultaneously reduce activation barriers for HER and HzOR. Therefore, the representative 10 %Pt-RuO2 nanofibers (NFs) exhibit remarkable catalytic metrics, with a potential of merely 0.084 V vs. RHE at 10 0 0 mA cm-2 for HzOR and an overpotential of 74 mV at identical current density for HER, surpassing Pt/C by a substantial margin. Long-term durability assessments reveal stable operation over 10 0 0 h under high-current-density HzOR conditions. In a practical overall hydrazine splitting (OHzS) configuration, the electricity requirement for H2 generation stands at only 0.146 kWh m-3 at 100 mA cm-2 , which is a 92 % reduction compared to conventional water electrolysis (1.74 kWh m-3 H2 ). This study offers a viable electronic-structure engineering strategy for designing efficient and durable bifunctional electrocatalysts toward practical energy-saving H2 production. (c) 2026 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.
Traditional thermodynamic vehicle air conditioning models rely on iterative computation and suffer from poor real-time prediction performance on vehicle onboard controllers. Conventional data-driven models have weak generalization capacity and fail to adapt to seasonal variations, extreme high ambient temperature and slow component performance degradation over the full service life of new energy vehicles. To solve this issue, this paper puts forward an AI-enabled adaptive digital twin system tailored for full-lifecycle thermal management. A high-precision neural network substitutes complex thermodynamic iteration modules to rapidly predict evaporator outlet air temperature and total cooling energy consumption. An online twin updating framework fusing factory baseline calibration data and accumulated real vehicle historical data is established, and the model performance is continuously optimized through periodic iterative retraining. Test results show that compared with the static baseline BP neural network, the prediction accuracy of temperature and energy consumption rose by 67 % and 87 %, respectively. This system effectively identifies external environmental disturbances and internal equipment performance drift and maintains stable full-lifecycle adaptive capacity. A lifecycle-oriented dynamic programming (LC-DP) strategy considering solar radiation thermal load is further proposed to cut energy consumption under standardized PMV thermal comfort constraints. Validations under near-real driving cycles demonstrate that LC-DP reduces air conditioning energy use by 4.4 % compared with PID control and 41.3 % compared with traditional on–off control. This research provides targeted technical references for the practical engineering deployment of intelligent thermal management systems for new energy passenger vehicles.
Weaving sections frequently experience congestion caused by complex interactions between merging and diverging vehicles. Although coordination among connected autonomous vehicles (CAVs) can help alleviate such bottlenecks, mixed-traffic conditions present a key challenge: how to manage local interactions between CAVs and connected manual vehicles (CMVs) while simultaneously improving global weaving efficiency. To address this issue, this paper proposes a hierarchical vehicle–infrastructure decision framework for cooperation in weaving section via coalition game (CWS-CG). At the global level, targeting the upstream–downstream density imbalance in the weaving section, a conflict resolution method is proposed to coordinate merging and diverging flows and transmit decisions to CAVs via roadside units (RSUs). At the local level, individual CAVs adapt lane-changing strategies based on coalition game outcomes and real-time traffic density, supported by a payoff compensation mechanism that aligns individual decisions with global objectives. In addition, the farthest game point concept is introduced to explicitly capture the influence of downstream lane-changing behavior on upstream game payoffs. Simulation experiments conducted in SUMO demonstrate that, under medium to high traffic demand and CAV penetration rates exceeding 50%, the proposed CWS-CG framework significantly improves average flow speed (up to 43.62%) and enhances traffic flow stability, without sacrificing safety. The results indicate that integrating coalition game strategies with V2I support provides an effective and scalable solution for mitigating structural congestion in mixed weaving sections.
The resolution of knowledge-intensive multi-objective decision-making poses a fundamental challenge in intelligent electromechanical systems. Active suspension control exemplifies this challenge, requiring the reconciliation of inherently conflicting objectives under complex physical constraints. While Deep Reinforcement Learning (DRL) offers a model-free approach, its application to multi-objective control remains largely knowledge-agnostic: conventional linear reward scalarization neglects the structural mismatch between the algorithm’s frequency-dependent learning dynamics and the system’s physical dynamics, leading to gradient imbalance and inefficient Pareto front exploration. This study introduces the Physics-Informed Pareto-Optimal (PIPO) framework, a novel DRL paradigm that addresses this challenge by establishing a mathematically grounded Frequency-Physics Alignment (FPA) theory. Starting from the Linear Frequency Principle model of neural network convergence dynamics, the FPA theory analytically derives that reward weights must be proportional to the dominant modal frequency of each objective, equalizing the effective gradient contributions of competing objectives throughout training and removing the algorithmic distortion responsible for skewed Pareto front coverage. All reward coefficients, including the FPA-derived priority weights and the physics-grounded normalization scales, are computed deterministically from open-loop plant measurements, constituting a fully knowledge-driven parameter derivation pipeline free of manual hyperparameter selection. Experimental results demonstrate that PIPO significantly enhances the efficiency of this knowledge-intensive task: the PIPO-derived Pareto set achieves a superior hypervolume index and Comprehensive Quality Index score; the controller attains a 41.04 % optimization rate in ride comfort under strict handling-stability constraints, outperforming all evaluated benchmarks. Generalization experiments across road classes A-D, vehicle speeds from 10 to 50 m/s, and a composite urban road profile combining stochastic excitation with superimposed speed-bump and pothole transients demonstrate robustness to non-stationary disturbances beyond the training distribution. Within the offline simulation environment, robustness evaluations confirm performance degradation below 10 % under plant parameter perturbations, below 0.50 % under 10 % sensor noise intensity, and within 5.5 % under 10 % sensor data-loss probability, while hardware-level real-time validation on automotive-grade embedded targets is positioned as a dedicated subsequent research stage. This work establishes a foundational informatics methodology for physics-guided multi-objective DRL, with broad applicability to electromechanical systems characterized by distinct frequency-domain trade-offs.
