Browsing by Author "Rezai, Sina"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    Fault Identification of UPFC-Compensated Transmission Lines in Complex Microgrids Using an Intelligent Relaying Scheme Based on Discrete Wavelet Transform and ANN Classifier
    (2018-07-23) Rezai, Sina; Sesay, A. B.; Nowicki, Edwin P.; Pahlevani, Majid
    Transformation of traditional power infrastructures to modern inter-connected electrical grids has resulted in the emergence of microgrids as modernization of the existing aging transmission and distribution systems. Microgrids allow integration of distributed energy resources (DER) such as wind farms (WF) and may utilize Flexible AC Transmission Systems (FACTS) such as Unified Power Flow Controller (UPFC). FACTS increase power flow capacity of transmission lines while DERs supply additional electric energy. In these systems, current direction, as well as power injections, may change at any time; hence, the protection task is complicated during fault occurrences. The fault data is highly nonlinear and nonstationary, making a conventional distance relay dysfunctional. To fix this, many researchers have used signal processing and soft computing techniques. However, previous works are inadequate in terms of scope and type of fault data acquired from circuits, which usually have simple topologies and are applicable to specific systems. In contrast, this work proposes a single-ended intelligent relaying scheme for high voltage UPFC-Compensated transmission lines in complex interconnected power systems with mesh/loop configurations. This study uses a microgrid model created in Matlab/Simulink, which includes a constant power source, a WF, a steam turbine synchronous generator (STSG), and UPFC for line compensation. Interestingly, most unpredictable in this circuit is the fault signature as it simultaneously combines the UPFC three-phase symmetric response as well as the responses from other interconnected sources. The study uses discrete wavelet transform (DWT) for feature extraction and artificial neuron network (ANN) for feature classification of fault currents. The main objectives are automatic detection and identification of fault type with the best accuracy, reliability, and reduced computational complexity. Furthermore, to analyze the impact of UPFC on fault, the study considers also: (i) circuit without UPFC and (ii) circuit without UPFC & DER. Furthermore, using a two-block ANN consisting of two parallel subclassifiers based on whether the fault involves ground or not, further improved the overall performance suggesting a modular ANN is preferred. Finally, the method achieved an early detection time of less than half cycle of post-fault data that is quite useful for breakers’ fast three-phase tripping operations.