Online bad data outlier detection in PMU measurements using PCA feature-driven ANN classifier

This paper presents a method for online bad data outlier detection in PMU measurements. High resemblance of bad data in the form of outliers with that of system disturbances can cause control centers to take false decisions. The proposed method utilizes artificial neural network (ANN) in association with principal component analysis (PCA)-based feature extraction technique in order to classify bad data outliers and outliers caused by disturbances like faults. The extracted features using PCA from the reduced dimensional representation of the data is given as input to ANN. Bayesian regularization back-propagation-based learning algorithm is used for training ANN. The test results demonstrate online classification/detection of bad data of different amplitudes injected in measurements, both before and after disturbance, in two example power systems.