Search Results (2,761)

With the wide application of battery energy storage systems (BESSs) in DC microgrids, BESSs are facing increasingly severe cyber threats, among which, false data injection attacks (FDIAs) seriously undermine the accuracy of battery state estimation by tampering with sensor measurement data. To address this problem, this paper proposes an improved generative adversarial network (WGAN-GP)-based detection and defence method for FDIAs in battery energy storage systems. Firstly, a more perfect FDIA model is constructed based on the comprehensive consideration of the dual objectives of circumventing the bad data detection (BDD) system of microgrid and triggering the effective deviation of the system operating state quantity; subsequently, the WGAN-GP network architecture introducing the gradient penalty term is designed to achieve the efficient detection of the attack based on the anomalous scores output from the discriminator, and the generator reconstructs the tampered measurement data. Finally, the state prediction after repair is completed based on Gaussian process regression. The experimental results show that the proposed method achieves more than 92.9% detection accuracy in multiple attack modes, and the maximum reconstruction error is only 0.13547 V. The overall performance is significantly better than that of the traditional detection and restoration methods, and it provides an effective technical guarantee for the safe and stable operation of the battery energy storage system. Full article

In modern intelligent buildings, the control of Building Energy Systems (BES) faces increasing complexity in balancing energy costs, thermal comfort, and operational flexibility. Traditional centralized or flat deep reinforcement learning (DRL) methods often fail to effectively handle the multi-timescale dynamics, large state–action spaces, and strict constraint satisfaction required for real-world energy systems. To address these challenges, this paper proposes an energy policy-aware layered control architecture that combines Virtual Energy Storage System (VESS) modeling with a novel Dynamic Constraint-Aware Policy Optimization (DCPO) algorithm. The VESS is modeled based on the thermal inertia of building envelope components, quantifying flexibility in terms of virtual power, capacity, and state of charge, thus enabling BES to behave as if it had embedded, non-physical energy storage. Building on this, the BES control problem is structured using a hierarchical Markov Decision Process, in which the upper level handles strategic decisions (e.g., VESS dispatch, HVAC modes), while the lower level manages real-time control (e.g., temperature adjustments, load balancing). The proposed DCPO algorithm extends actor–critic learning by incorporating dynamic policy constraints, entropy regularization, and adaptive clipping to ensure feasible and efficient policy learning under both operational and comfort-related constraints. Simulation experiments demonstrate that the proposed approach outperforms established algorithms like Deep Q-Networks (DQN), Deep Deterministic Policy Gradient (DDPG), and Twin Delayed DDPG (TD3). Specifically, it achieves a 32.6% reduction in operational costs and over a 51% decrease in thermal comfort violations compared to DQN, while ensuring millisecond-level policy generation suitable for real-time BES deployment. Full article

To address thermal management challenges in CR400BF high-speed EMU electrical cabinets—stemming from heterogeneous component integration, multi-condition dynamic thermal loads, and topological configuration variations—a dual-metric-driven finite element model calibration method is proposed using ANSYS Workbench. A multi-objective optimization function, constructed via the coefficient of determination ($R^2$) and root mean square error ($RMSE$), integrates gradient descent to inversely solve key parameters, achieving precise global–local model matching. This establishes an equivalent model library of 52 components, enabling rapid development of multi-physical-field coupling models for electrical cabinets via parameterization and modularization. The framework supports temperature field analysis, thermal fault prediction, and optimization design for multi-topology cabinets under diverse operating conditions. Validation via simulations and real-vehicle tests demonstrates an average temperature prediction error $\le 10\%$, verifying reliability. A thermal management optimization scheme is further developed, constructing a full-process technical framework spanning model calibration to control for electrical cabinet thermal design. This advances precision thermal management in rail transit systems, enhancing equipment safety and energy efficiency while providing a scalable engineering solution for high-speed train thermal design. Full article

