Journal of Engineering Research and Reports

AI-Driven Anomaly Detection Frameworks in Critical Infrastructure Networks for Proactive Threat Mitigation

Oma Nlerum, Fisayo Fakinlede, Bashiru Ibrahim — Thu, 11 Jun 2026 00:00:00 +0000

Background: Critical infrastructure networks are increasingly interconnected cyber-physical systems, making them more vulnerable to sophisticated cyberattacks that traditional detection methods struggle to address. AI-driven approaches using machine learning and deep learning enable real-time anomaly detection, adaptive response, and predictive resilience, improving security and reliability in these systems.

Aim: This study aimed to systematically to map and synthesize the available AI-based anomaly detection frameworks, which are implemented in a critical infrastructure (CI) network, with an emphasis on typologies of models, data sources and system layers, evaluation practices, and trade-offs in operations that impact the proactive threat mitigation.

Method: A systematic literature review was carried out in accordance with PRISMA framework. Across major academic databases, peer-reviewed research published between 2020 and 2026 was identified. Qualified articles covered the AI/ML-based anomaly detection in the CI setting, such as SCADA, ICS, cyber-physical system, and IT/OT-integrated networks. The types of models, threats, data used, evaluation methods, and constraints in implementation were analyzed using a qualitative comparative synthesis approach.

Findings: The review indicates that supervised statistical machine learning is the most prevalent, and mainly it focuses on network based cyber intrusions, including denial-of-service and malware attacks. Deep hybrid schemes also exhibit greater abilities to model complex and distributed settings but have issues with latency, explanations, and can be deployed in safety-important systems. The data sources are strongly biased towards the telemetry and logs of the IT-layers, and the integration of the SCADA and process-level information is relatively limited. Practices in evaluation are based on benchmark datasets and simulations, and little real-world deployment evidence. Some of the trade-offs that are important are the accuracy versus false alarm rate, latency versus model complexity, and predictive performance versus explainability.

Conclusion and Recommendations: AI-based anomaly detection systems have good technical promise, even though there are few studies that have validated them in high-consequence CI settings. Future applications must focus on deploying in layers, edge aware architecture, human in the loop monitoring, and CI specific evaluation benchmark as well as constant model re-alignment to provide operational resilience and safety assurance.

Finite Element Analysis and Optimization Design of the Metal Structure for a Trough-Making Machine

Xu Zhang — Fri, 29 May 2026 00:00:00 +0000

Trough-making machine is a key special equipment for aqueduct construction, which effectively solves the construction difficulties of long-span, large-volume and high-altitude operation. At present, most trough-making machines are non-standardized equipment designed by empirical methods, leading to material waste and difficulties in controlling structural strength, stiffness and stability. Taking a trough-making machine as the research object, this paper selects the dangerous working condition of full-load pouring under 8-grade wind, calculates wind load and other design loads, establishes a finite element model by ANSYS Parametric Design Language (APDL), and carries out strength and stiffness analysis on its metal structure. On this basis, a zero-order optimization algorithm is used to carry out lightweight optimal design for the main beam, with the minimum volume as the objective function and strength and stiffness as constraints. The results show that the maximum equivalent stress of the whole machine is 216.34 MPa, which is less than the allowable stress of Q355B material; the maximum mid-span deflection of the main beam meets the stiffness limit of L/1000, and all metal structures satisfy the strength and stiffness requirements. After optimization, the mass of the main beam is reduced by about 15.9%, and the overall weight of the trough-making machine is reduced by 5.83%, equivalent to about 54 tons of steel. The optimized structure still meets the design requirements under critical working conditions such as pouring and hole-crossing. The research provides a feasible finite element analysis and structural optimization method for the design of trough-making machines, which can give full play to material performance and realize lightweight design on the premise of ensuring construction safety and pouring quality.

Comparative Analysis of Anti-skid Performance of Smooth and Studded Gripper Shoes for Hard Rock TBMs in Steeply Inclined Shafts

Mingzhao Li, Chenhao Wang — Thu, 04 Jun 2026 00:00:00 +0000

Pumped-storage hydropower is a key clean energy source for grid stability and emergency backup. Tunnel Boring Machines (TBMs) enable safe, automated, and efficient hard-rock excavation with minimal environmental disturbance. To mitigate the critical risk of tunnel boring machine (TBM) slippage and subsequent retrograde falling in steeply inclined shafts, this purely numerical study presents a comparative analysis of the anti-skid performance and mechanical mechanisms of smooth versus studded gripper shoes. Comprehensive multi-body finite element simulation models, coupling the high-strength aluminum alloy (7075-T6) gripper shoe structural matrix with various surrounding rock classes, were developed using HyperMesh and LS-DYNA software. The structural response and anti-skid capacities of both configurations were systematically evaluated under distinct rock conditions, with a specific focus on the potential plastic deformation of the anti-skid studs. The numerical results indicate that under a conservative baseline friction coefficient of 0.3, the anti-skid capacity of the studded gripper shoes in Class II and Class III surrounding rocks increases by 18.5% and 12.1%, respectively, compared to the smooth configuration. This mechanical enhancement is primarily attributed to the additional plowing resistance generated as the studs penetrate the rock mass. Furthermore, the maximum equivalent stresses induced during the full penetration and subsequent shearing displacement within both Class II and Class III rock matrices do not exceed the yield strength of the stud material, demonstrating that structural integrity is maintained without local plastic failure. These findings demonstrate that the optimized lightweight studded gripper configuration significantly enhances gripping stability while minimizing the dead-weight of heavy tunneling apparatus. Cultivating these mechanical insights provides valuable design guidelines and a scalable safety strategy for TBM selection, risk mitigation, and operational safety control in high-inclination underground excavation projects.

