Data validation and anomaly detection techniques for smart substations

Smart substation data quality has been a big concern for utilities since large scale Phasor Measurement Unit (PMU) deployments. Data quality directly impacts the validity of the applications based on such data. This paper proposes a system including a model-less pre-screening step, a model-based SLSE step, and data-driven statistical method to detect anomalies. The pre-screening step serves as pre-filtering process validating the raw PMU data and setting the data quality flag. The modelbased SLSE uses the quality flags to determine which data to use. The SLSE makes use of the three-phase voltage and current phasors to do three-phase current LSE and three-phase zero-impedance voltage LSE. The SLSE estimates values for the measurements and performs error analysis to be sure they are good values. The statistical analysis includes 5 main processes. The implementation of these steps is explored in this paper. The proposed solution has been validated with simulated substation data. The work addresses data quality issues and hence benefits downstream synchrophasor applications, such as oscillation detection, voltage stability, and system control.