SOFTWARE ALGORITHM DESIGN OF FAULT LOCATING AND CLASSIFICATION FOR TRANSMISSION LINE FAULTDIAGNOSIS AND DETECTION DEVICE

• Main Author: XAVERIUS EVAN, FRANSISKUS
• Format: Final Project
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/27398
• Institution: Institut Teknologi Bandung

<p align="justify"> A transmission line fault could cause some damage to the system. Nowadays, there is no method to detect a fault accurately. In this final project report a system to diagnose a transmission line fault is proposed. The system will classify, diagnose, and localizethe fault that occurred. The system proposed is implemented to a 150 kV three phase system. <br /> <br /> In transmission line faultdiagnosis and detection system, an analysis of high frequency transient signal due to fault occurrence has been done. From the analysis, the location and the type of the faulthas been determined. Fault locating is done using traveling waves method, along with wavelet transform, where the location is determined by calculating the time difference between the arrival time of the transient signal and its reflection to the power house. The location later calculated by multiplying the traveling waves velocity with the time difference. Discrete wavelet transform is used to represent the signal in time and frequency domain. The time difference is determined by counting the index differences of the first two consecutives level 4 wavelet coefficient. <br /> <br /> Fault classification is done by determining whether the fault is grounded or ungroundedfault from its ground mode current rms value. The faulted phase is being determined by applying rms voltage thresholding from every phase. Faulted phase will have its rms voltage reducedsignificantly due to the short circuit. <br /> <br /> Fault locating needs to be done as soon as possible after a fault occurrence to reduce downtime. Therefore, in this research a certain algorithm is developed where the algorithm could automatically determine the fault location and type as soon as it's occurred. The algorithm has to be able to determine the right peak of wavelet coefficients to calculate the right fault location and decide the right fault type. The whole algorithm has been implemented using Python programming language in Raspberry Pi 3 hardware. <br /> <br /> The proposed system has been implemented successfully. The fault localization reached 94,94% accuracy from 10 scenarios for each fault type. The fault classification has 100% accuracy on determining the fault types. The proposed system could automatically do the fault localization and classification as the fault happened. <p align="justify"> <br />