Anomaly-based intrusion detection systems (IDS) are commonly evaluated under the assumption that network traffic remains stationary over time. However, in real-world deployments, they are constantly faced with concept drift. This drift may make the overall performance of the detectors, e.g., the F1-score, look fine and mask substantial variations in detector behaviour. The goal of this study is not to conclude if concept drift affects overall performance, but to describe how it affects the anomaly detection performance over time and what are the different failure modes of the behaviour. For this purpose, three datasets for intrusion detection, namely CIC-IoT-2023, Edge-industrial IoT (Edge-IIoTset), and UNSW-NB15, with different class distributions are used and evaluated with a common temporal evaluation framework. The results show that concept drift doesn't lead to a consistent decrease in performance. In contrast, it results in specific behaviour changes, such as total blindness, sensitivity inflation and relatively stable detection. In all experimental conditions, the isolation forest (IF) was the most unstable in terms of behaviours. Its recall was still close to zero on Edge-IIoTset dataset while Gaussian mixture model (GMM) and autoencoder (AE) were able to detect the attacks with near-perfect accuracy in the attack-contaminated segments. The results indicate that the overall performance of an IDS may mask failure modes that are operationally relevant and that a time-based behavioural assessment more realistically assesses IDS reliability in a dynamic cyber security context.
Lightweight concrete (LWC) is widely used in construction due to its lower density, improved thermal insulation, and enhanced fire resistance. This study investigates the utilization of vermiculite and pumice stone as fine and coarse aggregate replacements, respectively, in LWC, together with micro-silica incorporated at 5%, 10%, 15%, and 20% by weight of cement and a superplasticizer dosage of 0.3%. A comprehensive experimental program comprising 32 mix proportions, each tested in triplicate, was conducted to evaluate the influence of these materials on the microstructural and mechanical performance of LWC. The novelty of this study lies in its detailed investigation of the synergistic effects of vermiculite, pumice stone, and micro-silica on key performance parameters, including density, void ratio, water absorption, and mechanical strength. Among all mixtures, mix number 31 (MN31), containing micro-silica, exhibited the best overall performance, achieving the highest compressive strength (CS) of 17.65 N/mm², satisfying the requirements of ASTM C330 for LWC. The MN31 mix also attained a density of 1438.48 kg/m³, a flexural strength (FS) of 2.84 N/mm², and a split tensile strength (STS) of 1.63 N/mm². Furthermore, microstructural characterization using field-emission scanning electron microscopy (FESEM) and energy-dispersive X-ray spectroscopy (EDS) revealed improved matrix integrity and refined pore structure, which contributed to the enhanced mechanical performance of the developed LWC.
Composite materials (CM) are widely used in aerospace, transportation, and civil engineering due to their high specific strength, stiffness, and wear resistance. However, predicting their dynamic mechanical response under non-stationary loading conditions remains a significant challenge. Stress fields often exhibit singularities at corner points, contact discontinuities, and dissimilar material interfaces, which cannot be accurately resolved using classical homogenization approaches or conventional finite element methods (FEM). This paper presents a second-order accurate computational framework for analyzing the behavior of CM within a unified strip-halfplane configuration. Three mechanical scenarios are investigated: impact resistance (Problem A), tensile strength (Problem B), and bending behavior (Problem C). The proposed framework employs an explicit finite difference method (FDM) based on the method of spatial characteristics and a splitting technique. The numerical scheme is implemented in Python 3.12.3 using NumPy and Matplotlib libraries. A dedicated algorithm is developed to compute stress and velocity fields at singular points and is validated through a three-level grid convergence study. The results indicate that the stress concentration ratio at the singular point P reaches 3.0 times the applied boundary load in Problem A. Under tensile loading in Problem B, the normalized maximum normal stress, σ11k , attains a value of 0.63 at point P. Furthermore, sensitivity analysis reveals that the extremal stresses in the half-plane material vary by a factor of 3.0 for strip embedment depths of 5h, 10h, and 15h. Grid convergence is confirmed with a root mean square error (RMSE) of 4.3 × 10⁻³ between the medium and fine grids. The observed order of convergence at the singular point P is 1.2, which is consistent with the second-order accuracy of the trapezoidal bicharacteristic integration scheme. These findings provide quantitative reference data for the design and analysis of composite structures containing strip inclusions subjected to impact, tensile, and bending loads.
