Efficient electrical connections are essential thus addressing the integrity of Insulation Piercing Connectors (IPCs) are critical in aerial bundled cable (ABC) systems. Proper torque application during installation is crucial where these connectors are subjected to a variety of mechanical stressors, such as shearhead torque application, continuity testing, and body testing, all of which affect their long-term performance in overhead cable reliability where in this study, all of these dataset were used by the means of Artificial Intelligence (AI) in order to provide further insight of potential failure risk by clustering the torque range. We have deployed the models by means of machine learning (ML), including Random Forest (RF), Decision Trees (DT) and Artificial Neural Networks (ANN) using all these Mechanical Testing dataset from TeknoBumi to foresee each algorithm capability of self-prediction of individual torque values and data set correlation. Additionally, clustering methods using K-Means and DBSCAN were employed to categorize the torque, identifying low, intermediate, and high torque based on Malaysia IPCs torque range standard. The failure risk for IPCs was evaluated using Z-score and Isolation Forest anomaly detection, enabling early identification of potential failures. The study proves that RF achieving the highest performance metrics demonstrating superior reliability in predicting torque values and data correlations. The result of comparison of K-means and DBSCAN Clustering mark that K-Means was effective in creating distinct and well separated clusters, making it the more reliable choice for standard classification in this IPCs case. The Failure risk for IPCs shows that most samples fall into a low to moderate risk range, but some higher-risk samples, typically marked in red on the 3D plots, stand out. These red points highlight areas that deviate from the normal and could indicate potential problems, such as installation issues or material defects. These findings enable early detection of potential failures caused by improper installation or material defects, allowing industry players to implement targeted quality control measures. By leveraging AI-based torque prediction and clustering, utility companies and manufacturers can reduce maintenance costs, improve installation accuracy, and enhance the overall reliability of Low Voltage Overhead Distribution Networks.

Machine Learning Models, Torque Prediction, Insulation Piercing Connectors, Clustering Algorithms, Failure Risk Assessment, Anomaly Detection, K-Means Clustering, DBSCAN Clustering