Diagnosis kerosakan talian penghantaran sistem kuasa mengguakan kaedah inferens neural kabur ubah suai

Transmission line protection system is one of the most challenging issues in power system protection due to requirements of fast and accurate technique. Fault diagnosis on the transmission line conventionally is implemented either for detection and classification or detection and location of fault o...

Full description

Saved in:
Bibliographic Details
Main Author: Azriyenni, Azriyenni
Format: Thesis
Language:English
Published: 2018
Subjects:
Online Access:http://eprints.utm.my/id/eprint/79256/1/AzriyenniPFKE2018.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Transmission line protection system is one of the most challenging issues in power system protection due to requirements of fast and accurate technique. Fault diagnosis on the transmission line conventionally is implemented either for detection and classification or detection and location of fault only. This leads to time consuming either to locate or classify fault. None of the researchers perform detection, classification and location of fault simultaneously due to huge data transactions and delays. To ensure the availability of supplies, the control centre requires fast and complete information data for immediate follow-up. Among typical intelligent techniques which are normally applied include Fuzzy Logic (LK), Back Propagation Neural Network (RNRB), Support Vector Machines. This thesis develops a fault diagnosis based on the hybrid intelligent algorithms for detection, classification and location of fault simultaneously. The proposed technique is a combination of RNRB and LK techniques known as Adaptive Neuro-Fuzzy Inference System (SINKUS). SINKUS algorithm has been proposed to enhance the integrated data training for detection, classification and location of faults in power system transmission lines. For each of the stated assignment, four different SINKUS structures were developed producing a total of twelve SINKUS structures, each with six inputs and fuzzified using triangular membership function. During fault detection the system detects the affected phase through four different SINKUS structures and feeds the information to the next stage for classification. At this stage, the system is able to classify the type of the fault in a detected phase, using another four different SINKUS structures. On successful classification, another group of four SINKUS structures can be able to position the fault location. The effectiveness of the proposed algorithm is demonstrated on 13-bus IEEE Test System, 118-bus IEEE Test System and the Riau Power System. The types of fault performed in the diagnosis are symmetrical and asymmetrical faults. The simulation results are compared with the LK and the RNRB techniques to verify its performance. From the observed results, it is found that the relay operating time using SINKUS technique is only within 0.11 to 0.16 seconds or has been reduced to about 20% to various scales of power systems with 99.95% accuracy. Hence, the proposed technique is able to provide complete, fast and accurate fault information and can be applied to large-scale power system.