DIYALA JOURNAL OF ENGINEERING SCIENCES

This paper presents a portable, customized low-cost data acquisition system using the ESP32-C6 microcontroller and the INA219 sensor. The use of traditional multi-meters is time-consuming and subject to human error, especially for experiments with longer durations,which require continuous logging and repeated readings. Manual data recording increases the risk of transcription mistakes and limits sampling resolution. Since DC systems have gained significant attention in recent years across renewable energy and IoT applications, the need for more accurate, automated DC power acquisition is increasingly critical. We conduct a comparative analysis of the INA219 sensor under low and elevated voltage applications to evaluate the accuracy of the INA219 sensor with the calibrated reference DC power supply. Our experiment indicates perfect linear correlation between power and current in both modes (r≈1.000 for low-power, r = 0.992 for high-power) applications, respectively. A Python-based application was developed to assess the data logging ability of the COOLTERM software. This application will perform statistical analysis of the time-series power measurement datasets, including a visual representation in graphical form to clearly display the energy trend in the DC power system. This system provides a practical solution for real-time power measurement across a variety of DC systems, including IoT applications, solar renewable energy, laboratory experimentation, and related DC power systems.

Sugarcane(Saccharum officinarum)is one of the most economically important cropsin the world. However, outbreaks of leaf diseasescause extensive annual yield losses and threaten sugarcane agricultural productivity. Therefore, early and accurate disease detection is vital for effective crop management, yield protection, and food security. This study presents a deep ensemble learningframework, termed CBAM-SugarcaneNet, for the automated recognition of sugarcane leaf diseases under field conditions. The proposed framework integrates progressive multi-resolution training, a Convolutional Block Attention Module (CBAM), focal loss, Sharpness-Aware Minimization (SAM), ensemble learning, and test-time augmentation to address major challenges such as class imbalance, inter-class visual similarity, background variation, and limited training data. A three-stage progressive training strategy was employed using ConvNeXt-Large and EfficientNet-B5/B7 architectures enhanced with CBAM attention modules to improve the accuracy of disease-relevant feature extraction. Focal loss was used to emphasize minority and difficult-to-classify disease classes, whereas SAM improved model generalization byencouraging convergence towardflatter minima. Experiments were conducted on an 11-class sugarcane leaf disease dataset comprising 6,748 images. The proposed framework achieved 97.2% accuracy for single-model deployment, 97.6% for the ensemble configuration, and 98.1% with test-time augmentation, with a mean 5-fold cross-validation accuracy of 98.05%± 0.14%. Grad-CAM visualizations further confirmed that the model focused on disease-affected leaf regions. The developed framework offers a robust and deployable solution for precision-agriculture-based disease diagnosis in sugarcane

Rotary friction welding (RFW) is a solid-state joining process widely used for cylindrical components; however, accurate prediction of transient temperature fields remains challenging due to complex frictional heating, material property mismatch, and convective losses. This study develops an analytical model for transient thermal behavior during RFW of 20 mm diameter rods made of similar (316L/316L) and dissimilar (316L/A60) materials at rotational speeds of 710–2000 rpm.Interfacial heat generation is formulated based on thermal effusivity and partitioned between the contacting materials. The three-dimensional transient heat conduction equation in cylindrical coordinates is solved analytically using separation of variables, yielding a Fourier–Bessel series solution. For computational efficiency, the dominant axisymmetric mode (n = 0) is implemented in MATLAB. The model assumes constant thermophysical properties, semi-infinite geometry, axisymmetric heat generation, and lateral convective heat losses.Resultsshow symmetric temperature distributions for similar materials and pronounced asymmetry for dissimilar joints due to differences in thermal conductivity and diffusivity. The predicted peak interface temperature increases nearly linearly with rotational speed, reaching 780.1 °C at 1000 rpm, with deviations below 5% from experimental data (770–800 °C). The heat-affected zone (T > 500 °C) is also accurately predicted.The proposed model provides a fast, physically consistent, and computationally efficient tool for thermal analysis and optimization of RFW processes.

Traditional power systems rely on large synchronous generators (SGs), where the mechanical rotor’s inertia naturally stabilizes grid frequency during demand fluctuations.However, the rapid growth of inverter-based renewable energy sources (RES) has transformed modern power systems.Because the inverter-based RES systems lacked the same inertia as normal SG during a disturbance, any sudden changes in frequency and voltage could occur at the point of common coupling (PCC) that reducing system stability. High-RES system penetration will lower system fluctuations and overall power system inertia. The frequency stability of a power grid that primarily uses RES is proposed to be improved by using virtual synchronous generators (VSG). To reduce DC-link voltage fluctuation and stabilize the power system's frequency, VSG with enhanced whale optimization algorithm (WOA) technique for optimizing the PI controller's settings are used in this work. The proposed enhanced WOA integrates chaotic initialization, nonlinear adaptive parameters, elite guidance, and dynamic probability adaptation to ensure faster convergence, prevent premature stagnation, and achieve robust optimization. The system was modeled and simulated in MATLAB/Simulink, where its performance was validated under a scenario of rapid load change. The simulation results demonstrate the benefits of PI tuning with the WOA in 40% reduction in settling time and 65% reduction in system frequency overshoot.Employing the enhanced WOA in the proposed VSG system leadsto lower the error rates of the voltage regulator within (800 ±5)V and stabilize the frequency within (50±0.04)Hz of the VSG

