Search Results (351)

This study examines the systemic integration of entrepreneurship education and the culture of self employment within higher education as a component of sustainable institutional reform. Using a comparative analytical approach, it analyzes international practices across five higher education systems. Finland, the United States, Canada, the United Kingdom, and South Korea were selected to represent diverse yet mature models of entrepreneurship education integration. The findings reveal significant variation in the depth and coherence of integration across national contexts. Rather than identifying a single transferable model, the study shows that effective integration depends on the interaction of key institutional dimensions, including policy alignment, curricular embedding, faculty capacity, infrastructure, external partnerships, and impact evaluation. Finland demonstrates the most coherent configuration, while other systems exhibit partial or fragmented integration shaped by contextual factors. The study concludes that entrepreneurship education is most sustainable when embedded as a system-level institutional strategy rather than implemented through isolated initiatives. It offers an analytical framework, supported by an adapted ADKAR change model, to guide context-sensitive reform. For Arab higher education systems, the primary implication is diagnostic, emphasizing contextual adaptation over direct replication. Full article

A model of nanoparticles has been designed to partially resemble self-similar ferroelastic star-like domain textures. Numerical computations have been used to find the equilibrium configurations of magnetisation in such systems. As expected from the symmetry, the self-similar initial states give room to other types of domain structure as a function of the star parameters. When relaxed without an external field, the self-similar pattern mostly turns into a massive vortex in the centre with radially oriented domains in the star’s peripheral arms. In contrast, a random initial state ends up in a configuration of a triple valve with one input and two outputs, or vice versa, analogous to logical gates. A treatment with an in-plane magnetic field always leads to the valve configuration. The triple-valve states turn out stable, whereas the vortex ones are metastable. The results may be in the design of magnetic-based logic devices. Full article

Food waste remains a significant challenge in the European Union, reflecting structural differences across economic sectors and member states. This study examines how macroeconomic conditions relate to sectoral food waste using harmonized Eurostat data for the EU-27, covering five stages of the food chain and three economic indicators: GDP (Gross Domestic Product) per capita, adjusted gross disposable income per capita, and the Harmonized Index of Consumer Prices for food. The research design integrates factor analysis, structural equation modeling, and hierarchical clustering. Results show that income-related variables have a positive, statistically significant effect on overall food waste, particularly in manufacturing and distribution. In contrast, food prices show a negative, statistically non-significant relationship with waste generation. Cluster analysis identifies two statistically distinct country groups; however, substantial internal heterogeneity indicates that these clusters reflect structural economic configurations rather than typological or behavioral categories. The findings suggest that macroeconomic factors partially explain cross-country differences in food waste and support the need for context-sensitive, sector-specific policy interventions. Full article

Accessible and comfortable public transportation is a cornerstone of sustainable and inclusive urban mobility. However, there is a knowledge gap in how interior layout influences riders’ comfort perception under constant occupancy conditions. We conducted a pilot laboratory experiment in Valparaíso, Chile using a full-scale urban bus mock-up. Twenty-five participants each experienced four seating scenarios (yielding 100 total observations per outcome) that varied seat pitch (20, 30, 45 cm) and seat orientation (forward-facing vs. side-facing). Cumulative link mixed models were used to estimate seat pitch and orientation effects on the comfort outcomes, with participant-specific random intercepts. Increased seat pitch dramatically improved comfort ratings (e.g., virtually no participants felt comfortable at 20 cm, whereas nearly all did at 45 cm). Side-facing bench seating (longitudinal orientation) yielded significantly higher comfort, legroom, and ease-of-movement ratings than the forward-facing configuration at ~30 cm pitch (p < 0.001). Within the tested mock-up conditions, the results suggest that seat pitch is a major driver of perceived comfort and in-vehicle usability, and that a side-facing bench layout (tested at ~30 cm spacing) can improve perceived spaciousness relative to forward-facing seating. Because this is a small, non-probability pilot sample and a partial factorial design, these findings should be considered preliminary design sensitivities that warrant validation in larger, in-service studies before informing fleet-wide standards. Full article

This study presents a detailed experimental investigation of wave impact loads generated by a 3D dam break flow over a dry horizontal bed. Three-dimensionality is induced by a rigid obstacle partially blocking the channel, tested in both symmetric and asymmetric configurations. Impact pressures have been measured at three transverse locations on a downstream vertical wall, and peak pressures, rise times, and pressure impulses have been statistically characterized based on repeated experiments until convergence is achieved. The results show that three-dimensional effects significantly modify the spatial distribution and intensity of impact pressures compared to classical 2D dam break cases. In the asymmetric configuration, the obstacle induces strong lateral redirection of the flow, leading to highly impulsive loads at unshielded locations and substantial pressure attenuation in shadowed regions. In contrast, the symmetric configuration produces more uniform pressure distributions with reduced peak values and weaker impulsive behavior. A probabilistic description of pressure peaks, rise times, and impulses is provided. The dataset offers new experimental benchmarks for the validation and calibration of numerical models aimed at predicting wave-induced structural loads in complex three-dimensional impact flows. Full article

