DC Line Fault Prognosis Using Deep Recurrent Neural Network Over Sensor Data

dc.contributor.authorVissakodeti, Akhil Venkat
dc.contributor.departmentChalmers tekniska högskola / Institutionen för elektrotekniksv
dc.contributor.examinerHammarström, Thomas
dc.date.accessioned2023-03-08T12:13:15Z
dc.date.available2023-03-08T12:13:15Z
dc.date.issued2023
dc.date.submitted2023
dc.description.abstractAbstract The HVDC technology has become prominent because of its increased long-distance bulk power transmission efficiency and facilitation of asynchronous interconnections. The loaded cable can, however, fail due to flashover or short circuit in the power system. As a result, this can cause a grid failure and damage the equipment by introducing a high level of current in the system. To detect fault is therefore considered a cost-efficient and non-destructive technique to monitor the cable operating condition. The main aim of this thesis is to predict faults in a DC cable using measured data from the sensors present in the system. Moreover, this method helps to identify the cable fault before power failure with possible catastrophic consequences occurs. This thesis examines the prospect of employing deep neural networks to capture the hidden patterns from the time series sensors to predict DC cable fault at early stages. This is justified because deep learning approaches are well suited to incorporating feature extraction into the predictive model. In this regard, long short-term memory (LSTM) is considered to get a remarkable accuracy of 99.93%. A lower Relative value of the absolute error of the signals proves that the model predicts the accurate results for the fixed window size.
dc.identifier.coursecodeEENX30
dc.identifier.urihttp://hdl.handle.net/20.500.12380/306002
dc.language.isoeng
dc.setspec.uppsokTechnology
dc.titleDC Line Fault Prognosis Using Deep Recurrent Neural Network Over Sensor Data
dc.type.degreeExamensarbete för masterexamensv
dc.type.degreeMaster's Thesisen
dc.type.uppsokH
local.programmeElectric power engineering (MPEPO), MSc
Ladda ner
Original bundle
Visar 1 - 1 av 1
Hämtar...
Bild (thumbnail)
Namn:
Chalmers_University_of_Technology__Master_Thesis_Template_2021.pdf
Storlek:
2.75 MB
Format:
Adobe Portable Document Format
Beskrivning:
License bundle
Visar 1 - 1 av 1
Hämtar...
Bild (thumbnail)
Namn:
license.txt
Storlek:
2.35 KB
Format:
Item-specific license agreed upon to submission
Beskrivning: