Voltage Sag Detection Based on PDF Parameters Changes in Tensor Index

This paper presents a proposal of voltage sag detection algorithm based on cumulative sums algorithms (CUSUM) and harmonic component decomposition of Tensor Theory (TT) unified index through Kalman Filter (KF). Algorithm provide a reduce computational burden at same time it granted small detection delays and high-effectivity detection respect to state-of-the-art sag detection algorithms. Computational burden reduction was possible by using of TT voltage representation and reduction of the KF order. Abrupt changes on TT representation and rapid estimation on KF permit evidence a large deviation on signal probability function parameters (PFP) during voltage sag. Two-sided CUSUM algorithm take advantage of PFP deviations to make fast and accurate detection. Test for proposal and state-of-the-art algorithms was development. A large voltage sag record data base was used to test algorithms performance under low sampling conditions. Effectiveness, accuracy and computational burden were the test approaches.