A similarity-based PMU error detection technique

This paper presents a two stage error detection technique for a power system time series data. It implements a local similarity method to isolate an anomalous time series data, and further applies a window scanning technique to detect instances of inconsistent data segments. The requirement for few parameter definitions and small computation time makes this technique attractive for data error detection. Validation of the technique is carried out using data obtained from prototyped PMU clock delay and GPS signal loss.