POWER SIGNAL RECOGNITION METHOD AND SYSTEM WITH HIGH ORDER MOMENT FEATURES FOR NON-INTRUSIVE LOAD MONITORING

摘要

Provided are a system and a method for recognizing a power signal, capable of determining which electric device is turned on every moment by recognizing the power signal as a small fragment unit when the power single is inputted. The method includes the steps of: (1) extracting a higher-order moment feature vector from a power signal fragment; (2) extracting a secondary feature using the higher-order moment feature vector; (3) classifying a signal fragment using the secondary feature (i.e., determining which home appliance is currently turned on); and (4) predetermining (learning) medium variables which are required for the steps (2) and (3) using a learning material set.