Engineering reports : open access最新文献

A comprehensive review of the current trends, applications, and innovations within the realm of smart fisheries was performed. Particular focus was placed on the integration of advanced technologies such as Artificial Intelligence (AI), Machine Learning (ML), and the Internet of Things (IoT). Recognizing the critical role fisheries play in global economies—especially in developing countries like Bangladesh—this study examines how these technological advancements can tackle urgent issues such as overfishing, disease management, and environmental monitoring. Through an in-depth exploration of recent literature, we highlight successful implementations, pinpoint key knowledge gaps, and outline future research directions. The ultimate aim of this review is to shed light on how smart fishing can enhance sustainability, improve productivity, and strengthen the resilience of the fishing industry.

Urban-fringe zones represent critical regions for forest fire prevention, yet culturally driven fire risks—particularly those induced by ritual activities—remain underexplored. This study proposes a Ritual Ignition Probability Model (RIPM) to decipher the spatiotemporal coupling mechanisms of wildfire risk in the urban–rural transitional areas of Kaifu District, Changsha, China. By integrating multi-source data (2018–2022)—including ritual activity intensity, ecological factors, and resource allocation metrics—the model quantifies the synergistic effects of lunar festival cycles, fuel accumulation dynamics, and delayed response mechanisms. RIPM employs a Bayesian hierarchical framework to address data heterogeneity and incorporates cultural drivers such as ritual activity risk (R) to optimize risk prediction. Empirical validation demonstrates that RIPM improves prediction accuracy by approximately 30% and reduces emergency response time by 69%. Key findings reveal that 68% of historical wildfires originated from ritual activities, with 87% concentrated within a 1-km buffer of urban boundaries. Policy recommendations include dynamic resource allocation (e.g., increasing fire suppression equipment reserves by 1.5× during peak ritual periods) and culturally adaptive governance innovations (e.g., designated e-incineration zones). By bridging cultural practices and ecological vulnerability, this study advances wildfire risk management theory and provides a replicable analytical framework for global urbanizing regions.

In dynamically evolving supply chain networks, identifying high-risk nodes and abnormal behavior patterns is crucial for risk early warning and system stability. Existing methods mainly rely on static graph modeling or discriminative learning, which struggle to capture temporal evolution and often fail to detect “camouflaged normal” risk nodes under feature obfuscation. To address this, we propose the Temporal-Generative Relation-aware Risk Network (TG-RRNet) to systematically tackle key challenges in dynamic high-risk node identification. TG-RRNet first constructs a time-driven heterogeneous dynamic graph sequence, integrating three types of multimodal information including attribute similarity, historical interaction intensity, and historical risk factor to model the structural evolution process. A temporal-aware graph neural network with temporal decay and graph attention extracts dynamic node representations and captures risk propagation paths. To model latent abnormal patterns, a generative anomaly detection module uses a variational autoencoder to learn latent representations and jointly measures potential risks through reconstruction errors and KL divergence. Finally, a multimodal cross-attention mechanism dynamically fuses structured features, graph representations, and unstructured logs to generate unified risk representations for prediction. Experiments on real-world supply chain datasets show that TG-RRNet significantly outperforms state-of-the-art methods in high-risk node identification and anomaly detection, demonstrating strong practical value and generalization. Code is available at: https://github.com/PinmengLi/TG-RRNet.git.

Grinding in conventional air atmospheres is affected by the formation of oxide and passivation layers, which alter friction, material removal behavior, and surface integrity. This study investigates the influence of an oxygen-free atmosphere on surface grinding by eliminating atmospheric oxygen through argon purging and the introduction of an Ar/SiH₄ gas mixture, achieving an extremely low oxygen partial pressure. Four materials with different oxygen affinities (Ti-6Al-4 V, AlSi10Mg, C45 steel, K40-UF) were machined under both air and oxygen-free conditions. Process forces, residual stresses, and surface roughness were evaluated to identify atmosphere-dependent effects. The oxygen-free atmosphere led to reduced normal grinding forces, most notably for the cemented carbide K40-UF, while tangential forces remained largely unchanged. Residual stresses shifted toward more favorable compressive levels for all materials except AlSi10Mg. Surface roughness parameters were mostly unaffected, with measurable changes in Svk and Sk only for Ti-6Al-4 V and minor variations for C45. The results indicate that oxygen suppression reduces friction and modifies surface interaction mechanisms, particularly under higher thermal loads. This study provides a systematic assessment of atmospheric oxygen as an influential process variable in grinding and highlights the material-dependent sensitivity of grinding mechanisms to oxygen-free conditions.

The establishment efficiency of the surrogate model is often affected by the multi-output problem during the establishment process. It is an urgent issue to solve how to establish a multi-output joint surrogate model more quickly while ensuring a certain level of accuracy. In recent years, the advancement of artificial intelligence technology has provided a more efficient measure for establishing a multi-output joint surrogate model. Multilayer perceptron (MLP) is one of the most widely employed deep learning models and is commonly used to establish the surrogate model. How to establish a reasonable MLP surrogate model is the presumption and basis of establishing a surrogate model. Based on a review of the pertinent literature pertaining to MLP as a surrogate model, this paper examines the techniques and methods of MLP establishment. This paper proposes a framework for the establishment of a multi-output MLP joint surrogate model based on the aforementioned techniques and methods, as well as the existing problems associated with its establishment. On the basis of this framework, a surrogate model for the behavior of dam structural is developed. By confirming the model evaluation index, the performance of the surrogate model for dam structural behavior can be determined to be satisfactory. In addition, the feasibility of this framework is demonstrated by comparing it with independent models that establish surrogate models one by one for multi-output.

