Power data quality improvement through PMU bad data detection based on deep complex

Phasor measurement units (PMUs) allow devices to be switched in a variety of power signal modes. PMU data from spike produces jitter or signal glitches and disturbance in transmission creates bad data. Due to these challenges, PMU data experience varying degrees of data quality issues. Several methods have been already used to detect fake data, but they come with drawbacks like complexity. Similarly, it has not been possible to identify faulty data caused by topology fluctuations optimally. To resolve these problems a deep complex neural network (DCNN) has been proposed in which robust bad data detection technique analyses complex numbers with both voltage magnitude and phase angles. Comparisons are made between the proposed methods and existing methods in terms of accuracy, bad data detection capabilities, bad data detection range, running time, F1-score, and computational cost. The proposed technique provides an accuracy of 99.5% which is higher than the existing techniques. © 2023 Inderscience Publishers. All rights reserved.