Wind turbine blades feature complex geometries and operate under harsh conditions, including high curvature gradients, nonlinear deformations, elevated humidity, and particulate contamination. This study presents the design and kinematic analysis of a novel climbing robot based on a 10R folding metamorphic mechanism. The robot employs a hybrid wheel-leg drive and adaptively reconfigures between rectangular and hexagonal topologies to ensure precise adhesion and efficient locomotion along blade leading edges and windward surfaces. A high-order kinematic model, derived from a modified Grubler–Kutzbach criterion augmented by rotor theory, captures the mechanism’s intricate motion characteristics. We analyze the degrees of freedom (DOF) and motion branch transitions for three representative singular configurations, elucidating their evolution and constraint conditions. A scaled-down prototype, integrating servo actuators, vacuum adhesion, and multi-modal sensing on an MDOF control platform, was fabricated and tested. Experimental results demonstrate a configuration switching time of 6.3 s, a single joint response time of 0.4 s, and a maximum crawling speed of 125 mm/s, thereby validating stable adhesion and surface tracking performance. This work provides both theoretical insights and practical validation for the intelligent maintenance of wind turbine blades. Full article

In this work, we propose the use of a reinforcement learning (RL) agent for the control of a cascade hydropower plant system. Generally, this job is handled by power plant dispatchers who manually adjust power plant electricity production to meet the changing demand set by energy traders. This work explores the more fundamental problem with the cascade hydropower plant operation of flow control for power production in a highly nonlinear setting on a data-based digital twin. Using deep deterministic policy gradient (DDPG), twin delayed DDPG (TD3), soft actor-critic (SAC), and proximal policy optimization (PPO) algorithms, we can generalize the characteristics of the system and determine the human dispatcher level of control of the entire system of eight hydropower plants on the river Drava in Slovenia. The creation of an RL agent that makes decisions similar to a human dispatcher is not only interesting in terms of control but also in terms of long-term decision-making analysis in an ever-changing energy portfolio. The specific novelty of this work is in training an RL agent on an accurate testing environment of eight real-world cascade hydropower plants on the river Drava in Slovenia and comparing the agent’s performance to human dispatchers. The results show that the RL agent’s absolute mean error of 7.64 MW is comparable to the general human dispatcher’s absolute mean error of 5.8 MW at a peak installed power of 591.95 MW. Full article

Additive manufacturing offers new perspectives for creating complex geometries with improved design features at lower cost and with reduced manufacturing time. It may even become possible to print a micro-turbojet engine in one single print, but then unconventional geometrical constraints on compressor and turbine designs are inevitable. If a radial machine were printed through additive manufacturing as a standalone component, the most logical print direction would be from the radial outlet/inlet to the axial inlet/outlet to ease the process and limit the supports, with limited additional constraints compared to traditional manufacturing methods. If the rotor comprising a radial compressor and turbine needs to be printed in one single print, one of the components will be printed in a direction that is not favorable. In the present work, the radial turbine is considered to be printed in the unfavorable direction, namely, from the axial outlet to the radial inlet. These geometrical constraints orient the geometry towards a mixed-flow configuration with a trailing-edge cutback. Such design features reduce the available design space for improvement and will clearly have an unfavorable impact on performance. Therefore, a multi-disciplinary gradient-based adjoint optimization of the mixed-flow turbine is performed, striving to limit the adverse impact on total-to-total efficiency while respecting the mass flow rate and power matching with the upstream compressor. The structural constraint limits the p-Norm von Mises stress to a maximum threshold based on the material yield strength at the operating temperature. The results show that a satisfactory compromise can be found between manufacturability constraints, material limits and aerodynamic performance. Full article

Background: Tumor-educated platelets (TEPs) represent a promising biosource for non-invasive multi-cancer early detection (MCED). While machine learning (ML) has been applied to TEP data, the integration of explainability to reveal gene-level contributions and regulatory associations remains underutilized. This study aims to develop an interpretable ML framework for cancer detection using platelet RNA-sequencing data, combining predictive performance with biological insight. Methods: This study analyzed 2018 TEP RNA samples from 18 tumor types using seven machine learning classifiers. SHAP (Shapley Additive Explanations) was applied for model interpretability, including global feature ranking, local explanation, and gene-level dependence patterns. A weighted SHAP consensus was built by combining model-specific contributions scaled by Area Under the Receiver Operating Characteristic Curve (AUC). Regulatory insights were supported through network analysis using GeneMANIA. Results: Neural models, including shallow Neural Network (NN) and Deep Neural Network (DNN) achieved the best performance (AUC ~0.93), with Extreme Gradient Boosting (XGB) and Support Vector Machine (SVM) also performing well. Early-stage cancers were predicted with high accuracy. SHAP analysis revealed consistent top features (e.g., SLC38A2, DHCR7, IFITM3), while dependence plots uncovered conditional gene interactions involving USF3 (KIAA2018), ARL2, and DSTN. Multi-hop pathway tracing identified NFYC as a shared transcriptional hub across multiple modulators. Conclusions: The integration of interpretable ML with platelet RNA data revealed robust biomarkers and context-dependent regulatory patterns relevant to early cancer detection. The proposed framework supports the potential of TEPs as a non-invasive, information-rich medium for early cancer screening. Full article