Deep Learning-Based Object Detection and Segmentation Methods: A Narrative Review

Yipin Wang — Tue, 26 May 2026 00:00:00 +0000

Object detection and image segmentation are foundational tasks in computer vision, enabling machines to localise, classify, and delineate objects within images and video streams. Over the past decade, deep learning has transformed these fields beyond recognition, delivering performance gains that consistently surpass earlier handcrafted-feature approaches. This article presents a comprehensive narrative review of deep learning-based methods for object detection and segmentation, tracing the evolution from seminal convolutional architectures to contemporary transformer-based frameworks and foundation models. The literature for this review was identified through searches conducted in Web of Science, Scopus, Google Scholar, and PubMed. The review examines two-stage and one-stage detection paradigms, anchor-based and anchor-free detectors, semantic and instance segmentation, panoptic unification, self-supervised representation learning, data augmentation strategies, and transfer learning. Key benchmark datasets and evaluation metrics are discussed, as are applications across autonomous driving, medical image analysis, remote sensing, and small-object detection. The article concludes by identifying persistent challenges—including small-object detection, domain shift, annotation scarcity, and computational efficiency—and by outlining directions likely to define the next phase of progress in the field. The integration of temporal information for video-domain detection and segmentation continues to develop. And the fundamental challenge of robust generalisation across diverse, uncontrolled environments—which benchmarks have never fully captured—remains the field's most important open problem.

Optimization Analysis of Rotating Cuplok Scaffolding System in the Capping Process of an Ultra-Large Diameter Coal Silo: A Critical Review

Zhiyuan Guo — Fri, 29 May 2026 00:00:00 +0000

The construction of ultra-large diameter coal silos presents formidable engineering challenges, particularly during the capping phase, when temporary support structures must span vast distances within constrained, enclosed cylindrical environments. The rotating Cuplok scaffolding system has emerged as an efficient and cost-effective solution to this challenge, enabling the progressive, rotational assembly of silo cap structures without the prohibitive material and labour demands of full-coverage falsework. This critical review synthesises the current state of knowledge across three interconnected domains: the structural mechanics and optimisation of Cuplok modular scaffolding systems; the design requirements and construction constraints of ultra-large diameter silos; and the safety and risk management frameworks applicable to complex temporary structures. Literature searches were conducted across the following bibliographic databases: Web of Science, Scopus, Google Scholar, PubMed (for ergonomic and occupational health dimensions), Engineering Village (Compendex), CNKI (China National Knowledge Infrastructure, for Chinese-language technical literature), J-STAGE (for Japanese-language engineering research), DOAJ (Directory of Open Access Journals), and the structural engineering repository of the ASCE Library. Drawing upon recent peer-reviewed literature spanning structural engineering, construction technology, and safety science, the review identifies principal load-bearing mechanisms, stability-governing parameters, and optimisation strategies relevant to rotating scaffolding configurations. Key findings indicate that second-order geometric non-linearity, joint semi-rigidity, and dynamic load redistribution during rotation are the critical factors determining structural performance. Construction safety is identified as a major concern during non-working transition phases, particularly when the platform is being repositioned between sectors. The review further identifies significant gaps in the literature regarding the specific interaction between rotational kinematics and scaffold structural stability, and calls for more dedicated experimental and computational investigation into this construction scenario. Recommendations are offered for both structural design practice and future research directions, including reliability-based optimisation, integrated digital monitoring, and system-level advanced analysis approaches.

Mixed Reality Systems for Real-time Construction Project Monitoring to Improve Schedule Performance

Ukasha Tiibu Mohammed, Uthman Abba Ndayako — Wed, 03 Jun 2026 00:00:00 +0000

Construction projects continue to face coordination inefficiencies, schedule pressure, and limited visibility of actual site conditions despite increasing investment in digital tools. This scoping review examines the use of mixed reality systems for inspection, real-time monitoring, coordination and schedule-oriented control of construction projects. A PRISMA-ScR-guided process and a PCC-framed review question were used to identify studies published between 2015 and 2025 through the ASCE Library, Scopus, IEEE Xplore, and Web of Science. These studies were then screened and charted in duplicate using a standardised template. Eighteen empirical studies were included, covering buildings, tunnelling, modular construction, bridges, mechanical, interior works, electrical, and plumbing and (MEP) installation. These studies found that XR workflows linked with BIM, SLAM/VSLAM, computer vision, mobile platforms, 4D scheduling, laser scanning, and robotic capture improved the resolution of issues, communication of designs, collaborative planning, visibility of progress, the efficiency of inspections, remote inspections, constructability reviews and the shared understanding of site conditions. Quantitatively, inspection efficiency improved by 78.63% in one study, coordination overhead fell by up to 75% in another, and one MR progress-monitoring system reported 97.3% recall, 96.5% precision, and 95.2% accuracy. However, deployment remains constrained by interoperability issues, user adoption, safety concerns, hardware limitations, and weak integration with formal project control systems. This review consolidates dispersed evidence and identifies priorities for workflow integration, field validation, and governance.

User Preference Segmentation for Movie Recommendation Systems Using Clustering Algorithms

Mirali Mammadzade — Sat, 23 May 2026 00:00:00 +0000

Background: Online streaming platforms generate vast amounts of user interaction data, and recommendation systems use clustering and machine learning techniques to analyze this data and deliver personalized movie suggestions.