Student feedback sentiment analysis for co-curricular activities, employing deep learning, ascertains and classifies job placements based on students' feedback.Though several works have focused on student sentiment analysis, there remains room to analyze job placement and the effects of co-curricular activities on students' academic performance. This work proposes a method, called L1-Regularized Graph Convolutional Neural Network (L1-GCNN), to identify the relationship between co-curricular activities and students' placement performance. Initially, the raw student placement data is considered as input. After that, Euclidean Synthetic Minority Over-sampling-based pre-processing, including normalization and class balancing, is applied to the student placementdataset. Synthetic samples are generated efficiently, thereby avoiding distortions in local distributions and achieving class-balanced results. Following this, a normalized, class-balanced sample is used to train an L1-regularised Graph Convolutional Neural Network-based Student sentiment analysis, which is then applied to identify the most representative and optimal feature subset for estimating the impact of co-curricular activities on students’ academic performance. Employing the graph structure, the L1-regularised or logistic function highlights the interdependencies between co-curricular metrics and student placement performance via feedback. Moreover, the L1-regularisation function improves the student performance system by optimizing the network and utilizing features more effectively. Experiments are conducted on the proposed L1-GCNN and existing methods using several metrics. The simulation results show relative improvements in student sentiment analysis for placement performance based on co-curricular activities, reducing the mean absolute error by 0.053, precision by 0.9, recall by 0.93, and accuracy by 0.89 compared to existing methods.

This paper presents a novel control strategy aimed at improving power quality and operational stability in a multi-rotor wind energy conversion system based on a Doubly Fed Induction Generator (DFIG). Conventional Direct Power Control (DPC) techniques often suffer from significant active and reactive power ripples, variable switching frequency, and limited robustness under parameter uncertainties and changing wind conditions. To overcome these drawbacks, a feedback-based Fractional-Order Proportional–Integral (FOPI) controller is integrated into the DPC scheme. The controller parameters are optimally tuned offline using the Rooted Tree Optimization (RTO) algorithm to achieve enhanced dynamic performance and robust operation.The proposed control approach is implemented on the machine-side converter, where instantaneous power estimation is employed to determine the control error, while Pulse Width Modulation (PWM) is used to generate the switching signals and regulate converteroperation. By combining the flexibility of fractional-order control with the optimization capability of the RTO algorithm, the proposed method effectively improves power tracking accuracy and reduces system oscillations.To evaluate its effectiveness, the developed strategy is compared with the conventional 12-sector DPC method under various operating conditions. Simulation results obtained using MATLAB/Simulink demonstrate that the proposed controller significantly decreases power and stator current ripples, enhances transient and steady-state responses, and provides superior robustness against disturbances and parameter variations. Consequently, the proposed DPC-FOPI-RTO scheme contributes to improved energy quality, increased system reliability, and better overall performance of DFIG-based wind energy conversion systems

Power grid reliability and voltage stability are important factors for the stable operation of hydropower plants. Automatic Voltage Regulators (AVRs) contribute to maintaining stability by controlling generator excitation and terminal voltage. Achieving the best possible AVR performance is considered an important challenge. This study presents a comparison of two optimization algorithms—the Grey Wolf (GWO) algorithm and the Whale (WOA) algorithm—for selecting the optimal proportional-integral-derivative (PID) controller parameters in an AVR system. The optimization process uses two performance index equations, the integral of the time absolute error (ITAE) and the integral of the time square error (ITSE). The performance evaluation and Simulations were conducted using MATLAB/Simulink. The results showed that the ITSE method based on the Whale Optimization Algorithm (WOA) achieves a faster response, while the Gray Wolf Optimization (GWO-ITSE) algorithm was superior in stability performance. By using GWO-ITSE, the maximum deviation was decreased from 1.58 p.u. to 1.03 p.u. with 34.8% approximate enhancement. Moreover, the settling time (sec) was reduced from 18 seconds to 7.5 seconds with 58.3% approximate improvement. As a result, the GWO-ITSE proposed more stability in these hydropower plants and can be more suitable for similar applications.The obtained results further highlight the improved AVR performance using optimized PID parameters

