Municipal solid waste incineration (MSWI) is an important waste-to-energy technology, but its operation is challenged by nonlinear combustion dynamics, variable waste properties, and the conflicting objectives of high combustion efficiency (CE) and low nitrogen oxide (NOx) emissions. This study proposes a data-driven dynamic optimization framework, termed Koopman-DMOO, that integrates a residual-corrected controlled Koopman surrogate with dynamic multi-objective optimization and a residual-based Dynamic Response Strategy. A time-delay augmented state represents short-term process memory while preserving a leakage-free chronological prediction structure, and the dynamic response mechanism supports Pareto-front tracking under changing operating conditions. The framework was evaluated using dynamic multi-objective benchmark problems and data from a full-scale industrial MSWI facility. Under the same chronological prediction protocol, the residual-corrected Koopman model reduced RMSE by 75.1% for CE and 73.4% for NOx relative to the persistence baseline and achieved lower prediction errors than XGBoost, LSTM, GRU, and TCN. On the FDA1-FDA5 dynamic benchmarks, Koopman-DMOO achieved a mean hypervolume of 0.851, a mean inverted generational distance of 0.022, and the best average rank of 1.1 among the evaluated algorithms. Over the 120-sample chronological industrial evaluation window, the recommended trajectories increased mean physical CE from approximately 99.575-99.724% (+0.149 %age points) and reduced mean NOx from approximately 274.84-239.77 mg/ Nm3 (-35.07 mg/Nm3, approximately 12.76%). The optimized decision trajectories remained within the implemented feasible domain and were smoother than the corresponding historical controls. These results support Koopman-DMOO as a computationally efficient supervisory decision-support framework for dynamic CE-NOx trade-off optimization, while plant-level closed-loop validation remains necessary.
Photoluminescence (PL) enhancement in lead halide perovskites under continuous illumination is commonly attributed to light-induced ion migration and defect healing. Here, we report a significant light-induced PL enhancement in methylammonium lead bromide (MAPbBr3) perovskite thin films, achieved through a simple one-step spin-coating nanocrystal-pinning method. A comparative study of unencapsulated (Pristine) films and films embedded in a poly(methyl methacrylate) matrix (Pvsk@PMMA) under continuous UV irradiation (273 nm) reveals pronounced PL enhancement in both, but with starkly different kinetics and spectral evolution. The Pristine film shows a gradual, sustained increase of ≈ 2352% in PL intensity over 120 min, accompanied by a red-shift and a 42.8% reduction in emission linewidth. In contrast, the Pvsk@PMMA film exhibits a rapid ≈ 123% enhancement that saturates within 15 min, along with a slight blue-shift and a 15.8% linewidth broadening. Based on the combined spectral, kinetic, and structural characterization, we propose that light-driven ion migration and Frenkel-pair annihilation are responsible for defect healing in both cases. Crucially, the PMMA matrix plays a dual role: it passivates surface traps and suppresses grain agglomeration, while interfacial interactions contribute to the observed distinct spectral blue-shift and kinetic saturation. Remarkably, the Pvsk@PMMA film retained 97.91% of its initial PL intensity after 18 months of ambient storage, demonstrating the protective role of the PMMA matrix. This study demonstrates a straightforward pathway for tailoring the intensity and spectral properties of green emission from MAPbBr3 thin films for next-generation optoelectronic devices.
This paper presents a novel hybrid intelligent framework with a Biological-AI approach for fault diagnosis of power transformers based on DGA. The proposed model is inspired by Biologically Inspired Survival (BIS) concepts, where the transformer is considered as a patient and its faults are detected based on gas data. The biological component of the system uses the Cox model, which was initially used in medicine to analyze patient survival. This model enables risk assessment and prediction of the failure risk. The AI component of the system also includes three advanced models. First, DNNs are used for learning complex features. Second, gradient boosting trees are employed for nonlinear modeling. Finally, a meta-learning layer is utilized for the intelligent fusion of predictions. Experimental results on 849 real-world examples demonstrate that this hybrid framework achieves an accuracy of 92.3%, representing an improvement of 24.5% over the best traditional method and 5.1% over the best independent AI model.
The local entropy inequality in continuum thermodynamics is conventionally formulated by associating heat exchange with entropy transfer through the Clausius relation. This formulation, while appropriate for conventional thermal interactions, does not by itself establish that every energy-transfer contribution appearing in an internal-energy balance constitutes Clausius-effective heat. The present work addresses this unresolved thermodynamic attribution by introducing a structural decomposition of exchanged energy into Clausius-effective and non-Clausius contributions. The proposed framework establishes a necessary and sufficient criterion: an exchanged-energy contribution enters the local entropy-transfer term if and only if it admits a reversible thermodynamic representation consistent with the Clausius relation. Accordingly, the total energy balance remains unchanged and retains all energetic mechanisms, whereas the entropy-transfer term is restricted to thermodynamically admissible Clausius-effective exchanges. Non-Clausius contributions—including mechanically mediated dissipation, irreversible deformation, and internal structural energy conversion—remain fully represented in the energy balance and may contribute to internal entropy production through the irreversible evolution of the system, without being identified as Clausius heat. A general identification procedure is formulated based on the separation of thermal and non-thermal energy pathways, reversible and irreversible mechanical contributions, and independent and constrained energetic components. The framework is further connected to the structural viewpoint of Energy Structure Theory and General Energy Behavior (EST-GEB), providing a broader basis for classifying energy-transfer pathways across complex continua. For viscous heat-conducting fluids, the formulation distinguishes Fourier heat transfer from viscous dissipation without conflating a volumetric dissipative power with a heat-flux contribution. The resulting framework preserves the first law and the second law while sharpening the thermodynamic attribution of exchanged energy and providing a systematic basis for extending local entropy analysis to strongly coupled, multiphysics, and nonequilibrium systems.
Double-wall cylindrical shells are widely used in applications where resistance to acoustic loading is critical. These structures often feature mechanical interconnections to ensure structural integrity, load sharing, alignment, and enhanced resilience under dynamic loads. While the transient vibroacoustic behavior of single-walled shells has been extensively investigated, extending such analyses to interconnected double-shell configurations remains challenging due to multiple inter-shell acoustic reflections, and the resultant strong fluid-structure coupling, as well as further complexities introduced by the mechanical links between the shells. This study addresses the transient vibroacoustics of an interconnected piezoelectric double cylindrical shell subjected to an acoustic shock. A fully coupled acoustoelastic model is developed in polar coordinates, incorporating the surrounding medium, the fluid occupying the inter-shell gap, and the fluid inside the inner shell. The governing equations are derived based on Hamilton's principle, Kirchhoff-Love shell theory, Maxwell's electrodynamics, and linear acoustic wave theory, and a semi-analytical solution method is employed to capture the time-domain evolution of acoustic fields and shell vibrations. A key contribution of this work lies in integration of nonlinear vibration absorbers mechanically coupling the shells with piezoelectric actuators to realize an efficient hybrid architecture for controlling the vibroacoustic behavior. A parametric analysis has been carried out to quantify the effect of absorber design as well as active control parameters, yielding an optimized control configuration. Numerical results highlight a pronounced reduction in shock-induced acoustic pressure levels and mechanical stress, demonstrating the effectiveness of the proposed hybrid control strategy for improving structural resilience.