Aims: Movie recommendation platforms collect a large amount of information about user activity through ratings and viewing behavior. Since users usually have different interests, recommendation systems may struggle to provide equally accurate suggestions for everyone. The aim of this study is to examine whether clustering algorithms can group users according to their movie preferences and rating habits. The study also explores how unsupervised machine learning methods can help identify hidden behavioral patterns in recommendation data.

Study Design: The research was conducted as a quantitative study using unsupervised machine learning and clustering analysis.

Place and Duration of Study: The study was performed using the MovieLens 25M dataset in duration of 45 days between end of February 2026 and beginning of April 2026.

Methodology: The MovieLens 25M dataset was selected because it contains a large number of movie ratings created by users with different viewing interests. Several preprocessing steps were carried out before model implementation. These steps included removing incomplete records, filtering inactive users with insufficient rating activity, scaling numerical variables, grouping genre-related features, and converting sparse rating information into user-based feature matrices. After preprocessing and activity-based filtering, the final analysis was conducted on 861 active users from the original MovieLens dataset.

Principal Component Analysis was applied to reduce dimensionality and simplify visualization of user groups. Different clustering algorithms were implemented using Python and scikit-learn, including K-Means Clustering, Hierarchical Clustering, Agglomerative Clustering, and DBSCAN. Cosine Similarity was additionally used to compare similarity between users according to their rating behavior. The quality of clustering results was evaluated using Silhouette Score, Davies–Bouldin Index, inertia values, and visual interpretation of cluster separation.

Results: The analysis showed that clustering methods were able to separate users into several meaningful groups based on movie interests and rating behavior. K-Means produced the clearest and most balanced cluster structure among the selected algorithms. Some user groups mainly preferred action, adventure, and science fiction movies, while others showed stronger interest in drama, romance, thriller, and documentary genres. PCA visualization showed visible separation between the major user groups after dimensionality reduction. DBSCAN also identified smaller groups of users with unusual or inconsistent rating activity. Overall, the clustering results helped reveal hidden behavioral differences between users inside the recommendation dataset.

Conclusion: The findings of this study show that unsupervised machine learning techniques can be useful for analyzing user behavior in movie recommendation systems. Clustering methods make it easier to identify users with similar viewing habits and genre preferences. The results also show that clustering quality depends strongly on how user features are prepared before modeling. In practical applications, this type of analysis may help recommendation platforms better understand audience behavior and provide more personalized movie suggestions.

Modelling the Transport of Crude Oil in the Unsaturated Sandy and Loamy Soil Using Experimental and Statistical Methods

Uzochi Robert Njoku, Ejikeme Ugwoha, Victor Emeka Amah — Mon, 25 May 2026 00:00:00 +0000

Unregulated extraction and transport of crude oil pose significant environmental and public health risks through contamination of air, soil, and groundwater, particularly within the vulnerable vadose zone, where pollutants migrate toward subsurface water resources. This study modelled crude oil transport in the unsaturated zone using laboratory experiments, with sandy and loamy soils at varying column heights (15cm to 60cm). Many models have been developed and used, but fail to effectively depict the actual complexities, hence the need for this study. Polyvinyl chloride (PVC) soil columns were used, and after a 14-day drying period simulating drought, 0.0005m³ of crude oil was introduced into each column. Leaching time and Total Petroleum Hydrocarbon (TPH) concentrations were measured at different depths. In sandy soil, shorter columns had higher TPH concentrations and faster leaching, with leaching time ranging from 37 to 3600 seconds and TPH concentrations decreasing from 28,913mg/l to 103mg/l at depth (C₂). In loamy soil, leaching times ranged from 33 to 1500 seconds, with TPH concentrations decreasing from 11,085mg/l to 309mg/l at depth (C₂). Statistical analysis revealed a significant difference in concentration at depth between sandy and loamy soils. Predictive models based on Dimensional Analysis were developed for each soil type. The models ignored chemical and biological interactions, assumed steady-state flow, 1-dimensional vertical transport, homogenous media, constant viscosity, absence of preferential flow and a uniform initial crude oil distribution. In conclusion, the sandy soil model (R² = 0.97) showed better reliability for use than the loamy soil model (R² = 0.66). It is recommended that prediction be done cautiously with supplementary data when relying on the loamy soil model due to its moderate strength.

A Smart Water Quality Monitoring System Using Arduino IDE and Proteus IDE for Environmental Monitoring

A. Bubu, O. Horsfall — Tue, 26 May 2026 00:00:00 +0000

Water quality monitoring systems are crucial for ensuring the safety and sustainability of water resources. This paper discusses the design, implementation, and applications of a water quality monitoring system, highlighting its role in addressing global water challenges, promoting public health, and supporting environmental conservation. This system integrates advanced technologies, such as IoT (Internet of Things), remote sensing, and data analytics, to continuously assess water quality parameters, including pH sensor for pH, DS18B20 temperature sensor for temperature, turbidity sensor for turbidity, salinity sensor, and SIM800L GSM module for call activation and delivering short message service. By providing real-time monitoring and data visualisation, these systems enable early detection of pollution events, support regulatory compliance, and facilitate efficient water resource management. Modern implementations often leverage machine learning algorithms to predict water quality trends and optimise maintenance schedules. The integration of cost-effective sensors and cloud-based platforms is also explored as a means to enhance accessibility and scalability for communities worldwide.