The strong coupling of the six-degree-of-freedom (6-DoF) electro-hydraulic Stewart parallel mechanism manifests as adjusting the elongation of one actuator potentially inducing motion in multiple degrees of freedom of the platform, i.e., a change in pose; this pose change leads to time-varying and unbalanced load forces (disturbance inputs) on the six hydraulic actuators; unbalanced load forces exacerbate the time-varying nature of the acceleration and velocity of the six hydraulic actuators, causing instantaneous changes in the pressure and flow rate of the electro-hydraulic system, thereby enhancing the pressure–flow nonlinearity of the hydraulic actuators. Considering the advantage of artificial intelligence in learning hidden patterns within complex environments (strong coupling and strong nonlinearity), this paper proposes a reinforcement learning motion control algorithm based on deep deterministic policy gradient (DDPG). Firstly, the static/dynamic coordinate system transformation matrix of the electro-hydraulic Stewart parallel mechanism is established, and the inverse kinematic model and inverse dynamic model are derived. Secondly, a DDPG algorithm framework incorporating an Actor–Critic network structure is constructed, designing the agent’s state observation space, action space, and a position-error-based reward function, while employing experience replay and target network mechanisms to optimize the training process. Finally, a simulation model is built on the MATLAB 2024b platform, applying variable-amplitude variable-frequency sinusoidal input signals to all 6 degrees of freedom for dynamic characteristic analysis and performance evaluation under the strong coupling and strong nonlinear operating conditions of the electro-hydraulic Stewart parallel mechanism; the DDPG agent dynamically adjusts the proportional, integral, and derivative gains of six PID controllers through interactive trial-and-error learning. Simulation results indicate that compared to the traditional PID control algorithm, the DDPG-PID control algorithm significantly improves the tracking accuracy of all six hydraulic cylinders, with the maximum position error reduced by over 40.00%, achieving high-precision tracking control of variable-amplitude variable-frequency trajectories in all 6 degrees of freedom for the electro-hydraulic Stewart parallel mechanism. Full article

This study addresses the critical need for accurate tourism demand (TD) forecasting in Bulgaria using economic indicators, developing robust predictive models to navigate post-pandemic market volatility. The COVID-19 pandemic exposed tourism’s vulnerability to systemic shocks, highlighting deficiencies in traditional forecasting approaches. Bulgaria’s tourism industry, characterized by strong seasonal variations and economic sensitivity, requires enhanced methodologies for strategic planning in uncertain environments. The research employs comprehensive comparative analysis of machine learning (ML) and deep machine learning (DML) methodologies. Monthly overnight stay data from Bulgaria’s National Statistical Institute (2005–2024) were integrated with COVID-19 case data, Consumer Price Index (CPI) and Bulgarian Gross Domestic Product (GDP) variables for the same period. Multiple approaches were implemented including Prophet with external regressors, Ridge regression, LightGBM, and gradient boosting models using inverse MAE weighting optimization, alongside deep learning architectures such as Bidirectional LSTM with attention mechanisms and XGBoost configurations, as each model statistical significance was estimated. Contrary to prevailing assumptions about deep learning superiority, traditional machine learning ensemble approaches demonstrated superior performance. The ensemble model combining Prophet, LightGBM, and Ridge regression achieved optimal results with MAE of 156,847 and MAPE of 14.23%, outperforming individual models by 10.2%. Deep learning alternatives, particularly Bi-LSTM architectures, exhibited significant deficiencies with negative R² scores, indicating fundamental limitations in capturing seasonal tourism patterns, probable data dependence and overfitting. The findings, provide tourism stakeholders and policymakers with empirically validated forecasting tools for enhanced decision-making. The ensemble approach combined with statistical significance testing offers improved accuracy for investment planning, marketing budget allocation, and operational capacity management during economic volatility. Economic indicator integration enables proactive responses to market disruptions, supporting resilient tourism planning strategies and crisis management protocols. Full article