Bridges in Basra Governorate are among the most critical infrastructure components, serving as essential links between different regions of the province. Their structural importance becomes even more significant when seismic effects are considered. The 2020 American Association of State Highway and Transportation Officials (AASHTO) code represents one of the latest developments in bridge structural and seismic design. It adopts a performance-based design approach to ensure that bridge behavior under seismic loading meets life-safety requirements and prevents structural collapse. In this context, the present study aims to compare the seismic performance of three existing bridges located at different sites within Basra Governorate. Although all three bridges are multi-span structures, they differ in their beam support systems. The bridges were modelled using ABAQUS, and nonlinear analyses were performed to evaluate their responses under seismic loading conditions. The results indicate that the seismic performance of the bridges varies depending on their structural configurations. Nevertheless, all three bridges satisfy the safety requirements and remain within the permissible seismic performance limits specified by the AASHTO 2020 seismic code. Among the bridges studied, Al-Tarbya Bridge (T. Br.) demonstrated the highest capacity to withstand the maximum expected seismic intensity in the governorate. This superior performance can be attributed to its greater structural rigidity, where the bridge curvature contributes additional stiffness and improves load distribution during seismic events.
Heart failure (HF) significantly contributes to global morbidity and mortality, highlighting the need for strategies for early diagnosis and prognosis. This study investigates the application of machine learning (ML) techniques to improve the accuracy of HF diagnosis and survival analysis (SA). A comprehensive framework was developed by integrating four feature selection methods Kruskal-Wallis, analysis of variance (ANOVA), Relief, and Chi-squared (χ²) to identify the most effective clinical features for classification. Among the evaluated classifiers, the light gradient boosting machine (LightGBM) with noise augmentation achieved the highest classification accuracy of 99.98%. Concurrently, the random survival forest (RSF) model achieved a strong concordance index (C-index) of 0.85, demonstrating its effectiveness in survival prediction. The survival modelling framework further incorporated deep survival (DeepSurv), RSF, and extreme gradient boosting (XGBoost). In addition, the Cox proportional hazards (CPH) model was employed to enable robust prognosis prediction, while dataset augmentation techniques such as synthetic minority over-sampling technique (SMOTE)-based mixup, noise injection, and three-level augmentation were used to enhance model generalisability. The performance was evaluated using several ML models, including random forest (RF), k-nearest neighbours (KNN), gradient boosting machine (GBM), XGBoost, decision trees (DT), and logistic regression (LR). Comparative analysis with existing studies demonstrated the superior accuracy and detection capability of the proposed approach. The findings confirm the significant potential of ML techniques in improving HF patient management and predicting long-term clinical outcomes.
The rapid proliferation of misinformation across online platforms has intensified the need for reliable automated fake news detection systems, particularly those capable of processing multimodal content that combines textual and visual information. This study investigates the effectiveness of attention-based fusion mechanisms, namely self-attention (SA), multi-head attention (MHA), and co-attention (CoAtt), for enhancing multimodal fake news detection through the integration of textual features extracted using bidirectional encoder representations from transformers (BERT) and visual features derived from residual network (ResNet) architectures. Two model configurations, BERT+ResNet18 and BERT+ResNet50, were evaluated using a fine-grained multimodal dataset comprising six news categories: true, satire, misleading, manipulated, false connection, and imposter content. Feature fusion was performed at the representation level, and attention mechanisms were employed to improve cross-modal feature interactions and representation learning. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics. The experimental results demonstrate that MHA consistently outperforms the other attention-based fusion methods and remains competitive with conventional fusion approaches across both model configurations, achieving accuracies of 73% and 74% for the BERT+ResNet18 and BERT+ResNet50 models, respectively. Furthermore, per-class analysis reveals that MHA provides a more balanced classification performance across both majority and minority classes, whereas SA and CoAtt exhibit reduced effectiveness when handling underrepresented categories. Overall, the findings suggest that attention-based fusion strategies, particularly MHA, provide a robust and effective framework for fine-grained multimodal fake news detection.