Rheological Characterization and Enhanced Oil Recovery Performance of Plant-Derived Biopolymers and Nanoparticles

Hezekiah -Braye Oritom, Kenneth Lucky Igiks — Wed, 27 May 2026 00:00:00 +0000

Enhanced oil recovery is a technique mostly used to increase oil production from mature oil fields. This is achieved by altering reservoir and/ or fluid properties. There is a quest globally to meet the ever-increasing demand for hydrocarbons in a safe and environmentally friendly way. The aim of this study is to ascertain the potential of locally available environment-friendly plant leaf extract as a green polymer for enhanced oil recovery. This research evaluates rheological behavior and oil recovery performance of biopolymers extracted from plant leaves: Thevetia peruviana, Physalis angulata, and Solanum tuberosum, benchmarks them against commercial xanthan gum, and assesses their performance with silica oxide and aluminum oxide nanoparticles. A core flooding experiment was conducted with core samples from a Niger Delta reservoir. The rheological properties of the leaf extracts were determined with a Fann 35 rheometer. The results from the core flooding experiment showed that Xanthan gum without nanoparticles had 40% oil recovery; Thevetia peruviana had 26%; Physalis angulata had 18%; and potato leaf had 10%. Combining with a nanoparticle gives an additional 6% increase in oil production for xanthan gum. A combination of nanoparticles with Thevetia peruviana increased oil production by an additional 4%. Rheological characterization using Power Law, Herschel-Bulkley, and Bingham models shows that all plant leaf extracts exhibit shear thinning behavior indices of n of 0.24-0.34, comparable to xanthan gum. The findings provide a framework for selecting plant-derived biopolymers for enhanced oil recovery.

Desalination and Carbon Footprint: A Meta-Analysis of Plant‑Scale Operational Emissions for MSF and RO Technologies

Wed, 27 May 2026 00:00:00 +0000

As global freshwater scarcity intensifies, driven by rapid population growth, climate change, and industrialization, desalination has increasingly become a critical water supply strategy, particularly in arid and water-stressed regions. Despite its growing adoption, desalination remains highly energy-intensive, especially due to its operational processes that contribute significantly to carbon emissions, raising significant concerns about its environmental sustainability. Available life-cycle carbon intensity estimates vary widely across studies due to differences in plant design, energy sources, system boundaries, inconsistent definitions, and outdated data, which limit comparability. We present a plant- and pilot-scale meta-analysis to quantify operational carbon dioxide emission intensities (CO₂-eq/m³) for reverse osmosis (RO) and multi-stage flash (MSF) desalination technologies. For multi-effect distillation (MED), membrane distillation (MD), electrodialysis (ED), and nanofiltration (NF), available data are primarily modeled or experimental; as such, they were synthesized qualitatively and shown for context but not pooled. Results show that reverse osmosis (RO) has an operational carbon intensity of 2.52 kg CO₂-eq/m³ (95% CI: 1.28–3.76), while multi-stage flash (MSF) averages 8.98 kg CO₂-eq/m³ (95% CI: 4.66–13.30). The higher MSF values reflect reliance on thermal energy, whereas RO emissions align with electricity intensity and specific energy consumption. Sensitivity analyses confirm the robustness of these estimates. For MED, MD, ED, and NF, reported values were highly sensitive to energy source assumptions and system boundaries. These results establish transparent, plant-scale carbon benchmarks for desalination technologies under real-world energy pathways. The findings provide a decision-relevant framework for utilities and policymakers to evaluate technology selection, guide low-carbon infrastructure investment, and align desalination expansion with climate mitigation targets.

Comparative Performance of New and Remanufactured Lithium-Ion EV Battery Packs in Ghana

Thu, 28 May 2026 00:00:00 +0000

Background: The rapid transition toward electric mobility has increased the demand for lithium-ion electric vehicle (EV) batteries; however, the high cost of new battery packs remains a major barrier to EV adoption in developing countries such as Ghana. This study comparatively assessed the real-world performance of new and remanufactured lithium-ion EV battery packs under Ghanaian driving and environmental conditions through practical road testing.

Objective: The study aims to evaluate and compare the technical and environmental viability of using remanufactured lithium-ion batteries as a sustainable alternative to new or degraded batteries for electric vehicles in Ghana.

Method: An experimental comparative research design was employed using a converted electric vehicle fitted alternately with a new battery pack and a remanufactured battery pack. The batteries were evaluated under morning, afternoon, and evening driving scenarios across level roads, hilly terrains, highways, and traffic-congested conditions. Key performance indicators assessed included state of charge (SOC), battery percentage drop, charging efficiency, internal resistance, thermal performance, voltage stability, and driving range.

Results: The results showed that the new battery exhibited superior overall performance compared to the remanufactured battery. Internal resistance measurements revealed that the new battery recorded lower resistance values of 20 mΩ at the cell level and 22 mΩ at the module level, whereas the remanufactured battery recorded higher values of 35 mΩ and 41 mΩ, respectively, indicating increased energy losses and degradation tendencies in the remanufactured system. The charging analysis further demonstrated that the new battery reached 100% SOC within 90 minutes using a 60 kW DC fast charger, while the remanufactured battery required 120 minutes under similar charging conditions.

Road test results indicated a strong negative correlation between travel distance and battery percentage for both battery systems across all test periods. During the morning test, the Pearson correlation coefficients were for the new battery and for the remanufactured battery, with p-values less than 0.0001, confirming statistically significant discharge behavior. Similar trends were observed during the afternoon ( and ) and evening ( and ) sessions. The average battery percentage drop across all operational periods was higher for the new battery (186.99%) than the remanufactured battery (173.33%), with performance differences increasing progressively from morning (2%), afternoon (4%), to evening (7.33%) conditions.

