Search Results (2,453)

The heat release rate (HRR) indicates the scale of a tunnel fire, and inverting it in real time from ceiling sensors supports fire detection and ventilation control. Purely data-driven (deep learning) models are accurate within the training range but cannot extrapolate to larger fires, whereas a purely physics-based formula is less accurate and fails during the fast-growth transient. This paper proposes a data and knowledge dual-driven HRR inversion method. The data component is an encoder-only Transformer on ceiling thermocouples, and the knowledge component is a slope-corrected plume-scaling inversion. The two are coupled by a training-time soft constraint and an inference-time two-layer gate: a magnitude gate raising the physics weight beyond the training power ceiling, and a steady-state gate down-weighting it during transients. On 24 simulated cases (six slopes × four powers, 0.5–4 MW), data are split by slope and power into mutually exclusive training, validation, and test subsets, the test covering unseen slopes and powers. The method outperforms the physics formula at every power tier; on power extrapolation it far surpasses the pure deep learning model (R² = 0.84), and on slope extrapolation it matches that model (R² = 0.94). The results demonstrate, within the present single-geometry FDS tunnel configuration and the investigated working conditions (0–5% slopes, 0.5–4 MW, t² growth, natural ventilation), that physics-guided gated fusion can improve HRR estimation under a 4 MW single-power extrapolation test while retaining the accuracy of the data-driven model under slope extrapolation; the conclusions are not claimed to be directly transferable to other tunnel configurations. Full article

Groundwater contamination in chemical industrial parks (CIPs) poses a significant global threat due to complex pollutant compositions, high source intensity, and accidental release risks. Traditional manual monitoring methods often fail to capture transient contamination pulses or incipient leakages, necessitating a transition toward intelligent, real-time surveillance. This review comprehensively synthesized the state-of-the-art in intelligent online monitoring and early warning systems for CIP groundwater. First, integrated sensing architectures for high-frequency data acquisition were evaluated, and their capacity to resolve spatiotemporal data gaps within highly heterogeneous industrial environments was critically assessed. Analysis of online early warning platforms demonstrated that integrating fundamental hydrogeological principles and advanced data analytics within Digital Twin platforms significantly enhanced predictive reliability and enabled real-time risk quantification. Furthermore, this integration effectively overcame the inherent limitations of purely data-driven black-box models. Field-scale case studies demonstrated the practical efficacy of these systems in improving early warning precision and source tracking across diverse industrial sites. Finally, the persistent technical bottlenecks of sensor fouling and data silos were identified as primary drivers for proposed future research advancing autonomous self-healing hardware and federated learning protocols. This study provided a comprehensive scientific framework to support proactive, data-driven groundwater protection in high-risk CIPs. Full article

Reinforcement learning (RL) can guide autonomous underwater vehicle (AUV) maneuvering to improve the relative source–target–receiver geometry for bistatic sonar target tracking. However, learning a reliable RL policy typically requires substantial interactions with the environment. This paper proposes a risk-aware hybrid decision-making framework that combines an RL policy with selective receding horizon control (RHC) to reduce policy training requirements. Specifically, soft actor-critic (SAC) serves as the nominal decision maker, while a tracking-risk detector assesses target existence probability and estimation uncertainty. When a high-risk belief state is identified, RHC temporarily overrides the SAC action and performs finite-horizon planning based on predicted tracking uncertainty and acoustic detectability. Monte Carlo tree search is employed to efficiently solve the resulting planning problem. Numerical simulations show that the proposed framework improves tracking performance and reduces target-loss events under limited SAC training budgets. In particular, the hybrid framework using a SAC policy trained for 160,000 interaction steps achieves tracking performance comparable to that of pure SAC trained for 500,000 steps, while requiring only sparse RHC intervention. These results demonstrate that selective online planning provides an effective trade-off among policy training requirements, online computational cost, and target tracking performance. Full article

