NTU Journal of Engineering and Technology

Monitoring transformations in road infrastructure is critical for disaster response and recovery in conflict regions. However, manual analysis of road damage is time-consuming and limited in scale, while the speed of changes requires near real-time monitoring. This paper presents a deep learning framework using Hybrid Spectral Networks (HybridSN) for change detection in road networks during conflicts, applied to the city of Old Mosul, Iraq. The HybridSN model achieved high accuracy in road detection, with recall of 0.990 in 2014 and 0.982 in 2022. Change detection analysis revealed a substantial reduction of 13,249.83 m in total road length from 20,186.93 m in 2014 to 6,937.10 m in 2022, indicating widespread damage. The quantitative results demonstrate the capabilities of the proposed approach in assessing road network changes through conflict.

Nowadays,The danger of cyberattacks grows as technology develops, requiring more advanced detection and prevention methods. With an emphasis on e-mail phishing detection, the study explores the use of machine learning (ML) to improve cybersecurity measures. Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Logistic Regression (LR), and CatBoost are among the ML models that are assessed to determine how well they can differentiate between secure and phishing e-mails. F1-score, recall, accuracy, and precision are among the evaluation measures. The results show that all models perform well, with SVM showing perfect accuracy. These findings highlight the importance of cutting-edge technologies in strengthening cybersecurity defenses against changing cyber threats.

Quantum computing provides a powerful solution by using quantum parallelism and state rotations to perform the analysis at a far superior rate. Key contributions of this study include: proposing a multi-layer Quantum Neural Networks (QNN) architecture where classical weights are replaced with quantum gates to achieve scalability learning; showing possible use of quantum entanglement in QNN to find features as edges of the tumour; and comparing the QNN model to conventional and semi-quantum models to prove the effectiveness of the quantum model. The methodology involves Magnetic Resonance Imaging (MRI) brain tumour datasets and normalizes and augments the datasets to achieve good results. Qubit utilization is optimized through amplitude and basis encoding, while QNN layers employ rotation and entanglement gates. The model is trained with simulators available in the Qiskit environment and is tested on quantum devices (D-Wave). The QNN obtained an accuracy of 97.2% for biometric data; moreover, it had 96.5% precision and 98.1% recall, and an AUC-ROC score of 0.98 was higher than the classical and hybrid models. However, problems with quantum hardware and data encoding remain understudied and form the basis of possible subsequent experimentation in the healthcare area.

White Blood Cells (WBC) contain vital elements for identifying several health complications such as infections, immunity problems and even cancer. Most of the existing traditional Artificial Intelligence (AI) architectures for WBC detection and classification problems rely mainly on model accuracy, while other important aspects such as model size, time for inference, and energy consumption are critical in restricted use cases like mobile devices and edge computing platforms. To this end, this paper proposes a new framework called Multi-Objective Neural Architecture Search (MO-NAS) that aims to search for architectures that optimize for multiple objectives at once – accuracy, model size, inference time, and energy consumption. The performance of the proposed MO-NAS framework was tested using 8,500 AI images acquired from Hiwa Hospital with an overall classification accuracy of 96.4% within five WBC subtypes. Thereby also, the model achieved a faster inference time of 145ms per image, a smaller model size of 32.4 MB from 45MB, and a 25% energy advantage. It provides a path to scientific, high-quality AI models in WBC diagnostics and other medical imaging applications, towards improving individualized medicine and real-time clinical usage.

A two-dimensional numerical simulation was performed using the program (Ansys-Fluent 22.1), for air flow around three circular cylinders to get the knowledge of the characteristics of aerodynamic forces (drag & lift) where they are arranged in tandem at various ratios of spacing (L/D =1, 1.2, 1.5, 2, 3, 45) for constant Reynolds number (Re =1171). This study was done using the turbulence model (k-ω SST). The findings demonstrated that there is vortex simple shedding when spacing ratios where (L/D=1, 1.2, 1.5), vortex shedding begins when spacing ratios (L/D=2, 3), while when spacing ratio (L/D=45) the cylinders behave as if they were a single and independent cylinder because of the big distance of the cylinders from each other. Also, the pressure coefficient values for the first cylinder are greater than the pressure coefficient values for the second and third ones, the drag coefficient values for the first cylinder are positive and greater than the drag coefficient values for the second and third cylinders in all the study cases, the pressure fluctuations and the vortex shedding affecting the lift coefficient values so it oscillates between positive and negative and are close to be zero.

