Symbolic Dynamic Filtering for Online Power Quality Anomaly Detection

A methodology for anomaly detection using Symbolic Dynamic Filtering (SDF) is proposed for detection of power quality events in this paper. The methodology overcomes the limitation of bulk data processing by compressing the signature feature information. It improves the sensitivity of early evolving anomaly detection in case of power quality events, which might not be evident by manual inspection. SDF is constructed based on the knowledge of symbolic encoding and finite state automata that generates signature patterns of histograms. Quantification of anomalous condition is done by comparing it with reference conditions, which depicts the deviation from normalcy. The proposed approach is validated on simulated data of various power quality events. The proposed SDF approach is found to be promising for online power quality event detection.