This paper presents a unified Euler–Lagrange methodology for deriving the dynamic models of wheeled mobile robots (WMRs) across heterogeneous kinematic architectures. The work extends the authors’ previously developed kinematic methodology to the dynamic level, handling both holonomic platforms, where free roller motion eliminates kinematic constraints, and non-holonomic platforms, where pure-rolling conditions impose non-integrable velocity constraints. Through null-space-based reduction, constraint forces are eliminated, yielding control-affine state-space models that preserve structural properties essential for control synthesis, including inertia matrix symmetry, positive definiteness, and skew symmetry of the Coriolis matrix. The methodology is applied to four canonical architectures: differential drive, Ackermann steering, three-wheeled omnidirectional, and four-wheeled Mecanum drive. Distinct dynamic features are observed across topologies with direct implications for controller design. To confirm the correctness and practical utility of the derived models, trajectory tracking experiments are conducted in both ideal simulation and a realistic ROS2/Gazebo environment, with a model-based controller and a model-free strategy. Both successfully track the reference trajectories across all four architectures, with the model-based controller consistently achieving lower tracking errors. Full article

Beveled edges are inevitably formed in micro AT-cut quartz mesa resonators (QMRs) during the photolithography process. This local geometric variation alters the thickness distribution and edge stiffness, which in turn affects the coupling between the primary thickness-shear mode and adjacent parasitic modes. To elucidate the mechanism, we establish a vibration analysis model for the micro AT-cut QMR with beveled edge profiles based on the first-order Mindlin plate theory. The coupling between thickness-shear and flexure is the focus of the investigation. The uniform thickness regions and the beveled edge region are expressed using analytical solutions and power series expansions, respectively. The eigenfrequency equation is then derived through the interface continuity and free boundary conditions. The theoretical frequency spectra show good agreement with finite element results so that the accuracy of the proposed analytical approach is validated. A mode coupling intensity index is constructed based on the discrete Fourier spectrum of the surface displacement in the mesa region, and a modal kinetic energy ratio is introduced to characterize the modal coupling. The influence of structural parameters on the spectral characteristics and mode coupling intensity is analyzed. The results indicate that mesa parameters mainly regulate the resonance frequency and energy trapping of the primary thickness-shear mode. In contrast, due to changes in flexure stiffness of the ends and the reflection angle of elastic waves, the beveled edge parameters affect the coupling intensity between the primary thickness-shear mode and parasitic modes. The optimized parameter combinations lead to low coupling states with nearly pure thickness-shear vibration, while the mesa- and bevel-length related low coupling intervals exhibit better fabrication robustness. Full article

Traffic signal control plays a critical role in mitigating congestion and improving urban mobility, particularly in multi-intersection networks where fixed-time strategies cannot adapt to fluctuating demand. Although reinforcement learning has shown strong potential for adaptive signal optimization, purely learning-based controllers often rely on reward shaping rather than explicit enforcement of traffic engineering constraints, which may lead to unstable phase switching and operational inefficiencies. This study proposes a Hybrid Reinforcement Learning and Rule-based algorithm (Hybrid-RL-RB), a constraint-aware traffic signal control algorithm that combines reinforcement learning with a rule-based supervisory layer for multi-intersection traffic signal control. In the implemented version, the learning component is based on tabular Q-learning with a discretized traffic state representation, while the rule-based layer supervises the final executable signal action. The objective is to improve adaptive signal control while preserving operational feasibility through minimum green time, maximum green time, spillback protection, and phase-safety constraints. The framework was implemented in SUMO through TraCI and evaluated under three scenarios of low, medium, and high traffic demand conditions across multiple network configurations, including a real-network topology (Casablanca-OSM). Experimental results show that Hybrid-RL-RB reduces average queue length by up to 51.47% and waiting time by up to 68.10% compared with Fixed-Time control. Compared with Simple-RL, the proposed method provides modest but consistent queue reductions on the 16 × 16 network, while MaxPressure remains the strongest queue-minimization baseline. In the high-demand Casablanca-OSM scenario, Hybrid-RL-RB reduces queue length by 20.50%, reduces waiting time by 21.41%, and increases throughput by 16.83% compared with Fixed-Time control. These results indicate that explicit rule-based projection can improve the operational feasibility and extensibility of RL-based traffic signal control, although further validation with additional seeds and longer real-network simulations is required. Full article