Spur gears made from metal serve as central equipment in multiple mechanical configurations. They experience various defects like fatigue cracks, abrasion, and adhesion wear, pitting and scuffing. This review delivers a detailed analysis and performance review of research material from recent studies regarding spur gear failure modes, together with monitoring techniques and predictive models. Where detailed analysis through real-world examinations is conducted, wind energy applications are combined with automotive and manufacturing sector work environments to evaluate diagnostic system performance in practice. In this paper, both benefits and drawbacks across time-domain, frequency-domain, and time-frequency domain techniques are analyzed. This includes Fast Fourier Transform, empirical mode decomposition, wavelet transform, and Hilbert-Huang transform, as well as contemporary developments in machine learning diagnostic systems. The research sector identifies three main missing elements: limited availability of fault-labeled data, difficulties in maintaining operational condition generalization, and real-time system implementation. Future work and industrial use of spur gear fault diagnosis solutions need guidance to develop robust interpretive fault detection systems at a large operational scale.

The objective of this study is to develop lightweight hollow concrete blocks for use as partition walls instead of load-bearing walls. These blocks aim to improve thermal insulation by reducing heat transfer through cavities of varying dimensions and shapes, thereby minimizing thermal bridges and enhancing thermal resistance. Six foam ratios (1.0, 1.5, 2.0, 2.5, 3.0, and 3.5 L/m³) and seven cement-to-sand ratios (1:1.5 to 1:2.5) were tested. Experimental tests included compressive strength, wet and dry density, water absorption, and thermal conductivity on blocks with different cavity shapes (B0–B6). To evaluate thermal behavior, the blocks were also exposed to heat at 300°C. Wall thicknesses of 30, 35, and 40 mm were examined. Overlapping openings were designed to extend the heat transfer path, improving insulation. The use of foam enhanced thermal performance, reducing energy consumption and supporting sustainability through efficient heat insulation.

Wind energy integration into the grid presents a major challenge to the conventional Permissive Underreach Transfer Trip (PUTT) distance protection scheme. This technical challenge arises from variations in power generated by wind turbines connected to double-circuit transmission lines. This paper proposes an enhanced PUTT scheme to improve the accuracy of the Mho-type distance protection relay in such systems. The Developed PUTT (DPUTT) scheme incorporates compensation techniques for residual and mutual coupling currents in the zero-sequence network. Since variations in wind generation impact the security of conventional compensation methods, the DPUTT scheme compares earth current magnitudes at both ends of the system to distinguish between faulted and healthy circuits. Simulation results from a MATLAB model show the apparent impedance trajectories observed by the relays at both terminals. These results demonstrate the mechanism of the proposed tripping scheme across three protection zones and show improved performance over the traditional PUTT. The supervision signal generated by the DPUTT effectively blocks tripping of healthy terminals while correctly initiating trips for faulted circuits. The simulation confirms that the proposed DPUTT scheme reduces false tripping events by up to 99% compared to PUTT, significantly enhancing protection reliability and system stability under variable wind conditions.

Shape memory alloys (SMAs) have been researched and used in numerous applications. This article discusses a general methodology for the bending behavior of SMA, and the validation of this bending behavior was achieved by reinforcing polyurethane and epoxy polymers with graphene nanotubes using a finite element model. Polyurethane and an epoxy were impregnated with different amounts of graphene volume percent (25, 40, 50, 60, and 75). Some of the mechanical properties that were being studied included tensile strength, strain response, and modulus of elasticity. The results show that with the rise of graphene concentration, reductions in deformation and material are enhanced by stiffness as it changes its ductility and flexibility. An examination of the bending moment of 177 N.mm to determine the variation in the composite stiffness was also done. The results showed that the deformation always decreased with the increase in the quantity of graphene, thus proving the strengthening effect of real bending forces.