Conclusion: The findings demonstrate that although remanufactured lithium-ion batteries exhibit moderate reductions in efficiency, thermal stability, and energy retention compared to new batteries, they still maintain acceptable operational performance for short-distance and urban EV applications. The study concludes that remanufactured EV batteries can provide a cost-effective and environmentally sustainable alternative for electric mobility deployment in Ghana, particularly within commercial transportation and circular economy frameworks.

Strength Capacity Characteristics of Welded and Mechanical Couple Spliced Steel Bars for Reinforcing Concrete

Fri, 29 May 2026 00:00:00 +0000

Lapping, the commonest technique used to splice reinforcement bars, needs no special instrument or skill to execute. However, their usage may result in excessive section reinforcement, which may result in a non-ductile response of the spliced region. The structural performance of reinforcement bar splices plays a critical role in ensuring the continuity and load transfer capacity of reinforced concrete members. This study experimentally investigated the tensile behaviour of steel reinforcement bars connected using threaded couplers and welded joints, with particular emphasis on the influence of connection development length and bar diameter. Tensile tests were conducted on 12 mm, 16 mm, and 20 mm diameter reinforcing bars with threaded engagement lengths of d, 3d, and 5d, and weld lengths of end-to-end, 3d, 5d, and 7d. The results show that connection length significantly affects the mechanical performance of both joining techniques. For threaded couplers, the 12 mm bars exhibited substantial reductions in yield and ultimate strengths at short thread engagement (d), achieving only 29–37% of the control strength. However, increasing the thread length to 3d and 5d significantly improved performance, with the 16 mm and 20 mm bars nearly recovering the full strength of the control specimens. In welded connections, increasing the weld length resulted in substantial strength recovery, with welded bars at 5d and 7d achieving strengths comparable to or slightly exceeding those of the control bars. The findings demonstrate that both threaded and welded reinforcement connections can achieve mechanical performance comparable to continuous bars when sufficient connection development length is provided. However, their efficiency varies with bar diameter and connection configuration.

Power Loss Mitigation Using Distributed Generation in Asaba Distribution Network

Raymond Onyeka NWAJUONYE, Chibuzor George OKONKWO, Innocent Ifeanyi OKONKWO — Tue, 02 Jun 2026 00:00:00 +0000

The increasing demand for electricity, coupled with rising economic and environmental concerns, has made renewable energy, particularly solar energy, a major area of interest. This research work evaluates the impact of distributed generation (DG) and presents a methodology for the optimal allocation and sizing of distributed generation within a distribution network in order to minimise power losses and improve voltage profiles. The study involved load flow analysis of the existing 15MVA, 33/11kV Benin Electricity Distribution Company (BEDC) Asaba injection substation distribution network and its 11kV radial feeders (SPC and Anwai Road), connected to an aggregate of ninety-six (96) 11/0.415kV secondary distribution transformers serving as load buses. Simulation, analysis, and integration of photovoltaic (PV) distributed generation into the Asaba injection substation distribution network were carried out using the Newton–Raphson algorithm and the Loss Sensitivity Factor (LSF) algorithm to determine the optimal location and size of the DG units. The network was modelled in ETAP version 12.6 using a detailed single-line diagram developed in the ETAP editing environment. The results revealed that, prior to DG installation, only ten (10) out of the ninety-six (96) buses operated within the statutory voltage limits of 394.25V–435.75V (0.95–1.05 p.u.), while eighty-six (86) buses violated the permissible voltage range. This indicated that the network was weak, unstable, and characterised by high active and reactive power losses of 1329.08kW and 2031.16kVAr, respectively. Following the optimal placement of DG units, active and reactive power losses were reduced by 57.5% and 70.7%, respectively. In addition, ninety-one (91) out of the ninety-six (96) buses operated within the statutory voltage limits. This improvement significantly enhanced the capacity, reliability, and operational efficiency of the distribution network.

A Process-controlled Synthesis of Corrosion-resistant Fe₃O₄/Au Nanoparticles as a Precursor for Stable Magnetic Fluids in Harsh Environments

Yun Long Wang — Tue, 02 Jun 2026 00:00:00 +0000

Background: Conventional direct reduction synthesis approaches encounter several significant challenges during the fabrication process. The acidic nature of chloroauric acid can induce etching of the Fe₃O₄ core, thereby compromising the structural integrity of the nanoparticles. In addition, rapid reduction kinetics frequently promote homogeneous nucleation of gold nanoparticles rather than controlled deposition onto the Fe₃O₄ surface. Furthermore, fluctuations in pH during the reaction process can result in uneven and non-uniform coating formation, ultimately affecting the stability, morphology, and reproducibility of the final nanocomposite material.

Aims: To develop a process-controlled synthesis of Fe₃O₄/Au nanocomposites with uniform Au decoration, high corrosion resistance, and superparamagnetism, serving as a stable precursor for magnetic fluids operable in harsh acidic environments.

Research Design: Experimental synthesis and characterization of Fe₃O₄/Au nanoparticles using a dynamic pH-buffered reduction strategy; evaluation of structural, magnetic, and anti-corrosion properties.

Methods: Fe₃O₄ nanoparticles (~12 nm) were prepared by coprecipitation. Fe₃O₄/Au nanocomposites were synthesized using hexamethylenetetramine (HMT) as an in-situ pH buffer, with real-time pH feedback and ultra-slow addition of ascorbic acid. Morphology, structure, composition, optical properties, magnetic properties, and corrosion resistance (0.1 mol/L HCl, 0.5 mol/L NaOH, 60 °C, 24 h) were characterized by TEM, SEM, XRD, FT-IR, UV-Vis, VSM, and TGA.