A novel axiomatic foundation of entropy has recently been proposed, overcoming the limitations of classical and information-theoretic entropy foundations and eventually unifying probabilistic and physical entropy. A new set of postulates leads to a rigorous, uncertainty-based definition of entropy consistent with the principle of maximum entropy. Entropy is thus a purely stochastic concept quantifying uncertainty, thereby enriching Kolmogorov’s probability system. Applied to gas thermodynamics, the new framework reproduces classical results and derives, rather than assumes, the laws of thermodynamics. In atmospheric applications, entropy maximization yields an isothermal state as the molecular equilibrium. Gravitation does not alter the isothermal state but differentiates it from the isentropic one of macroscopic air parcels, whose motion drives the atmosphere away from equilibrium. Radiatively active gases, through interactions with shortwave and longwave radiation, sustain non-equilibrium vertical profiles of the atmospheric variables. Combined with the Stefan–Boltzmann law, these mechanisms provide a simple, parsimonious, and coherent explanation of observed atmospheric behaviors and the climatic system. The framework highlights thermodynamics as emergent from stochastics, offering new insights into molecular uncertainty, emergence of macroscopic structures, and radiation in shaping Earth’s climate. It also suggests a broader stochastic view of nature and atmospheric processes. Full article

This study presents a novel detailed mathematical model of vibrational relaxation in pure methane. Based on the critical analysis of the available experimental data, a new kinetic scheme of vibrational energy exchanges is constructed. A reduced five-process scheme for the bending modes is proposed for comparison with experiments. The state-to-state rate coefficients of vibrational–translational (VT) and vibrational–vibrational (VV) processes are calculated on the basis of the forced harmonic oscillator (FHO) model, which is for the first time applied to CH₄–CH₄ collisions in a unified two-oscillator formulation covering intermolecular and intramolecular energy exchanges. The model parameters are calibrated against experimental relaxation times in the temperature range 140–1100 K. The state-to-state, three-temperature, and two-temperature descriptions of the bending mode relaxation are assessed by solving the isothermal bath problem. It is shown that the three-temperature model yields excellent agreement with the state-resolved solution for the relaxation time. The two-temperature model is valid mainly for moderate and high temperatures. The roles of individual energy transitions in the relaxation are identified, the VT deactivation of the triply degenerate bending mode dominating the relaxation. Full article

Identifying the associations between non-coding RNAs and drug resistance (RDRAs) is crucial for uncovering resistance mechanisms and screening effective biomarkers. However, existing graph-based methods typically rely on purely data-driven learning and commonly assume a simplified noise independence hypothesis, overlooking the complex dependencies among feature, structural, and label noise in biomedical networks. This leads to issues such as spurious associations, poor generalization, and lack of interpretability for noisy association prediction tasks. To address these challenges, we propose Prior-RDRGSE, a prior-knowledge-guided dependency-aware graph learning framework. This framework integrates both dependency-aware graph noise modeling and domain knowledge into graph representation learning. Specifically, we first construct a heterogeneous bipartite graph and employ a deep generative inference encoder to jointly infer the underlying clean graph structure and the association signals, thereby explicitly modeling and purifying the intertwined complex noise within the network. Next, we design a resistance-semantics-conditioned interaction module that injects disease-specific and mechanism-related semantic priors into attention queries, explicitly guiding subnetwork interactions in a biologically plausible manner. Furthermore, we introduce a resistance consistency constraint based on KL divergence, which regularizes model training by aligning the learned association distribution with prior distributions derived from clinical and literature data. Comprehensive experiments demonstrate that Prior-RDRGSE achieves state-of-the-art performance in RDRA prediction and significantly outperforms existing methods. Full article

Multiphase pumps are key equipment for the efficient transport of multiphase fluids in the petroleum industry, and their transient stability under variable-speed conditions directly affects system reliability. By combining numerical simulation with experimental validation, this study systematically investigates the evolution of external characteristics, energy conversion mechanisms, and the dynamic response of the internal flow field during a 0.4 s variable-frequency speed regulation cycle at inlet gas volume fractions (IGVFs) of 10% and 20%. The numerical model was validated against experimental measurements of a four-stage multiphase pump under pure-water steady-state conditions, with deviations in head, efficiency, and power all within 5%. The results show that during acceleration, the increase in hydraulic efficiency at the lower IGVF is greater than that at the higher IGVF; once deceleration begins, IGVF has no significant effect on hydraulic efficiency. At the investigated IGVFs of 10% and 20%, a higher IGVF increases the transient sensitivity of the internal flow field to speed variation, and increasing IGVF suppresses energy conversion in the impeller. The principal novelty of this work lies in the temporal decomposition of impeller work into dynamic and static pressure components during transient speed variation, revealing that static pressure power consistently accounts for more than 50% of the total power throughout the speed regulation cycle. As rotational speed increases, dynamic pressure power rises because the circumferential velocity of the fluid increases with impeller peripheral speed, while static pressure power also increases continuously owing to the enhanced static pressure work of the blades. During deceleration, the impeller’s energy transfer capability weakens with decreasing rotational speed, and both dynamic and static pressure power decline. These findings elucidate the coupled evolution of gas–liquid two-phase flow under variable-speed conditions and provide a theoretical basis for the operational optimization and speed control of multiphase pumps. Full article