Harmful brain activity can significantly impact an individual’s life, leading to debilitating seizures, epileptic episodes, and long-term cognitive impairments that affect daily functioning and overall quality of life. These abnormalities can be effectively detected using electroencephalography (EEG) image data, which captures the underlying neural activity of the brain in a non-invasive manner. In this study, Retentive Networks (RetNets) are employed to classify EEG spectrogram images for identifying different types of harmful brain activity. While deep learning architectures such as EfficientNet and Vision Transformers have demonstrated strong performance in image-based classification tasks, RetNets offer a compelling advantage in terms of reduced computational complexity and efficient sequence modelling. The dataset used in this work is a publicly available EEG dataset comprising approximately 17,000 spectrogram images derived from EEG recordings, ensuring sufficient diversity and representation for robustmodel training and evaluation.Experimental evaluation demonstrates that the proposed RetNet model achieved the highest overall performance, with a precision of 94%, recall of 100%, and an F1-score of 97%, indicating balanced and highly reliable classification capability. In comparison,EfficientNet achieved a precision of 91%, recall of 90%, and an F1-score of 96%, while the Vision Transformer achieved precision, recall, and F1-scores of 94%, 86%, and 90%, respectively. Furthermore, the proposed approach integrates model interpretability through Explainable Artificial Intelligence (XAI) techniques. The novelty of the proposed work lies in combining RetNet-based EEG spectrogram classification with XAI-driven interpretability for harmful brain activity detection

The durability and strength of single-lap adhesive joints are influenced by numerous factors. Among which are stress and strain concentrations at the overlap edges. This research investigates a method for improving these joints through a proposed geometricdesign incorporating a circular overlap region with a central hole. The primary objective is to reduce strain concentrations within the adhesive layer, which is an important factor governing joint strength and durability. An aluminum alloy was employed asthe adherent material, in conjunction with an epoxy adhesive.Numerical and experimental analyses, utilizing the Finite Element Method (FEM) and Digital Image Correlation (DIC) techniques, were used to evaluate strain distribution and validate the proposed design. Three models of different inner circle diameter were tested. The tested where made such that the adherent area are the same of corresponding conventional rectangular area. Results indicate that the modified configuration achieved a reduction in normal strain within the adhesive layer ranging from 2.8% to 8.6% compared to the conventional design. Furthermore, strain concentrations at the overlap edges were redistributed, promoting a more uniform stress distribution and potentially enhancing joint strength. Very good agreement was observed between FEM and DIC results, varying from 69% to 95%, with an average exceeding 80% for most measurement points. These findings demonstrate the possibility of simple geometric modifications to enhance adhesive joint performance.

A novel control strategy was introduced for enhancing the performance of photovoltaic (PV) systems integrated into the power grid. The proposed method applied an Improved Grey Wolf Optimizer (IGWO) directly for Maximum Power Point Tracking (MPPT) in PV systems. The main goal is to maximize the output power extracted from the PV panels under varying environmental conditions while in the same stage improving the power quality delivered to the grid through keeping the stable voltage and frequency at the same required levels. A detailed simulation model was obtained using MATLAB/Simulink, including the PV array, DC-DC boost converter, and grid-connected inverter. The IGWO algorithm was employed to determine the optimal controlling signals for MPPT, that providing in high dynamic performance compared to traditional techniques. Simulation was included list of three cases in solar irradiance and load variations to evaluate the effectiveness of the controller. The proposed IGWO-based MPPT strategy achieved higher output power from the PV system compared to the selected traditional method as P&O. In addition, the method effectively minimized the Total Harmonic Distortion (THD) and ensures voltage with frequency stability at the point of common coupling. The designing of a single optimization control for the PV framework based on IGWO technique was the main novelty of this work through the integration between the regulated grid voltage and frequency stabilization. This control strategy enhanced the reliability, making ita strong candidate for next-generation renewable energy integration

Chronic Kidney Disease (CKD) is a progressive renal disease that needs to be accurately and early diagnosed to eliminate irreversible damage to the kidney. Nevertheless, traditional methods of diagnosis might not be particularly effective at capturing the complex nonlinear patterns between clinical attributes. The paper offers a Hybrid Metaheuristic Optimization with Stacked Sparse Autoencoder framework as a superior method of CKD detection and classification. The suggested model is based on the combinationof Z-score normalization, hybrid Dwarf Mongese Optimizer-Giant Trevall Optimizer based feature selection, deep hierarchical feature representation with Stacked Sparse Autoencoder, and RMSProp-based adaptive optimization. The DMO aspect facilitates exploration of the needed clinical properties globally whereas GTO enhances local optimization of the chosen set of features. The SSAE also trains small sparse representations to minimize redundancy and enhance the interpretability of the diagnoses. The frameworkwas tested on the UCI CKD dataset of 400 patient records having 24 clinical attributes. The proposed hybrid optimization model achieved an accuracy of 99.25%, precision of 99.10%, recall of 99.40%, specificity of 99.05%, F1-score of 99.25%, MCC of 0.985, AUC of 0.997, and an error rate of 0.75%. The proposed model had a higher accuracy of 1.75% and 0.85 when compared with XGBoost+BSO and Two-Tier ACBPNN, respectively. These findings validate the hypothesis that hybrid feature-selection and sparse deep-representation enhance the diagnostic reliability, misclassification, and enable interpretable CKD screening. The suggested framework has a high potential to be a clinical decision-support model of early CKD detection