Background/Objectives: Early diagnosis of periodontitis remains challenging using traditional clinical methods. This systematic review and meta-analysis evaluated the diagnostic accuracy of artificial intelligence (AI) models trained on non-invasive or minimally invasive biomarkers—including saliva, gingival crevicular fluid (GCF), and immunologic profiles—for diagnosing and classifying periodontitis in human subjects. Methods: A comprehensive search of PubMed/MEDLINE, Scopus, Web of Science, EMBASE, and Cochrane CENTRAL was conducted from database inception to June 2025. Eligible studies used AI or machine learning models with patient-derived biomarker data and reported diagnostic performance metrics. Results: Seven studies were included, employing various AI models such as random forest, artificial neural networks, and gradient boosting. Biomarkers were derived from saliva (n = 4), saliva-derived biomarkers from oral rinse (n = 1), immunologic profiles (n = 1), and tissue-based gene expression (n = 1). Reported area under the receiver operating characteristic (ROC) curve (AUC) ranged from 0.83 to 0.96. Meta-analysis of studies with comparable outcomes showed a pooled sensitivity of 0.89 (95% CI: 0.84–0.93), a specificity of 0.87 (95% CI: 0.80–0.92), and a summary AUC of 0.92. Subgroup analysis revealed that models using salivary biomarkers achieved a higher pooled AUC (0.94) than those using GCF or immunologic markers (AUC: 0.89). Sensitivity analyses excluding studies with unclear bias did not significantly alter pooled estimates, affirming robustness. The overall certainty of evidence was rated as moderate to high. Conclusions: AI-based diagnostic models utilizing salivary, microbiome, or immunologic biomarkers demonstrated quantitatively high accuracy; however, the overall certainty of evidence was rated as moderate to high due to limitations in study design and validation. Full article

The rapid development of marine recirculating aquaculture systems (RASs) worldwide offers an efficient and sustainable approach to aquaculture. However, the slow start-up of the nitrification process under low-temperature conditions remains a significant challenge. This study evaluated multiple start-up strategies for moving bed biofilm reactors (MBBRs) operating at 13–15 °C. Among them, the salinity-gradient (SG) strategy exhibited the best performance, reducing the start-up time by 38 days compared to the control, with microbial richness (Chao1 index) reaching 396 and diversity (Shannon index) of 4.89. Inoculation with mature biofilm (MBI) also showed excellent results, shortening the start-up period by 26 days and achieving a stable total ammonia nitrogen (TAN) effluent concentration below 0.5 mg/L within 132 days. MBI exhibited the highest microbial richness (Chao1 index = 808) and diversity (Shannon index = 5.55), significantly higher than those of the control (Chao1 index = 279, Shannon index = 3.90) and other treatments. The hydraulic retention time-gradient (HRT) strategy contributed to performance improvement as well, with a 24-day reduction in start-up time and a Chao1 index of 663 and a Shannon index is 4.69. In contrast, nitrifying bacteria addition (NBA) and carrier adhesion layer modification (CALM) had limited effects on start-up efficiency or microbial diversity, with Chao1 indices of only 255 and 228, and Shannon indices were both 3.24, respectively. Overall, the results indicate that salinity acclimation, mature biofilm inoculation, and extended HRT are effective approaches for promoting microbial community adaptation and enhancing MBBR start-up under low-temperature marine conditions. Full article

The deformation behaviors of a gradient-structured (GS) Mg-Gd-Y-Zr alloy, prepared via surface mechanical attrition treatment (SMAT), were systematically investigated in comparison with those of a uniform coarse-grained (CG) counterpart by high-temperature tensile tests at high strain rates (≤400 °C and ≥0.01 s⁻¹). The results indicated that the uniform CG samples exhibited high flow stresses and low elongations (43.9% at 400 °C and 0.01 s⁻¹). Their fraction of dynamic recrystallization (DRX) during the hot deformation was very low, and the dislocations accumulated inside the deformed grains formed high residual stresses. Moreover, the solely operated prismatic <a> slips in the coarse grains implied insufficient deformation coordination. These resulted in their low deformability. By contrast, the GS samples formed by SMAT exhibited more stable flow behaviors, showing lower flow stresses and higher elongations (71.9% at 400 °C and 0.01 s⁻¹). The high dislocation density in the severely deformed (SD) layer provided sufficient driving force for DRX, promoting remarkable softening effect during the hot deformation. The grain boundary slip mechanism facilitated by DRX in the SD layer played a significant role in the hot deformation, enhancing the overall plasticity of the GS samples, although the deformed coarse-grained (DCG) layer deformed in a manner resembling that of the CG samples. Full article

