Palestinian Journal of Technology and Applied Sciences (PJTAS)

Distribution Network Performance Enhancement Using Reconfiguration Technique based on Gravitational Search Algorithm

2024-03-20T14:12:22+02:00

Objectives: The main goals of this work are to minimize network power loss and enhance the system's voltage profile (VF).

Methods: This work presents a novel methodology that simultaneously optimizes Distribution Network Reconfiguration (DNR), Distributed Generation (DG) sizing, and DG placement using the Gravitational Search Algorithm (GSA) optimization technique. The DNR approach helps reduce power loss, but its effectiveness is limited when applied alone. Similarly, optimizing DG sizing and placement can further minimize power loss, but improper integration with DNR may lead to increased power loss and voltage fluctuations. Hence, it is essential to develop an efficient optimization strategy that simultaneously determines the optimal DG size and location while achieving optimal DNR.

Results: For the IEEE 33-bus network, active and reactive power losses were reduced by 67.488% and 64.88%, respectively. Similarly, for the IEEE 69-bus network, the reductions in active and reactive power losses were 82.55% and 62.25%, respectively.

Conclusions: The findings show that adjusting the size and location of distributed generation units (DGs) while configuring the network significantly improves the voltage profile and reduces losses.

Feature Selection for Serving Medical Datasets Applying Heuristic Algorithms (Scatter Search within Decision Tree Classifier)

2024-03-20T09:27:22+02:00

Objectives: This research presents a feature selection process on different datasets of the medical domain with different aims and sizes using a wrapper approach based on a powerful metaheuristic algorithm which is the Scatter Search Algorithm and J48 decision tree classifier as the selection criteria.

Methods: The paper applied a modified approach of the basic Sequential Scatter Search algorithm called Improved Sequential Scatter Search follows the basic procedures of the original algorithm in addition to an early improvement mechanism choosing decision tree classifier to be the evaluator of the experiments.

Results: The experimental results show competition and superiority in feature selection compared to other metaheuristic algorithms for the same datasets in consideration of number of features selected and accuracy.

Conclusion: This research emphasizes the importance of wrapper approaches using metaheuristic algorithms to select the most dominant attributes in a dataset which is very important in reduction of the cost and complexity of all data analysis areas.

Using Fine Needle Aspiration Data to Classify Breast Cancer Types by Machine Learning

2024-07-10T09:23:01+03:00

Objectives: Breast cancer, a leading cause of death worldwide and the foremost in Palestine, often benefits from early diagnosis to improve patient outcomes. However, diagnosing small tumors accurately can be challenging, with a high risk of human error. This study seeks to enhance breast cancer classification by utilizing machine learning (ML) algorithms.

Methods: The research analyzed and utilized three machine learning techniques - Decision Tree Classifier (DTC), Support Vector Machine (SVM), and Random Forest Classifier (RFC) - to predict breast cancer tumors. The accuracy of the three algorithms was analyzed and evaluated using a confusion matrix as well as different metrics on a dataset containing 569 samples and 29 features.

Results: The result showed that the Decision Tree Classifier (DTC) has the high scores of 100% in accuracy, precision, sensitivity, and specificity.

Conclusions: In the conclusion, the research emphasizes the excellent performance of the Decision Tree Classifier in classifying breast cancer, which could significantly improve diagnostic accuracy and patient outcomes. The results indicate that DTC has the potential to be a useful ML model in decreasing human diagnostic mistakes and enhancing the early detection and care in medical environments, prompting additional studies to enhance and confirm its effectiveness.

Results: The result showed that the Decision Tree Classifier (DTC) has the high scores of 100% in accuracy, precision, sensitivity, and specificity.