Results: Uniform Au nanoparticles (~50 nm) were anchored on Fe₃O₄ clusters (30–100 nm). The composite exhibited superparamagnetism with saturation magnetization 33 emu/g. After 24 h in 0.1 mol/L HCl, magnetization remained 31.4 emu/g with negligible iron leaching. Alkaline corrosion caused no observable degradation. TGA showed thermal stability below 250 °C. Oleic acid modification enabled oil-phase dispersibility.

Conclusion: The dynamic pH-buffered epitaxial strategy effectively suppresses Fe₃O₄ core dissolution and Au self-nucleation, yielding Fe₃O₄/Au nanocomposites with excellent acid resistance and magnetic integrity. This material is a promising precursor for stable magnetic fluids in harsh environments.

The Influence of Intake Pressure on the Performance of a Free-piston Hydrogen Engine

Zhen Han — Wed, 03 Jun 2026 00:00:00 +0000

Background: Free-piston engines have attracted significant research interest as a promising alternative to conventional internal combustion engines due to their simplified mechanical structure and potential for improved efficiency and lower emissions. Early studies primarily focused on scavenging and gas exchange processes, highlighting the strong dynamic coupling between piston motion, in-cylinder flow characteristics, and scavenging performance resulting from the absence of a crank–connecting rod mechanism.

Aims: This paper selects the free-piston hydrogen engine as its research subject, concentrating on investigating the coupled impacts of intake pressure on in-cylinder combustion, airflow dynamics, heat transfer, and emissions, with the aim of identifying the optimal operating conditions.

Study Design: A simulation model was developed by employing a coupled dynamics and thermodynamics methodology, and iterative computations were utilized to dynamically update the piston motion trajectory and combustion model. Subsequently, a three-dimensional model was constructed, incorporating the ECFM 3Z combustion model and the Han Reitz heat transfer model, while the extended Zeldovich mechanism was adopted to delineate NO formation. Ultimately, the reliability of the simulation model was validated through a comparative analysis of in-cylinder pressure and heat release rate, thereby laying a groundwork for subsequent parametric investigations.

Methodology: This article conducts research on parameters such as combustion, heat transfer, and emissions under different intake pressures, aiming to obtain the optimal operating conditions.

Results: When the intake pressure increased from 0.10 MPa to 0.18 MPa, the total heat transfer increased from 35.72 J to 38.83 J. The multi-objective evaluation results in terms of power performance, fuel economy, and emissions indicate that both the indicated mean effective pressure (IMEP) and indicated thermal efficiency exhibited increasing trends. Specifically, the IMEP increased from 0.5866 MPa to 0.6446 MPa, while the indicated thermal efficiency increased from 41.999% to 45.329%. In addition, with increasing intake pressure, the NO mass fraction increased significantly, whereas the soot mass fraction continuously decreased.

Conclusion: This article employs the linear weighted sum method to conduct multi-objective optimization analysis for various operating conditions. The optimization results indicate that within the studied operating range, the comprehensive evaluation result is optimal when the intake pressure is 0.16 MPa. However, due to certain limitations in both the research scope and computational workload of this article, further detailed analysis of the aforementioned range was not conducted.

Generating Use Case Diagrams from User Stories Using Large Language Models

Abdelkareem M. Alashqar — Thu, 04 Jun 2026 00:00:00 +0000

Background: While user stories are effective for defining features and supporting release planning in agile software engineering, they do not provide a comprehensive view of the system being developed. UML use case diagrams address this limitation by providing a visual representation of the overall system. Large language models (LLMs) have recently been adopted to support software modeling, mainly for generating UML models such as use case diagrams.

Aim: This paper proposes a multistep methodology, guided by well-designed prompts, that enables LLMs to generate UML use case diagrams from user stories.

Method: The methodology consists of three sequential processing steps, each guided by a well-designed prompt and processed progressively by four LLMs: ChatGPT, Gemini, Claude, and DeepSeek, to generate the final scripts of the use case diagrams from user stories. The performance of the four LLMs is evaluated using an inter-LLM agreement approach to assess their consistency.

Results: The experimental results of applying the multistep methodology to the four LLMs across six datasets show high consistency among them in generating use case diagrams from user stories. The highest inter-LLM agreement values achieved are 95.88%, 99.1%, and 100% for Step 1, Step 2, and Step 3 of the methodology, respectively.

Conclusions: LLMs can consistently extract and transform user stories into structured UML representations, and well-designed prompts reduce variability and produce high agreement across models. Although a preliminary qualitative semantic evaluation was achieved, this study remains focused on consistency among LLMs.

Adaptive Behavioral Zero Trust Frameworks for Identifying Rogue and Compromised Autonomous AI Agents