When does a complex adaptive system (CAS) (like a thermostat, a market, a flock, or a large language model) become a complex adaptive autopoietic system (CAAS), one that actively produces and sustains itself rather than merely adapting to its environment? We argue that this transition requires two simultaneous conditions. The first, established in a companion work, is architectural sufficiency. The system’s decision structure must achieve recursive closure at the third cybernetic order, the so-called fractal-seed point. The second, derived here, is corporeal viability. The system must pay a structural cost (the enactment tension E) large enough to hold its organization in place against perturbation. Working within Fisher-Generative Informational Realism (FGIR), an informational-realist framework in which information, not matter or energy, is the primary generative substrate of reality, we derive a scalar invariant E = I_pc², where I_p is the integrated informational potential of the operative field and c is the global coherence conductance at which informational structure locks into place. This invariant marks the threshold at which a proto-autopoietic system crosses into full autopoiesis. Because FGIR treats physical mass-energy as the result of a freezing projection acting on an incorporeal informational manifold rather than as the foundation of reality, the same invariant that governs autopoietic closure in social and organizational systems also projects, under appropriate bridge conditions, for instance, onto the classical physical law $\widehat{E}=\widehat{m}\hspace{0.17em}{\widehat{c}}^2$, or in a quantum context, the Schrödinger equation. The physical projection is stated as a conditional equivalence (T-BRIDGE), but it is not the focus of this paper. Rather, the central contribution is the cybernetic threshold itself and its operationalization. We show how a systems theorist can assess a system’s distance from critical admissibility, even when the absolute magnitudes of I_p and c are unavailable in domains that lack R(6) closure; we connect the coercivity threshold explicitly to Ashby’s Law of Requisite Variety, supplying the energetic constraint that Ashby’s purely combinatorial criterion leaves implicit. The contemporary case of large language models illustrates the framework. LLMs display transient, externally-scaffolded R(3)-like decision structure but fail the coercivity threshold, and so remain proto-autopoietic. The paper thus offers systems science a derived criterion, additional to architectural closure, for distinguishing full from proto-autopoiesis, with a specified but as yet unexecuted test program across physical, biological, cognitive, and social domains. Full article

Visible and synthetic aperture radar (SAR) images exhibit substantial nonlinear radiometric differences and geometric deformations due to their fundamentally different imaging mechanisms, making high-precision registration between the two modalities a long-standing challenge in remote sensing image processing. Existing deep learning-based cross-modal registration methods mostly adopt purely convolutional architectures or hybrid convolution–Transformer frameworks, which struggle to achieve a favorable trade-off between long-range dependency modeling and computational efficiency. Moreover, current methods generally rely only on heterogeneous cross-modal supervision for end-to-end training, while overlooking the geometric deformation priors embedded in intra-modal consistency. To address these issues, this paper proposes MTCPNet, a Mamba-based registration network with tri-branch consistency projection for SAR–visible image registration. Specifically, a feature consistency projection module is designed to project SAR and visible images into a modality-invariant shared feature space, with a feature consistency loss introduced to explicitly constrain cross-modal geometric alignment. A Mamba-based hybrid architecture serves as the feature extraction backbone, integrating the linear-complexity long-range dependency modeling of selective state space models with the local contextual representation of window-based self-attention. Under a tri-branch training paradigm, intra-modal consistency supervision and cross-modal matching supervision are jointly incorporated to optimize network parameters. Experimental results demonstrate that MTCPNet consistently outperforms state-of-the-art methods on multiple benchmark datasets, providing a promising solution for high-precision multisource remote sensing image registration. Full article

