- Repository URL:
- electrical engineering.; Major electrical engineering.
thesis / dissertation description
Locating the faulted section of a distribution system is a difficult task because of lack of accurate system models and the presence of uncertainty in the data used for estimating the fault section. Many of the methods used to account for the uncertainty use fuzzy logic techniques to estimate bounds of possibility of the input data and calculated quantities, or probabilistic modeling of the input data to estimate the likelihood of the location of the fault on a particular section of the feeder. Heuristic knowledge of control center dispatchers has also been used for uncertainty management. This thesis presents the design and implementation of a phase selector algorithm and a fault distance algorithm for use in an automated modular scheme for fault section estimation on radial distribution systems. These two algorithms will be executed in combination with two other fault location algorithms. The scheme is executed using the data record of an abnormal event in a three-stage scheme. The phase selector algorithm was used to obtain event-phase possibility values representing the possibility of involvement of each of the phases and the neutral in an event. A section-event possibility value that indicated the possibility that a section of the feeder was involved in the fault was evaluated using the event-phase possibility values and line section phase topology information. The fault distance algorithm was used to eliminate sections of the feeder that were not likely to be possible faulted section candidates by assuming a bolted fault and estimating its location. Each line section was assigned a fault possibility value of zero or one according to its location relative to the location of the fault. The phase selector algorithm was tested using real data measured at feeder substations and the fault distance algorithm was tested using data obtained by staging faults on a model of an overhead feeder using EMTP/ATP simulation. The results obtained from the tests were promising. A simple illustration of the combination of the results of the two algorithms is given. The result of this combination shows the potential of the simultaneous use of the two algorithms.