Abstract—Reliable fault detection in communication transmission lines is crucial for modern infrastructure. This paper presents a simulation-driven method for classifying transmission line faults, normal, open, short, and mismatch, using physically grounded transmission line features. By leveraging key parameters such as the reflection coefficient, Voltage Standing Wave Ratio (VSWR), and material properties (complex permittivity, conductivity), a dataset that mimics real-world fault scenarios are built. A Support Vector Machine (SVM) was selected for its transparent decision boundaries, strong performance on moderately sized datasets, and ability to leverage physiochemically interpretable features. While SVM provides a robust baseline, other classifiers, including Random Forests and neural networks, may offer advantages in scalability and nonlinearity. A systematic comparison and benchmarking with additional classifiers are proposed for future studies to clarify their relative merits for transmission line fault detection.