Omar Ibrahim Alsaif

The main objectives of this research are to fill a gap in PV inspection systems, which consist of convolutional detectors with a limited amount of global context or transformer classifiers with weak localization. Defects micro-cracks, broken electrodes, and degradation regions decrease the PV energy efficiency considerably, and this encourages the use of an automated computer vision algorithm for the classification of these defects extracted from electroluminescence (EL) images. These objectives were achieved by solving the following task: a novel dual-stage architecture is used to solve these tasks, which are performed sequentially, YOLOv8m followed by the Swin Transformer architecture. In comparison with previous CNN-Transformer PV models that combine features by concatenation or averaging with predefined weights, the proposed model implements a learnable cross-attention fusion probability merging ensemble, which dynamically weights the localization and global attention evidence for each defect. YOLOv8m localizes the exact positions of defects, then the Swin Transformer analyzes the region by window-swiping self-attention to return a long-range degradation pattern that CNNs can't pass. The model was trained using EL images in 12 classes, such as, intact cells, cracks, point defects, finger damage, and EL images were augmented to overcome the class imbalance problem and cross-conditioning generalization. The most important results are a macro-averaged F1-score of 94.05% and 95.40% accuracy, with per-class F1 ranging from 98.98% (Mono-crystalline) to 89.04% (Micro-crack). The significance of the obtained results lies in the use of integrative localization-driven YOLOv8m convolutional features with transformer global attention and learnable cross-attention fusion instead of static cross-attention fusion, providing a new and more robust implementation to perform automated PV quality assessment than a single branch.

With an increasing rate of data exchange via networks rapidly, information security is a critical indicator of network effectiveness. Data transmission has to be secured, and that is crucially done through cryptography, where AES, DES, and multi-level encryption (AES+DES) are the significant methods. This research provides a comparative performance analysis of AES and DES. Execution times for encryption and decryption, efficiency ratios, and memory usage were recorded using various dataset sizes (1, 10, 100, 1,000, 10,000, and 50,000 images). Results show that AES is faster than DES in all cases with less memory. Experiments were conducted on both CPU and GPU: Results indicate that GPU acceleration does make a difference in accelerating encryption with up to a (6.68×) speedup for multi-level encryption and (8.13×) for AES with 50,000 images. GPU-memory usage was (35%) less than the respective CPU-based memory, thus being more efficient. Multi-level encryption presents a potential trade-off for more robust security, while DES is insufficient for bulk encryption because its performance is considerably lowered.

Nowadays, it is widely recognized that solar PV is one of the favorited sustainable energy options globally. Recently, new technological methods have been developed based on the use of thin films and nanostructures with the aim of reducing the solar cell cost and increasing its efficiency. This work includes simulation of Cu(In,Ga)Se2 nanostructured solar cell. Silvaco program was used with TCAD technology to obtain (ZnO/CdS/CIGS) tandem solar cell model. The thickness of three layers was changed for optical wavelength range (300 –1200) nm, then the optical quantum efficiency was studied and compared with the measured model. The maximum value of quantum efficiency (80%) was obtained when thickness of ZnO layer was doubled, while the maximum value of this efficiency was obtained (74.5%) when reducing the thickness of CdS layer to one-tenth. As for CIGS layer, by increasing its thickness three times, the optimum quantum efficiency obtained is (82.08%). Several peaks are obtained in each case for different values of wavelengths which are determined from (450-750) nm. The best results reached were relied upon to implement an optimal model for a solar cell with several values of maximum efficiency for different wavelengths, as the highest quantitative efficiency (92%) has been achieved

The world is now more connected through technology and this has given rise to cyber threats, and malware is one of the threats that a system and data need to guard against. In this paper, we propose a detailed framework wherein state of the art ML methodologies can be employed for malware categorization and identification. In the evaluation of the performances of various ML algorithms, we analyze Random Forest, CatBoost, XGBoost, K-Nearest Neighbors (KNN), Histogram-based Gradient Boosting (Hist GB), and AdaBoost. The algorithms are assessed in this study using an assembly command dataset and a static and dynamic analysis approach to improve the detection rate and stability. Out of all algorithms discussed, Random Forest, CatBoost, XGBoost, Hist GB are ranked highest with 99% accuracy. As for the accuracy, KNN yielded an accuracy of 97%. Performance analysis based on metrics shows that Random Forest, CatBoost, and Hist GB not only have high accuracy but also precision, recall, and F1-score. Particularly, the accuracy of Random Forest was 99% for both the precision, recall, and F1-score. These results confirm the use of ML-based solutions in the analysis and counteraction of modern malware threats and their advantages over traditional detection methods as well as strengthening cybersecurity

Mustansiriyah Journal of Pure and applied sciences was firstly commission in 2022. A quarterly journal that publishes four issues per year. The journal publishes academic and applied papers dealing with recent topics and scientific concepts. Papers considered for publication in Mathematic, Physics, Chemistry, Computer Science, Biology, Geology. Accepted papers will be freely downloaded by professors, researchers, instructors, students, and interested workers. Mustansiriyah Journal of Pure and Applied Sciences is an open access journal, therefore there are no fees required for downloading any publication from the journal website by authors, readers, and institutions. The journal applies the license of CC BY (a Creative Commons Attribution 4.0 International license). For more information, please follow the link https://creativecommons.org/licenses/by/4.0/. The journal is also supported by the College of Education , Mustansiriyah University.

Mustansiriyah Journal of Pure and applied sciences was firstly commission in 2022. A quarterly journal that publishes four issues per year. The journal publishes academic and applied papers dealing with recent topics and scientific concepts. Papers considered for publication in Mathematic, Physics, Chemistry, Computer Science, Biology, Geology. Accepted papers will be freely downloaded by professors, researchers, instructors, students, and interested workers. Mustansiriyah Journal of Pure and Applied Sciences is an open access journal, therefore there are no fees required for downloading any publication from the journal website by authors, readers, and institutions. The journal applies the license of CC BY (a Creative Commons Attribution 4.0 International license). For more information, please follow the link https://creativecommons.org/licenses/by/4.0/. The journal is also supported by the College of Education , Mustansiriyah University.

Protecting maritime borders is crucial to ensuring overall border security. Law enforcement agencies make great use of analyzing images of underwater debris to gather intelligence and detect illicit materials. Underwater image improvement contributes to better data quality and analytical. Nevertheless, underwater image analysis poses greater challenges compared to analyzing images taken above the water, factors like refraction of light and darkness contribute to the degradation of underwater image quality. In this paper, a novel approach is proposed to enhance underwater images, the proposed approach involves splitting underwater colored image to its three basic components, Subsequently, a point spread function is created for each component to describes image blurring factor, The deblurring process is then applied by using wiener filter, the result sharped by sharping filter to clarify edges, contrast linear stretch is performed to improve contrast and visual details. and the resulting image is finally reassembled from the three basic components. The proposed method showed effective results in evaluating the main metrics and gave better results when compared to a number of different methods. These results prove the effectiveness of the proposed method and its ability to practical applications in improving image quality.