Comparison of Methods for Determining Inception and Recovery Points of Voltage Variation Events

Abstract

Voltage sags and swells occur often in power systems; however determining the duration of these events is not straightforward. Seven methods for estimating the inception and recovery points (and consequently, the duration) of voltage sags and swells are surveyed: threshold rms voltage, waveform envelope, discrete wavelet transform, missing voltage, dq transformation, numerical matrix, and peak detector. Each method has its own strengths and weaknesses; however, no one method works on all conditions. An algorithmic approach to determining inception and recovery points is suggested. Such approach employs one method in combination with other methods based on the event characteristics.

Accurate Identification of Point-on-Wave Inception and Recovery Instants of Voltage Sags and Swells

“…The lowest time latency for the traditional method is obtained through short sliding windows (half-cycle) and high update rate (rms values computed for each new voltage sample). However, even this approach has been shown to introduce a relatively high time latency [22]. Additionally, the estimated event duration depends on the instants at which the rms values are computed.…”

“…A previous survey indicated that the discrete wavelet transform (DWT) presents the best performance among the methods for determining the inception and recovery points of a voltage sag [22]. Fig.…”

“…6a depicts an example of a sag pertaining to this class of events. Although it is still possible to estimate the inception and recovery instants by visually inspecting the voltage waveform, automating this process through the techniques found in the literature is challenging [22]. The rms voltage difference method is able to accurately determine the inception and recovery instants for this voltage sag, regardless of the presence of transient components.…”

“…5, the traditional method underestimates the event duration in this case. There is no general rule to predict whether the event duration calculated by the traditional method will be an underestimation or overestimation in relation to the 22 exact value. This estimate depends on the points on waveform at which rms values are updated, as discussed in Section II.…”

“…However, none of these methods provide accurate estimates for the point-on-wave inception and recovery instants for all types of sags/swells. For example, the threshold rms voltage method introduces an error up to 1 cycle in the estimation of the inception and recovery instants, while the discrete wavelet transform method performs well only if the voltage variation event is accompanied by transients [22]. This paper proposes a method to accurately determine the point-on-wave inception and recovery instants of voltage sags and swells.…”