Tue, 09 Jun 2026 00:00:00 +0000

This article presents the Adaptive Behavioral Zero Trust Framework (AB-ZTF), a novel governance architecture integrating continuous behavioral profiling, Bayesian dynamic trust scoring, and adaptive policy enforcement for real-time identification of rogue and compromised autonomous AI agents. Rapid deployment of autonomous AI agents introduces complex vulnerabilities that conventional Zero Trust Architecture frameworks cannot address, as none integrate continuous behavioral profiling, dynamic trust scoring, and adaptive policy enforcement for agent governance. The framework was evaluated through a three-stage ensemble detection pipeline comprising unsupervised anomaly scoring, sequential LSTM-based behavioral classification, and Temporal Graph Neural Network-based lateral interaction analysis, applied to the CICIDS2017 benchmark dataset with SMOTE-based class balancing across five mapped AI agent threat classes. LightGBM achieved the highest weighted F1-Score of 0.9989 and AUC-ROC of 0.9999, while the Bayesian Trust Engine accurately classified 85.1% of agents, recording a 3.2-step Mean Time to Detection, a False Positive Rate of 0.0031, and a False Negative Rate of 0.0004. Under sustained adversarial feature perturbation, ensemble models maintained F1-Scores above 0.74. Unlike prior behavioral analytics and Zero Trust proposals that treat trust as static at deployment, the AB-ZTF achieves continuous runtime trust updating, surpassing existing benchmarks such as the 0.942 F1-Score reported for Isolation Forest-based non-human entity authentication and representing the first framework to empirically unify behavioral anomaly detection, Bayesian trust scoring, and adaptive Zero Trust policy enforcement for autonomous AI agent governance. These results are obtained using network intrusion data as a structured behavioral proxy, and the framework's direct applicability to live autonomous AI agent telemetry streams remains subject to further validation with agent-native datasets. AB-ZTF provides security layer for enterprise AI operations enforcing adaptive access control across AI ecosystems.

Comparative Performance Analysis of LCL, LLCL, L(LCL)₂ and L(LCL)₃ Filter Topologies for High-power Grid-connected Solar PV Inverters

Samson Dauda Yusuf, Unogwu, Daniel Ogbu, Abdulmumini Zubairu Loko — Tue, 09 Jun 2026 00:00:00 +0000

Advanced passive filter topologies are increasingly required in modern high-power grid-connected solar PV systems to ensure compliance with stringent harmonic and power quality standards. Conventional LCL-based filters often face limitations in multi-frequency harmonic attenuation, stability under weak grid conditions, and scalability for high-voltage applications. This paper presents a rigorous comparative analysis of four passive filter topologies, LCL, LLCL, L(LCL)₂, and the novel L(LCL)₃, for harmonic mitigation in high-power grid-connected solar PV inverter systems. Evaluation metrics encompass Total Harmonic Distortion (THD), attenuation at the switching frequency and its harmonics, Power Spectral Density (PSD) characteristics, filter losses, efficiency, component count, physical size, material cost, stability margin, and weak grid suitability. Results from MATLAB/Simulink simulation of a 100 MVA, 33 kV reference system demonstrate that the L(LCL)₃ filter achieves the best harmonic performance: grid-side THD of 1.3% (vs. 4.2% for LCL), switching-frequency attenuation of −90 dB (vs. −60 dB for LCL), and PSD reduction of 3,000:1 at the second harmonic. Filter efficiency is comparable across all topologies (99.51–99.53%), with total losses ranging from 445.8 kW (LLCL, minimum) to 468.6 kW (LCL, maximum). The L(LCL)₃ filter exhibits excellent stability characteristics (stability margin: Excellent; Q = 2.55) and the highest suitability for weak grid conditions, making it the preferred choice for Nigeria's power grid environment. Material cost is approximately 65% higher than the LCL baseline, offset by superior performance and lifetime operational benefits.

Designing Interesting Opponents through Online Learning in Predator–Prey Environments

Tue, 09 Jun 2026 00:00:00 +0000

Predator–prey environments are widely used benchmarks for multi-agent reinforcement learning (MARL) because they capture simultaneous cooperation among predators and competition against prey, yet many deployed predator opponents rely on static or pre-trained policies that become predictable, reduce behavioural diversity, and limit long-term engagement. This study investigates how online learning can generate adaptive and interesting opponents that continuously challenge prey agents. We propose an online-learning framework that integrates reinforcement learning with dynamic opponent adaptation and opponent modelling in a discrete 20×20 grid world containing four coordinated predator agents and one evasive prey. Agents use a state representation comprising relative agent positions, Euclidean distances, velocity vectors, and historical actions. Predator learning combines temporal-difference updates (Q-learning) with PPO/MAPPO-style policy optimization for stability, while an opponent model is updated online to predict behaviours and support coordinated decision-making. Interestingness is quantified using behavioural diversity (entropy), novelty (distance between current and historical behaviours), and challenge, alongside standard performance indicators and confusion-matrix-based evaluation of action prediction (Chase, Surround, Ambush). Across 10,000 training episodes and multiple runs under identical conditions, the proposed online-learning predators achieved the highest cumulative rewards with faster convergence than random, scripted, and offline-RL baselines, and attained an 87% capture success rate (56 percentage points above the random baseline). Online learning also produced the greatest behavioural diversity (entropy=2.21) while remaining strategically effective. Opponent modelling showed strong classification performance (85% Chase, 81% Surround, 82% Ambush; precision/recall/F1=0.83), and training yielded emergent cooperative strategies including coordinated trapping, dynamic flanking, ambush positioning, and adaptive pursuit. Overall, continuous online adaptation improves robustness and engagement by preventing behavioural stagnation, though it introduces computational overhead and potential instability in non-stationary MARL settings; future work should address scalability and explore hierarchical, graph-based, transformer, and meta-learning extensions. The work was conducted entirely in simulation without human or animal data.

