Fault Simulation Dataset for 110 kV Power Transmission Lines

This dataset contains simulated fault events for fault detection, classification, and localization in 110 kV power transmission lines. The simulations were conducted using DIgSILENT PowerFactory, with a Python script automating fault generation, data extraction, and dataset compilation. The simulations include three fault types: single-phase-to-ground fault (spgf), two-phase short circuit (2psc), and three-phase short circuit (3psc). Each fault type was simulated with fault locations varying from the beginning to the end of the transmission line in 5% increments, ensuring comprehensive spatial coverage along the lines. Additionally, a no-fault scenario was included to represent normal operating conditions, where system loads were incrementally varied by 0.1 MW to simulate typical fluctuations in power demand.

The dataset contains 24 input variables that describe the electrical state of the system under both fault and no-fault conditions.
Features
Voltage Magnitudes: Line-to-ground voltage magnitudes for phases A, B, and C at busbars:
Ua1, Ub1, Uc1 - Bus 1
Ua2, Ub2, Uc2 - Bus 2
Ua3, Ub3, Uc3 - Bus 3
Short-Circuit Current Magnitudes: Initial (subtransient) short-circuit current magnitudes for phases A, B, and C at the beginning of transmission lines:
Ia1, Ib1, Ic1 - Line 1
Ia2, Ib2, Ic2 - Line 2
Ia3, Ib3, Ic3 - Line 3
Ia4, Ib4, Ic4 - Line 4
Ia5, Ib5, Ic5 - Line 5

Targets
The target variables for classification and regression tasks are:
Fault: Type of fault (spgf, 2psc, 3psc, or no-fault)
Line: Faulted line (Line1, Line2, Line3, Line4, Line5, or NONE in case of no-fault)
Position: Fault position along the line (0%, 5%, ..., 100%, or NONE in case of no-fault)

The dataset consists of 618 samples, covering various fault and normal operating scenarios. For each sample:
Fault events are labeled by fault type, faulted line, and fault position.
No-fault scenarios are included to improve model robustness in distinguishing between fault and normal operating conditions.

The dataset provides a comprehensive and systematic representation of electrical behaviors under faults and normal conditions, facilitating fault detection, classification, and localization tasks. The integration of DIgSILENT PowerFactory simulations with Python ensures accuracy and reproducibility in data generation.