Show simple item record

AuthorBhuiyan, E. A.
AuthorMuyeen, S. M.
AuthorFahim, S. R.
AuthorSarker, S. K.
AuthorDas, S. K.
Available date2022-03-23T08:22:45Z
Publication Date2021
Publication Name2021 3rd International Conference on Smart Power and Internet Energy Systems, SPIES 2021
ResourceScopus
Identifierhttp://dx.doi.org/10.1109/SPIES52282.2021.9633925
URIhttp://hdl.handle.net/10576/28911
AbstractThis paper introduces a greedy layer-wise learning algorithm to diagnose open-circuit faults of grid-connected inverters. Inverters play important roles in energy conversion, especially when converting direct current to alternating current. The accurate functioning of inverters is essential for successful energy conversion. The diagnosis of inverter faults is the primary requirement to guarantee the reliability of the entire energy conversion operation. In this work, a multilayer learning algorithm based on a restricted Boltzmann machine (RBM) is presented for fault diagnosis of an inverter topology. It uses both supervised and unsupervised layer-wise learning and hierarchically extracts the features from a given data. A three-phase two-level grid-connected PV inverter test model has been operated for twenty-two conditions to assess the effectiveness of the proposed algorithm. The investigation results in diagnostic accuracy of 99.786% for twenty-two operating conditions of the inverter.
Languageen
PublisherInstitute of Electrical and Electronics Engineers Inc.
SubjectDeep learning
Electric inverters
Energy conversion
Fault detection
Learning algorithms
Timing circuits
Alternating current
Conversion operation
Deep learning
Direct-current
Faults diagnosis
Generative model
Grid-connected
Layer-wise
Open-circuit fault
Photovoltaic inverters
Failure analysis
TitleA Greedy Layer-Wise Learning Algorithm for Open-Circuit Fault Diagnosis of Grid-Connected Inverters
TypeConference Paper
Pagination72-76
dc.accessType Abstract Only


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record