The efficient operation of tower cranes is critical in the construction and industrial sectors, where precise control of load positioning and swing angle is essential for ensuring safety, productivity, and operational efficiency. This study proposes a model predictive controller (MPC) to regulate the crane cart position while simultaneously minimizing the load swing angle to zero. The performance of the proposed MPC was compared with that of the widely used proportional-integral-derivative (PID) controller for tower crane systems. A series of simulation studies was conducted to evaluate the effectiveness of the proposed control strategy. The results obtained using MATLAB/Simulink demonstrate that the MPC controller outperforms the PID controller in terms of tracking accuracy, response speed, and disturbance rejection capability. Furthermore, the MPC significantly enhances crane positioning accuracy and reduces the swing angle amplitude by more than 72.4% compared with the PID controller, thereby improving operational safety, reliability, and overall efficiency.
Wire electrical discharge machining (WEDM) of tool steels involves a persistent trade-off between machining quality and productivity, as discharge energy conditions simultaneously influence crater formation and material removal behaviour. This study proposes an integrated multi-objective optimisation framework for WEDM of SKD61 tool steel by combining the Taguchi experimental design, analysis of variance (ANOVA), preference selection index (PSI)-based ranking, and Pareto front analysis. The effects of pulse-on time (T-on), pulse-off time (T-off), servo voltage (SV), and wire feed rate (WFR) on surface roughness (Ra) and material removal rate (MRR) were investigated using a Taguchi L27 experimental design. ANOVA results revealed that SV and T-on were the most influential factors affecting Ra, whereas T-on and T-off predominantly governed MRR. The PSI-based optimisation identified the optimal machining parameters as T-on = 2 µs, T-off = 28 µs, SV = 50 V, and WFR = 8 mm/min, yielding the highest PSI value of 0.8337. Pareto front analysis further illustrated the trade-off relationship between Ra and MRR, while comparative and sensitivity analyses using PSI, grey relational analysis (GRA), and the technique for order preference by similarity to ideal solution (TOPSIS) demonstrated stable optimisation performance under varying weighting conditions. Furthermore, confirmation experiments and external validation using six previously unseen parameter combinations showed good agreement with the optimisation and prediction results within the investigated machining range, thereby supporting the practical applicability of the proposed framework. Overall, the integrated PSI–Pareto approach offers an interpretable, robust, and practically applicable strategy for the multi-objective optimisation of WEDM processes within the investigated machining domain.
This research proposes a novel high step-up non-isolated modified Cuk converter (MCC) that integrates switched-capacitor (SC) and active switched-inductor (SI) techniques for renewable energy source (RES) applications. While preserving the key advantages of the conventional Cuk converter, including low capacitor voltage stress, common grounding, and continuous input current, the proposed MCC achieves a significantly higher voltage gain. Furthermore, the converter parameters are optimized using the electric Eel foraging optimization (EEFO) algorithm, resulting in enhanced performance and reduced voltage stress across individual power switches. The paper presents a comprehensive analysis of the operating principles and steady-state characteristics of the proposed converter, demonstrating its reliability and effectiveness. In addition, the parameters of the tilt-integral-derivative (TID) controller are optimally tuned using the EEFO algorithm to improve dynamic performance. The proposed converter topology is modeled and validated using MATLAB/Simulink simulations, and the obtained results are thoroughly analyzed. The converter achieves a peak efficiency of 94.74%, highlighting its excellent operational performance. Comparative analysis with existing converter topologies further confirms the superiority of the proposed approach in terms of voltage gain, efficiency, and overall performance.
Protection of computing resources and software applications against malicious attacks remains a major security concern due to the increasing prevalence and sophistication of cyber threats. The complexity of distinguishing between benign and malicious behavior is further exacerbated by network traffic data characterized by numerous traffic features and diverse attack profiles. This paper presents a bio-inspired intrusion detection system (IDS), termed the tournament binary grey wolf optimization-stacked ensemble (TBGWO-SE). In the proposed framework, tournament binary grey wolf optimization (TBGWO) is employed to identify an optimal subset of features, while a stacked ensemble (SE) classifier is developed by integrating an adaptive Gaussian support vector machine (SVM) and a decision tree ensemble (DTE). For binary attack detection, the proposed IDS achieves classification accuracies of 99.67%, 99.66%, and 99.67% on the KDDTest+, KDDTest-21, and UNSW-NB15 datasets, respectively. For multi-class attack classification, it attains accuracies of 99.66%, 98.98%, and 95.32% on the corresponding datasets. The effectiveness of the proposed model is demonstrated through comprehensive performance evaluation and comparative analysis against representative state-of-the-art methods. Furthermore, the IDS exhibits strong generalization capability, indicating its potential for future extension toward the detection of previously unseen attacks.