Industrial water treatment plants are increasingly dependent on cyber–physical systems (CPS) and automated control processes for their operational safety and efficiency. However, the embedding of digital control networks exposes these critical infrastructures to sophisticated cyber–physical attacks, including malicious tampering with chemical dosing units and physical actuators. This paper proposes a robust, AI-enhanced anomaly detection framework designed to identify multi-stage malicious activities in water treatment systems using real-world industrial datasets. The proposed system is developed and validated on the Secure Water Treatment (SWaT) dataset, which contains multivariate sensor and actuator time-series data collected from a fully operational physical testbed under both normal operations and targeted cyber–physical attacks. First, high-frequency sensor noise is filtered, and cross-channel measurement reliability is maximized using a Kalman filter-based sensor fusion module. Subsequently, the fused-state vector is analyzed using an unsupervised Isolation Forest algorithm optimized for high-dimensional boundary isolation. To eliminate false negatives caused by stealthy, low-amplitude data injections that bypass purely statistical models, a deterministic, rule-based verification layer derived from physical process control logic is integrated. By integrating a discrete linear Kalman filter with an unsupervised Isolation Forest and deterministic physical rules, the framework effectively suppresses high-frequency sensor noise, achieving a 67.8% reduction in root mean square error (RMSE), while maintaining high detection accuracy across complex industrial attack scenarios. Experimental results demonstrate that the proposed hybrid framework yields superior detection capability, achieving a Precision of ≈95%, a Recall of ≈93%, a scenario-level F1-score of $94.1\%$ (alongside a sample-level F1-score of $21.5\%$) and an edge inference latency of $0.6 \mathrm{ms}$, effectively demonstrating its suitability for deployment within simulated real-time industrial edge computing environments. The findings further confirm that combining statistical machine learning, state-space sensor fusion, and invariant physical process logic provides a resilient defense paradigm for securing critical industrial infrastructure against modern cyber–physical threats. Full article

Network delays are correlated, so treating hub airports as independent can severely understate simultaneous delay events. This study formulates system-wide flight-delay scenario generation with a pure-state quantum circuit Born machine (QCBM), in which each qubit represents one hub and each computational-basis measurement produces one joint binary delay scenario. Within this unitary pure-state model, a non-entangling circuit yields a factorized distribution; entangling operations therefore provide the resource required to represent cross-hub dependence. Using 24 months of U.S. Bureau of Transportation Statistics data, the five-hub exact-probability finite-difference MMD simulation reduces correlation-matrix error from 0.380 for the independent model to 0.028 ± 0.010. A finite-shot SPSA simulation using 8192 shots per circuit evaluation reaches 0.060 ± 0.029 under ideal sampling and 0.066 ± 0.026 after 0.5% noise-injected training with noiseless post-training evaluation. The independent model underestimates the observed system-wide probability by 6.6-fold for five hubs and 116-fold for 10 hubs. However, second-order classical point estimates lie within the observed five-hub bootstrap interval, and classical models evaluated over the complete 32–1024-state spaces fit efficiently. No quantum advantage or physical-hardware demonstration is claimed. Full article

The standard interpretation of quantum measurement on entangled systems holds that measuring one particle nonlocally collapses the wavefunction of its spacelike-separated partner. We argue that this conclusion rests on a false presupposition: that subsystems of entangled systems possess independent ontic states. If the global wavefunction is the sole ontic object ($\psi$-ontic holism), then for entangled systems, there is no “state of B” to be affected by measurement at A. Measurement is a local dynamical process—concretely modeled by continuous spontaneous localization—that destroys one local wavefunction component at the measurement site; the global state factorizes as a consequence, and subsystem ontology emerges for the first time. The transition of the distant particle’s reduced density matrix from mixed to pure reflects this emergence of separability, not a physical change at the distant location. The framework satisfies no-signaling and embraces the contextuality required by the Kochen–Specker and GHZ theorems. We are explicit about its relation to Bell’s theorem: Bell local causality (factorizability) fails, as it must in any empirically adequate theory, but the failure is confined to outcome independence and is identified with the nonseparability of the global ontic state, while parameter independence—and with it the locality of the dynamics—holds exactly. Decoherence provides the mechanism by which the global wavefunction factorizes and classical separability emerges. The apparent nonlocality of quantum mechanics is thus reinterpreted as nonseparability: the fundamental ontology is holistic, but the dynamics are local. Full article