Use this URL to cite or link to this record in EThOS:
Title: Artificial intelligence based fault location in a transmission system with UPFC
Author: Zhou, Xiaoyao
ISNI:       0000 0001 3577 7440
Awarding Body: University of Bath
Current Institution: University of Bath
Date of Award: 2006
Availability of Full Text:
Access from EThOS:
Access from Institution:
Continuing pressure to minimize capital expenditure and the increasing difficulties involved in obtaining transmission rights of way have focused the attention of the utility community on the flexible AC transmission system (FACTS) concept resulting in the initiation of studies and implementation programmes which are now well underway. Accurate fault location for FACTS-compensated transmission lines is a crucial part of the complete protection scheme to maintain the integrity of power systems. This research is devoted to the investigation and development of accurate fault location techniques for a transmission system with Unified Power Flow Controller (UPFC). Many current fault location techniques are based on the measurement of apparent impedance of the transmission line, distance relay principle being one of them. In this thesis, a comprehensive study is thus carried out based on the fault data attained from an improved UPFC transmission system model, to ascertain how the apparent impedance is affected under different faults by the UPFC, and also its adverse impact on the commonly employed distance relay performance. In order to overcome the drawbacks of the conventional fault location approach, this thesis proposes the application of discrete wavelet transform (DWT) integrated with artificial neural network (ANN) to the development of an accurate fault location technique. The ANN based fault location comprises of three stages: fault classification, fault discrimination and fault location. The fault data obtained from the sending end of UPFC-compensated transmission line are decomposed into a series of wavelet components by utilising DWT. The salient features are then chosen as inputs to different fault classification, discrimination and location ANNs. The extensive simulation studies have demonstrated that a very high classification rate of over 99% and a maximum fault location error of 2% are achieved under a vast majority of practically encountered system and fault conditions.
Supervisor: Aggarwal, Raj ; Wang, Haifeng Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available