Thermal barrier coating (TBC) around film-cooling holes is a key failure location for turbine blade TBC. This study built a numerical model. The model used conjugate heat transfer (CHT) and sequential thermal-stress calculation methods. It analyzed the temperature and stress fields in the TBC around film-cooling holes. The holes had different inclination angles (30°, 45°, and 60°). It also explored the balance between cooling effectiveness and stress at these angles. Results show that increasing the film-cooling hole angle reduces the cooling film coverage area significantly. Cooling effectiveness becomes worse. The temperature field near the holes is complex. Sharp temperature gradients exist there. An inverse temperature gradient appeared in the top coat (TC) layer at the hole exit. Stress in the TBC was analyzed next. Analysis was conducted under rated operating conditions. Analysis was also completed after 500 h of creep under these conditions. Stress concentration around the holes is obvious. At room temperature, Mode I cracks easily form upstream of the holes. Mode II cracks easily form downstream. Under rated conditions, mixed-mode cracks (I + II) easily form downstream. The coating experiences larger stress at room temperature. This means that the coating is more likely to spall during cooling. Increasing the hole angle can reduce stress concentration. It can also lower the chance of crack formation. However, a larger angle increases the normal momentum of the cooling jet. This reduces film coverage. Therefore, after considering both cooling effectiveness and TBC failure, the 45° film-cooling hole is optimal. Full article

Fraudulent online payment operations represent a persistent challenge in digital commerce, particularly in sectors like air travel, where credit and debit card payments dominate. This study presents a novel fraud detection framework tailored to airline ticket purchases, combining a synthetic dataset generator with a modular, customizable feature engineering process. These are two machine learning models—support vector machines (SVMs) and the light gradient boosting machine (LightGBM)—for real-time fraud detection. A synthetic dataset was generated, including a rich set of engineered features reflecting realistic user, transaction, and flight-related attributes. While both models were evaluated using classification-evaluation metrics, LightGBM outperformed SVMs in terms of overall performance with an accuracy of 94.2% and a recall of 71.3% for fraudulent cases. The main contribution of this study is the design of a reusable, customizable feature engineering framework for fraud detection in the airline sector, along with the development of a lightweight, adaptable fraud detection system for merchants, especially small and medium-sized enterprises. These findings support the use of advanced machine learning methods to enhance security in digital airline transactions. Full article

When control room operators encounter mental fatigue, the accuracy of their work will decline. Accurately predicting the mental fatigue of industrial control room operators is of great significance for preventing operational mistakes. In this study, facial data of experimental participants were collected via cameras, and fatigue levels were evaluated using an improved Karolinska Sleepiness Scale (KSS). Subsequently, a dataset of fatigue samples based on facial features was established. A novel early-warning framework was put forward, framing fatigue prediction as a time series prediction task. Two innovative data processing techniques were introduced. Reverse data binning transforms discrete fatigue labels into continuous values through a random perturbation of ≤0.3, enabling precise temporal modeling. A fatigue-aware data screening method uses the 6 s rule and a sliding window to filter out transient states and preserve key transition patterns. Five prediction models, namely Light Gradient Boosting Machine (LightGBM), Gated Recurrent Unit (GRU), Temporal Convolutional Network (TCN), Transformer, and Attention-based Temporal Convolutional Network (Attention-based TCN), were evaluated using the collected dataset of fatigue samples based on facial features. The results indicated that LightGBM demonstrated outstanding performance, with an accuracy rate reaching 93.33% and an average absolute error of 0.067. It significantly outperformed deep learning models. Moreover, its computational efficiency further verified its suitability for real-time deployment. This research integrates predictive modeling with industrial safety applications, providing evidence for the feasibility of machine learning in proactive fatigue management. Full article