Influence of Rice Husk Incorporation on the Strength of Structural Concrete: An Experimental Approach in San Martin, Peru 2026

Kenji Quispe Medina, José Francisco Cabrera Vilca, William Moisés Prado Labam — Thu, 11 Jun 2026 00:00:00 +0000

Rice husk (RH) is an abundant agro-industrial residue that can be incorporated into concrete mixtures as a sustainable alternative material to improve their mechanical performance. The objective of this research was to evaluate the influence of RH incorporation on the compressive and flexural strength of structural concrete with a design strength of f'c = 210 kg/cm². An experimental methodology under controlled laboratory conditions was applied using concrete mixtures containing 0.00%, 0.50%, 0.80%, and 1.00% RH. Cylindrical specimens and prismatic beams were tested at curing ages of 7, 14, and 28 days according to ASTM standards.

The results demonstrated that the mixture containing 0.80% RH achieved the best mechanical performance, reaching a maximum compressive strength of 218.10 kg/cm² and a flexural strength of 43.30 kg/cm² after 28 days of curing, exceeding the performance of conventional concrete. Additionally, the incorporation of moderate percentages of RH improved the internal cohesion of the concrete and reduced segregation without significantly affecting workability.

It is concluded that the moderate incorporation of rice husk improves the mechanical properties of structural concrete and represents an environmentally sustainable alternative for the reuse of agricultural waste in the construction industry.

Predictive Analysis of Employee Attrition Using Machine Learning

Mirali Mammadzade — Fri, 12 Jun 2026 00:00:00 +0000

Background: Employee attrition is a critical organizational issue that affects productivity, operational efficiency, and workforce stability, making employee retention an important strategic priority. Advances in machine learning have enabled organizations to analyze complex workforce data and predict employee turnover more accurately, supporting proactive retention strategies.

Aims: Employee attrition has become an important organizational challenge because persistent workforce turnover may negatively affect productivity, operational continuity, institutional knowledge retention, and long-term organizational stability. In recent years, predictive analytics and machine learning approaches have increasingly been adopted in workforce analytics to identify employees who may demonstrate elevated resignation risk before actual turnover occurs. The primary objective of this study is to evaluate the predictive effectiveness of supervised machine learning classification algorithms for employee attrition prediction using HR analytics data. The research additionally investigates how workplace conditions, compensation-related variables, employee satisfaction indicators, and professional experience influence workforce turnover behavior and organizational retention patterns.

Study Design: This study was conducted using a quantitative experimental research design based on supervised machine learning classification methods and comparative predictive analytics evaluation.

Place and Duration of Study: The experimental analysis was performed using the IBM HR Analytics Employee Attrition dataset between February 2026 and April 2026.

Methodology: The dataset contained demographic, financial, behavioral, and workplace-related variables associated with employee retention conditions and organizational stability patterns. Before model implementation, several preprocessing procedures were applied, including duplicate inspection, categorical feature transformation, feature scaling, and exploratory data analysis. Employee attrition status was defined as the target variable of the supervised classification task. Four supervised machine learning classification algorithms were implemented and comparatively evaluated using Python-based machine learning libraries, including Support Vector Machine, XGBoost, LightGBM, and CatBoost classifiers. The dataset was divided into training and testing subsets using an 80:20 ratio to evaluate predictive capability on previously unseen employee observations. Model performance evaluation was conducted using Accuracy, Precision, Recall, F1-score, and ROC-AUC metrics. Feature importance analysis and ROC curve evaluation were additionally performed to identify the organizational and behavioral variables contributing most strongly to employee turnover prediction.

Results: The experimental findings demonstrated noticeable differences in predictive capability among the evaluated classification models. Support Vector Machine achieved the strongest overall classification performance with an Accuracy score of 0.864 and ROC-AUC value of 0.816. LightGBM additionally demonstrated stable classification behavior and maintained balanced Precision performance throughout testing evaluation. In contrast, XGBoost and CatBoost produced comparatively lower Recall and F1-score values when identifying minority attrition observations. The findings additionally revealed that TotalWorkingYears, Age, MonthlyIncome, OverTime, and WorkLifeBalance were among the most influential variables associated with workforce attrition behavior. Employees with lower professional experience, lower compensation levels, excessive overtime exposure, and weaker workplace satisfaction conditions appeared more likely to demonstrate resignation tendencies.

Conclusion: The findings of this research confirm that supervised machine learning techniques can provide effective support for employee attrition prediction and workforce analytics applications. Among the evaluated models, Support Vector Machine demonstrated the strongest overall predictive capability under the implemented experimental conditions. The study additionally highlights the practical importance of predictive HR analytics in supporting employee retention planning, organizational decision-making, and workforce stability management. Furthermore, the results emphasize that preprocessing consistency, balanced model evaluation, and feature importance analysis play important roles in improving classification reliability and interpretability within workforce analytics systems.

AutoNav: Enhancing Education, Entertainment, and Research through Wireless Interactivity and Color Sensing

S. S. Nipun, R. V. Biswas — Sat, 30 May 2026 00:00:00 +0000

AutoNav, a Bluetooth-enabled vehicle that can be operated with a Bluetooth-enabled device like a smartphone or tablet, is one type of robotic vehicle. A Bluetooth module is integrated into the vehicle to establish communication with the user's device, enabling wireless control of the vehicle's movements and actions. A chassis, two DC motors, a motor driver module, an Arduino or an alternative microcontroller, an HC-05 or HC-06 Bluetooth module, a breadboard and jumper wires, and a power source are the standard components of a robot car. The microcontroller interprets and translates the commands transmitted by the controller app to the robot car in a particular manner into physical motor movements. This car also has a color sensor capable of identifying specific colors. A Bluetooth-controlled robot car has the potential to be utilized for multiple purposes, including research, education, and entertainment. It provides a fun and interactive way to learn about robotics and programming while also allowing for experimentation and customization of automobile's features.