The integration of micro-nano hybrid architectures serves as a critical strategy for augmenting the thermal efficacy of ultra-thin heat pipes. In this work, a novel composite heat pipe is fabricated by synergistically combining silicon-based microchannel structures with anodic aluminum oxide (AAO) membranes. The liquid-vapor interface dynamics and thermal performance were systematically investigated under horizontal, positive gravity, and reverse gravity directions. Results show that gravitational assistance enhances capillary-driven flow with increasing inclination angles, accelerating liquid reflux and intensifying interfacial fluctuations in the evaporator. At 90° inclination angle, gravity significantly improves capillary liquid supply capacity, resulting in excellent heat transfer performance. Although HFE-7100 enables rapid startup, anhydrous ethanol demonstrates superior maximum heat load capacity. A higher filling ratio generally improves effective thermal conductivity and temperature uniformity by sustaining a stable liquid film. As the filling ratio rises, the liquid content within the ultra-thin heat pipe increases, accelerating evaporation and condensation rates and significantly lowering the evaporator section temperature. At a filling ratio of 1.7 and a 90° inclination, the maximum effective thermal conductivity reached 1193 W/(m∙K), highlighting the synergistic effect of optimal liquid charging and gravitational assistance on heat transfer enhancement.
Building on our previous work in [1], this paper establishes the local well-posedness of the Schrödinger–KdV system in H−3/16×H−3/4 for the resonant case. Combining this result with those in [1], we obtain local well-posedness for the parameter range max{−3/4,s1−3}≤s2≤min{4s1,s1+2}, and this range is sharp in the sense of the contraction mapping argument.
The reliability of dam-break scour prediction remains sensitive to the calibration of evolving bed resistance and to uncertainty in the associated model parameters. This study develops a probabilistic inversion framework that integrates a two-phase Smoothed Particle Hydrodynamics (SPH) model with Bayesian inference implemented through Markov chain Monte Carlo (MCMC) sampling for uncertainty-aware parameter inference in transient water-soil interaction. The framework captures the progressive loss of bed resistance through a time-dependent degradation of the internal friction angle. It is evaluated against the classical Louvain dam-break experiment using two spatial parameterization strategies: a vertically layered representation and a longitudinally partitioned representation. Bayesian updating improves the reconstruction of the observed transient soil–water interface relative to the deterministic baseline. Posterior inference and multi-chain convergence diagnostics reveal parameter-dependent practical identifiability, with the initial internal friction angle more clearly constrained than the degradation coefficients over the short observation window. A within-experiment temporal holdout and complexity-adjusted model comparison indicate that the longitudinally partitioned model better represents the spatially non-uniform scour response in this benchmark, although its improved morphological fidelity does not ensure reliable estimation of every regional parameter. Overall, the proposed framework provides a tractable route from deterministic calibration toward uncertainty-aware inference in SPH-based scour simulations and a physically interpretable basis for evaluating evolving bed response.
In human-centered flexible manufacturing systems under Industry 5.0, some human operations can be performed through cooperation among multiple workers. The selected worker number affects both processing time and total labor input, so scheduling involves three coupled decision dimensions: operation sequencing, resource assignment, and worker number selection. This study investigates a multi-objective flexible job shop scheduling problem with multi-worker cooperation, processing times that vary with the selected worker number, and transportation time. A mathematical model is formulated to minimize makespan and total labor input while describing the diminishing returns from multi-worker cooperation. To solve this problem, a preference-conditioned heterogeneous graph reinforcement learning method is proposed. The scheduling process is formulated as a Markov decision process, and a heterogeneous graph is used to encode operation precedence, resource states, station worker capacity, and transportation information. Based on this graph representation, a hierarchical policy network is designed with an operation-resource pair selection actor and a worker number selection actor to handle the additional worker number decision for human operations. Furthermore, the objective preference weight is embedded in both the state representation and the reward function, enabling a single model to generate multi-objective schedules under different objective preferences. Experiments on 130 instances constructed from a welding and assembly workshop show that the proposed method has advantages over ten classic and recent algorithms across multiple evaluation metrics. Preference interpolation and ablation analyses show that the preference input and hierarchical policy design help improve solution set quality. An industrial order case study further demonstrates that the method can generate executable schedules and has the potential to make fast decisions after disturbances.
Bus operations are subject to stochastic disturbances, often causing bus bunching and service unreliability. Conventional control strategies, such as bus holding and stop-skipping, may increase passenger delays, while existing bus speed optimization methods are prone to local optima and limited system-level coordination. To address these issues, this study proposes a deep reinforcement learning-based bus speed control framework leveraging an improved twin delayed deep deterministic policy gradient algorithm. The bus operation process is formulated as a Markov decision process, where both bus stops and signalized intersections are treated as decision points. The state incorporates headway deviation, bus speed, and signal information, while the action space consists of segment speeds. The reward function is designed to jointly minimize headway deviation and intersection delay while discouraging overtaking. To improve training stability and policy robustness under stochastic operating conditions, a dual-buffer experience replay mechanism, a dynamically adjusted loss function, and a residual network architecture are incorporated. A data-driven simulation platform built on smart card and automatic vehicle location data from Route 57 in Beijing enables systematic evaluation. Experimental results show that the proposed strategy outperforms both conventional control methods and deep reinforcement learning baselines, reducing headway deviation, average passenger waiting time, and average travel time by up to 49.40%, 34.81%, and 14.29%, respectively, while achieving a higher average operating speed. Furthermore, sensitivity analyses and additional tests on Route 23 verify the strategy’s high stability and robust generalization capability. The proposed strategy can provide support for improving bus service reliability and alleviating the bus bunching problem.