The deployment of wireless seismic nodal systems necessitates the efficient identification of optimal locations for sensor installation, considering factors such as ground stability and the absence of interference. Semantic segmentation of satellite imagery has advanced significantly, and its application to this specific task remains unexplored. This work presents a baseline empirical evaluation of the U-Net architecture for the semantic segmentation of surfaces applicable for seismic sensor installation. We utilize a novel dataset of Sentinel-2 multispectral images, specifically labeled for this purpose. The study investigates the impact of pretrained encoders (EfficientNetB2, Cross-Stage Partial Darknet53—CSPDarknet53, and Multi-Axis Vision Transformer—MAxViT), different combinations of Sentinel-2 spectral bands (Red, Green, Blue (RGB), RGB+Near Infrared (NIR), 10-bands with 10 and 20 m/pix spatial resolution, full 13-band), and a technique for improving small object segmentation by modifying the input convolutional layer stride. Experimental results demonstrate that the CSPDarknet53 encoder generally outperforms the others (IoU = 0.534, Precision = 0.716, Recall = 0.635). The combination of RGB and Near-Infrared bands (10 m/pixel resolution) yielded the most robust performance across most configurations. Reducing the input stride from 2 to 1 proved beneficial for segmenting small linear objects like roads. The findings establish a baseline for this novel task and provide practical insights for optimizing deep learning models in the context of automated seismic nodal network installation planning. Full article

This study evaluates the feasibility of Near-Infrared (NIR) spectroscopy combined with chemometric modeling for monitoring twin-screw melt granulation. Lactose monohydrate was used as a model excipient and polyethylene glycol (PEG 6000) (Sistemas Analíticos S.A, Buenos Aires, Argentina) as a meltable binder. Granules were produced under different processing conditions by varying binder content, screw configuration (kneading or conveying elements), and measurement temperature. NIR spectra were acquired on-line on a conveyor belt and analyzed using Principal Component Analysis (PCA) and Partial Least Squares (PLS) regression. The regression models showed excellent predictive performance for PEG 6000 content in lactose-based granules, with coefficients of determination higher than 0.998 for both raw and preprocessed spectral data. PCA successfully discriminated between granulated and non-granulated materials, as well as between granules produced with different screw configurations, demonstrating the sensitivity of the technique to processing conditions and granule formation mechanisms. In addition, spectral differences associated with measurement temperature were detected, with derivative-based preprocessing improving the discrimination between warm and cooled granules. Overall, the results demonstrate that NIR spectroscopy, coupled with multivariate analysis, is a robust and non-invasive tool for real-time monitoring of twin-screw melt granulation, with strong potential to enhance process understanding, control, and product consistency in continuous pharmaceutical manufacturing. Full article

Fault-tolerant control in safety-critical industrial systems demands adaptive responses to equipment degradation, parameter drift, and sensor failures while maintaining strict operational constraints. Traditional model-based controllers struggle under these conditions, requiring extensive retuning and dense instrumentation. Recent safe multi-agent reinforcement learning (MARL) frameworks with control barrier functions (CBFs) achieve real-time constraint satisfaction in robotics and power systems, yet assume comprehensive state observability—incompatible with sensor-hostile industrial environments where instrumentation degradation and contamination risks dominate design constraints. This work presents a safety-aware multi-agent deep reinforcement learning framework for adaptive fault-tolerant control in sensor-lean industrial environments, achieving formal safety through learned implicit barriers under partial observability. The framework integrates four synergistic mechanisms: (1) multi-layer safety architecture combining constrained action projection, prioritized experience replay, conservative training margins, and curriculum-embedded verification achieving zero constraint violations; (2) multi-agent coordination via decentralized execution with learned complementary policies. Additional components include (3) curriculum-driven sim-to-real transfer through progressive four-stage learning achieving 85–92% performance retention without fine-tuning; (4) offline extended Kalman filter validation enabling 70% instrumentation reduction (91–96% reconstruction accuracy) for regulatory auditing without real-time estimation dependencies. Validated through sustained deployment in commercial beverage manufacturing clean-in-place (CIP) systems—a representative safety-critical testbed with hard flow constraints (≥1.5 L/s), harsh chemical environments, and zero-tolerance contamination requirements—the framework demonstrates superior control precision (coefficient of variation: 2.9–5.3% versus 10% industrial standard) across three hydraulic configurations spanning complexity range 2.1–8.2/10. Comprehensive validation comprising 37+ controlled stress-test campaigns and hundreds of production cycles (accumulated over 6 months) confirms zero safety violations, high reproducibility (CV variation < 0.3% across replicates), predictable complexity–performance scaling ($R^2=0.89$), and zero-retuning cross-topology transferability. The system has operated autonomously in active production for over 6 months, establishing reproducible methodology for safe MARL deployment in partially-observable, sensor-hostile manufacturing environments where analytical CBF approaches are structurally infeasible. Full article

