The neuro-fuzzy technique (NFS) has been implemented for the
deterioration of the winding insulation paper (WIP) identification
using the dissolved gas-in-oil analysis for power transformers
and the effectiveness of NFS diagnosis has been greatly analyzed.
The real data sets are used to investigate its feasibility in forecasting
the electrical health assessment of power transformers on the
basis of DGA methods in power transformer oil.
From the test results, it is found that the neuro-fuzzy method is
more suitable method as a power transformer diagnosis procedure.
The NFS with the PNN has a better performance than the other
individual AI method (BPNN or fuzzy-logic method) or other combination
in terms of providing diagnosis accuracy and operating
time. The accuracy of the NFS for faults diagnosis is comparable
to conventional methods due to their great facilities for study.