The variable and unpredictable output from distributed generation (DG) like wind and solar creates new reliability concerns for distribution networks. Integrating DG on a large scale can unbalance the power supply and compromise quality, making accurate reliability assessment essential. This paper puts forward a new assessment method using a Transformer network. The proposed framework integrates physical modeling with deep learning. First, an improved minimum path algorithm is employed to theoretically evaluate system reliability, specifically modeling the load restoration capability of islanded microgrids. The resulting reliability indices are then discretized into specific intervals to construct a labeled dataset. Subsequently, the Transformer network is innovatively applied to learn the mapping between the stochastic output characteristics of DG and these reliability intervals. By transforming the difficult prediction challenge into a classification task, this method effectively overcomes the problem of non-smoothness in reliability data caused by discrete load restoration. We demonstrate the method's effectiveness on Feeder 4 of the IEEE RBTS 6-node test system. The proposed framework achieves fast online prediction, enabling dynamic monitoring, and proactive warnings against operational risks in the grid.

Existing object detection methods remain severely challenged by adverse weather and domain shifts. On the one hand, the significant distribution shift between clean and degraded samples under diverse weather conditions prevents models from robustly learning intrinsic object representations. On the other hand, drones are distant from objects, and even slight degradation may lead to significant loss of details. There is a lack of a unified and effective all-weather detection framework. To this end, a unified object detection method with degradation-aware and domain adaptive modeling is proposed. First, we design a degradation-aware module (DAM) that leverages amplitude characteristics in the frequency domain to explicitly model degradation patterns, enabling the detector to perceive various types of image quality deterioration. Second, we propose a domain-aware attention-based restoration expert system (DA-RES). It disentangles shared and domain-specific representations through a combination of domain-shared and domain-specific encoders, thereby suppressing category-irrelevant information while enhancing domain-specific useful cues. Finally, through embedding the degradation patterns identified by DAM into the target domain encoder, DA-RES performs multiscale feature restoration guided by degradation priors, thereby boosting downstream detection tasks against adverse conditions. Extensive experiments demonstrate that the proposed framework achieves robust detection performance under all-weather conditions, particularly in challenging degraded scenarios.

In the pursuit of lightweight, high-strength materials for automotive and aerospace applications, the improvement of hybrid metal matrix composites (MMCs) has gained significant attention. This work investigates the mechanical and microstructural characteristics of LM26 aluminum alloy reinforced with varying weight percentages (2–8 wt.%) of silicon carbide (SiC) and graphite particles using the stir casting method. The aim is to enhance the performance of conventional aluminum alloys by incorporating the synergistic effects of ceramic (SiC) and solid lubricant (graphite) reinforcements. The mechanical properties, such as hardness, tensile, compressive, and flexural strength, were evaluated. Mechanical testing revealed that the composite with 6 wt.% reinforcement exhibited maximum performance, with tensile strength of approximately 300 MPa, compressive strength around 480 MPa, flexural strength near 310 MPa, and hardness reaching 162 BHN. Unlike prior studies focusing on single reinforcements, this research systematically explores combined SiC–graphite effects on LM26 composites. SEM indicated relatively uniform dispersion of reinforcements with minimal agglomeration, while EDS and XRD confirmed phase and elemental composition without deleterious phases. An artificial neural network (ANN) model was developed to accurately forecast mechanical properties from reinforcement composition, showing strong predictive capability. The findings provide quantitative benchmarks and enhanced understanding crucial for designing advanced LM26/SiC/graphite hybrid composites for structural, automotive, and aerospace applications.

This study presents a mathematical analysis of electroosmotically modulated peristaltic transport of an Eyring-Powell fluid in a two dimensional microchannel. The walls of channel are propagating sinusoidal waves possesses an electric double layer (EDL) characterized by a constant zeta potential. Under the long-wavelength and low-Reynolds-number regime, the governing equations are simplified and solved analytically. The resulting nonlinear dynamical system is examined through a bifurcation analysis to identify critical points and characteristize their behavior under variations in the fluid flow parameters. Stream function plots and bifurcation diagrams reveal how electrokinetic forces govern flow regime transitions including the formation and destruction of trapped boluses. This work offers significant insight into electroosmotic control of complex biofluids in physiological and microscale pumping.

Characterization and testing of 3D-printed robotic compliant systems for lifespan assessment is time-consuming and costly. For this reason, this work introduces a computer vision approach for automated, non-invasive monitoring of grippers and evaluation of failures. The vision system first detects colored fiducial markers placed on key points of the gripper. The detection model was trained using synthetic data to ensure robustness to background, illumination, and gripper color variations. Then, the marker positions across frames are used to train and detect anomalies in the gripper's displacement. This is performed by thresholding the reconstructed signal over temporal analysis windows, using the reconstruction error as an anomaly score. Validation was performed on real 3D-printed grippers under controlled mechanical failures and uncontrolled lighting and background conditions, correctly classifying over 97% of actions corresponding to normal and anomalous gripper performance. The proposed framework offers a scalable and low-cost alternative to embedded sensors for monitoring gripper performance and detecting early failures.