Impact of Waveform Segmentation Accuracy on Disturbance Recognition Reliability

Mathieu Caujolle (1) , Marc Petit (1) , Gilles Fleury (1) , Louis Berthet (1)

1 E3S - Supélec Sciences des Systèmes

Power Quality disturbances can be identified by modern waveform analyzers using several recognition indicators. The indicators are computed after segmentation of the signal during either steady-state intervals or transient ones. This paper proposes indicators for the recognition of capacitor bank switching and transformer energizing events. These indicators computed over transient intervals prove to be robust to noise and harmonic disturbances. We also investigate their sensitivity to inaccurate detection instants. For each indicator, a validity interval is identified and its properties determined. Constraints on the segmentation stage are deduced from this analysis: a standard deviation less than 2.5 ms is required on the segmentation delay for reliable recognition of transformer energizing event, while a deviation less than 0.25 ms is necessary for capacitor switching recognition. For both event types, a near zero mean segmentation delay is to be expected.

Domains: Electric power; Signal and Image processing; Signal and Image Processing

Mathieu Caujolle, Marc Petit, Gilles Fleury, Louis Berthet. Impact of Waveform Segmentation Accuracy on Disturbance Recognition Reliability. International Conference on Power Systems Transients (IPST'11), Jun 2011, Delft, Netherlands. CD-ROM Proceedings (8 p.). ⟨hal-00633037⟩