Hybrid trolleybuses (HTBs) with in-motion charging (IMC) can extend zero-emission service using existing catenary, but high on-wire charging powers may concentrate loads and accelerate battery aging. We present a data-driven simulation that replays recorded high-resolution Controller Area Network (CAN) logs through a per-vehicle electrical model with (Constant-Current/Constant-Voltage) (CC/CV) charging and a stress-map aging estimator, a configurable partial catenary overlay, and fleet aggregation by simple summation and an iterative node-voltage analysis of a resistor-network catenary model. A parameter sweep across battery sizes, upper state of charge (SoC) bounds, and charging power caps compares a minimal “charge-whenever-possible” policy with a per-vehicle lookahead (“oracle”) policy that spreads charging over available catenary time. Results show that lowering maximum charging power and/or the upper SoC bound reduces capacity fade, while energy-demand differences are small. Fleet load profiles are dominated by timetable-driven concurrency using 40 recorded days overlaid into one synthetic day: varying per-vehicle power or target SoC has little effect on peak demand; per-vehicle lookahead does not flatten the peak. The node-voltage analysis indicates catenary efficiency around 97% and fewer undervoltage events at lower charging powers. We conclude that per-vehicle policies can reduce battery stress, whereas peak shaving requires cooperative, fleet-level scheduling. Full article

Automated insulin delivery (AID) systems depend on their actuators’ behavior since saturation limits, rate constraints, and hardware degradation directly affect the stability and safety of glycemic regulation. In this paper, we conducted an actuator-centric evaluation of five control strategies: Nonlinear Model Predictive Control (NMPC), Linear MPC (LMPC), Adaptive MPC (AMPC), Proportional-Integral-Derivative (PID), and Linear Quadratic Regulator (LQR) in three physiologically realistic scenarios: the first combines exercise and sensor noise to test for stress robustness; the second tightens the actuation constraints to provoke saturation; and the third models partial degradation of an insulin actuator in order to quantify fault tolerance. We have simulated a full virtual cohort under the two-actuator configurations, DG3.2 and DG4.0, in an effort to investigate generation-to-generation consistency. The results detail differences in the way controllers distribute insulin and glucagon effort, manage rate limits, and handle saturation: NMPC shows persistently tighter control with fewer rate-limit violations in both DG3.2 and DG4.0, whereas the classical controllers are prone to sustained saturation episodes and delayed settling under hard disturbances. In response to actuator degradation, NMPC suffers smaller losses in insulin effort with limited TIR losses, whereas both PID and LQR show increased variability and overshoot. This comparative analysis yields fundamental insights into important trade-offs between robustness, efficiency, and hardware stress and demonstrates that actuator-aware control design is essential for next-generation AID systems. Such findings position MPC-based algorithms as leading candidates for future development of actuator-limited medical devices and deliver important actionable insights into actuator modeling, calibration, and controller tuning during clinical development. Full article

Various TinyML models face a constantly challenging environment when running on emerging sixth-generation (6G) edge networks, with volatile wireless environments, limited computing power, and highly constrained energy use. This paper introduces DRL-TinyEdge, a latency- and energy-sensitive deep reinforcement learning (DRL) platform optimised for the 6G edge of adaptive TinyML. The suggested on-device DRL controller autonomously decides on the execution venue (local, partial, or cloud) and model configuration (depth, quantization, and frequency) in real time to trade off accuracy, latency, and power savings. To assure safety during adaptation to changing conditions, the multi-objective reward will be a combination of p95 latency, per-inference energy, preservation of accuracy and policy stability. The system is tested under two workloads representative of classical applications, including image classification (CIFAR-10) and sensor analytics in an industrial IoT system, on a low-power platform (ESP32, Jetson Nano) connected to a simulated 6G mmWave testbed. Findings indicate uniform improvements, with up to a 28 per cent decrease in p95 latency and a 43 per cent decrease in energy per inference, and with accuracy differences of less than 1 per cent compared to baseline models. DRL-TinyEdge offers better adaptability, stability, and scalability when using a CPU < 5 and a decision latency < 10 ms, compared to Static-Offload, Heuristic-QoS, or TinyNAS/QAT. Code, hyperparameter settings, and measurement programmes will also be published at the time of acceptance to enable reproducibility and open benchmarking. Full article

