Abstract:
This thesis emphasizes on the automatic classification of short duration voltage variation event data which are recorded by power quality monitors in the 22 kV distribution system. The rms voltage is computed and updated by the UNIPOWER PQSecure for each half cycle. The considered conditions used in the classification of these events are the changes of the voltage and current magnitude, duration, the recovery of voltage and current magnitude after a dip occurred, cause and direction of fault, and 1 minute window size. This thesis uses 3 steps to analyse and classify the events. In the first step, the events are classified by case and direction of fault which separated in 6 types, such as "Fault & Downstream", "Fault & Upstream", "Interruption & Downstream", "Interruption & Upstream", "Not Fault & NA.", and "Not Group". In each event, the magnitude, duration, time interval between shot, and cause and direction of fault are evaluated. This program is able to classify the events with more than 80% satification of all events in 1 year. In the second step, the "Fault & Downstream" events are analyzed in order to find out the pattern of the voltage and fault development in each event. These results are presented in statistic classification such as the summary table of voltage and fault development. Finally, it is to calculate the statistic of fault current from the results in step 2, such as ground fault, phase fault type 2L and 3L, phase fault develop start from 1LG and 2LG. The results are used for preventing the severity occurred from the short duration voltage variations event which caused by the fault development. This developed program can handle huge database by using SQL-DB and Delphi 7.