Phasor Measurement Unit Data - Labeled
Description
The PMU dataset is collected from the website of Ecole Polytechnique Fédérale de Lausanne (EPFL) institutions of Switzerland. The PMU dataset consists of 16 features and the data is normal without anomalies. The PMU provides a measurement for every 0.02 second. One hour's worth of PMU data contains 1,80,000 readings. The dataset takes 10 days of May month data into account and one measurement is chosen at random as a sample for every 500 measurements. As a result, 8640 samples are obtained per day and 86400 samples are obtained throughout a 10-day period. In the dataset first 86400 rows are normal data which are labeled as 0 and next 15247 rows are anomalies which are labeled as 1. The anomalies include Single Line to Ground fault, Line to Line fault and Double Line to Ground fault.
Files
Steps to reproduce
As PMU data anomalies are not available, they are generated for line faults such as single line to ground fault (SLG), line to line fault (LL), and double line to ground fault (LLG) for testing purposes. The condition for a single line to ground fault is when the current in phases B and C is zero and the current in phase A is maximum, as shown in Equation (1). I_b=I_c=0 and I_a is maximum (1) Because the current magnitudes of phases B and C are zero, the current angles of phases B and C are also zero, and the current magnitude in phase A is maximized by multiplying the phase A magnitudes of the data points by a factor of square root of 5. Other features in the observed dataset comprise the timestamp, delay at arrival, frequency, rate of change of frequency, phase A voltage magnitude, phase A voltage angle, phase A current angle, phase B voltage magnitude, phase B voltage angle, phase C voltage magnitude and phase C voltage angle remain unchanged. The condition for a line to line fault is when the current in phase A is equal to zero, the current in phase C is equal to the negative of the current in phase B and the voltage in phases B and C are equal, and, as shown in Equation (2). I_a=0,I_b= -I_c and〖 V〗_b=V_c (2) As the current magnitude of phase A is zero, the current angle of phase A is also zero and the remaining features namely timestamp, delay at arrival, frequency, rate of change of frequency, phase A voltage magnitude, voltage angle, phase B voltage and current angle and phase C voltage and current angle remain unchanged in the observed dataset. The condition for double line to ground fault is the current in phase A is equal to zero and the voltage in phase B and C are equal to zero as in Equation (3). I_a=0,V_b= V_c=0 (3) As the current magnitude of phase A is zero, the current angle of phase A is also zero and the remaining features namely timestamp, delay at arrival, frequency, rate of change of frequency, phase A voltage magnitude, voltage angle, phase B voltage angle, current magnitude, current angle, phase C voltage angle, current magnitude and current angle remain unchanged in the observed dataset. By incorporating the conditions mentioned in Equations (1), (2) and (3) on the normal data set, 5000 anomalous data is generated for each faulty condition.