In this paper, a novel 8-bit approximate multiplier, called INVCAM, is proposed in which the inverted partial products (PPs) are summed using approximate 4:2 compressors. This design allows for flexibility in applying approximations, enabling the multiplier to be tuned to the specific accuracy requirements of different applications. By adjusting the number of approximated bits, the multiplier can operate with a better balance between desirable hardware characteristics and acceptable levels of error. Our approach ensures that INVCAM is customizable for a wide range of applications. The results indicate that INVCAM reduces delay, power, and area by up to 21.5%, 70.0%, and 57.6%, respectively, compared to the state-of-the-art (SoTA) approximate multipliers within its mean relative error distance (MRED) range, and by 42.4%, 80.1%, and 68%, compared to an exact multiplier. The efficacy of INVCAM is evaluated in image processing and deep neural network (DNN) applications. The images processed by different configurations of INVCAM have PSNR and SSIM values greater than 28.9 dB and 0.81, respectively, which manifests the acceptable quality of the processed approximate images. In the DNN application, the classification accuracy of the models implemented using INVCAM(7) is within 0.6% of the original model accuracy. When the number of approximate bits is increased to nine, less than 5% accuracy reduction is observed compared to an exact model, while the power-delay-area product of the multiplier improves by 46%. Full article

Incorporating coarse aggregates into ultra-high-performance concrete (UHPC-CA) can reduce material costs, yet reliably predicting its strength-related behavior and overall performance remains challenging. This study examines UHPC-CA through a two-stage orthogonal experimental program comprising 18 mixtures with coarse aggregate, fly ash, and hybrid fiber reinforcements (steel, polypropylene, and composite fibers). Microstructural characterization using scanning electron microscope (SEM) and X-ray computed tomography (X-CT) was conducted to assess interfacial features and crack evolution and to link these observations to the measured mechanical response. Experimentally, fiber reinforcement markedly enhanced post-cracking performance. Compared with the fiber-free control mixture, the optimal hybrid configuration increased flexural strength from 6.9 to 23.5 MPa and compressive strength from 60.1 to 90.5 MPa. The steel–composite fiber system outperformed the steel–polypropylene system, which is consistent with the tighter composite-fiber interfacial bonding observed by SEM/X-CT and supports the feasibility of partially substituting steel fibers. An artificial neural network (ANN) model trained on 50 mixtures and evaluated on 10 unseen mixtures achieved an R² of 0.9703, an MAE of 1.22 MPa, and an RMSE of 2.11 MPa for compressive strength prediction, enabling sensitivity assessment under multi-factor coupling. Overall, the proposed experiment–characterization–modeling framework provides a data-driven basis for performance-oriented mix design and rapid screening of UHPC-CA. Full article

This article aims to describe a novel workflow designed for generating a new family of minimal surface-based lattice structures with improved performance in terms of material budget compared to the well-known cells like Gyroid and Schwartz. The implemented method is based on the iterative resolution of a dynamic model, where proper forces are applied to generate minimal surface lattices, considering the boundary conditions and the constraint configurations. The novelty of the approach is given by the ability to create a minimal surface without resolving the partial differential equation and without knowing the exact minimal surface generative function. The starting geometry used for the lattice generation is the hypercube, parametrized to create different lattice configurations. Creating five different starting geometries and two constraint configurations, ten different lattice cells were created. For the comparison, a representative parameter of the material budget has been introduced and used to define the two best cells. The material budget is crucial for particle accelerator components, sensors, and detectors. These cells have been compared with Gyroid and Schwartz of the same thickness and bounding box, highlighting improvements of a factor of 2.3 and 1.7, respectively, in terms of material budget. The same cells have also been 3D-printed and tested under compression, and the obtained force–displacement curves were compared with those from a finite element analysis, demonstrating good agreement in the elastic region. Full article

This paper designs a hardware-implementable joint denoising and demosaicing acceleration system. Firstly, a lightweight network architecture with multi-scale feature extraction based on partial convolution is proposed at the algorithm level. The partial convolution scheme can reduce the redundancy of filters and feature maps, thereby reducing memory accesses, and achieve excellent visual effects with a smaller model complexity. In addition, multi-scale extraction can expand the receptive field while reducing model parameters. Then, we apply separable convolution and partial convolution to reduce the parameters of the model. Compared with the standard convolutional solution, the parameters and MACs are reduced by 83.38% and 77.71%, respectively. Moreover, different networks bring different memory access and complex computing methods; thus, we introduce a unified and flexibly configurable hardware acceleration processing platform and implement it on the Xilinx Zynq UltraScale + FPGA board. Finally, compared with the state-of-the-art neural network solution on the Kodak24 set, the peak signal-to-noise ratio and the structural similarity index measure are approximately improved by 2.36dB and 0.0806, respectively, and the computing efficiency is improved by 2.09×. Furthermore, the hardware architecture supports multi-parallelism and can adapt to the different edge-embedded scenarios. Overall, the image processing task solution proposed in this paper has positive advantages in the joint denoising and demosaicing system. Full article