This　paper　proposed　a　method　for　evaluating　the　oil-paper　insulation　of　transformers　based　on　wavelet　analysis　and　probabilistic　neural　network　(PNN).　First,　the　six-layer　decomposition　of　daubechies　6　(db6)　is　performed　on　the　depolarization　current　spectra　to　obtain　the　signal　components　of　each　frequency　band.　Then,　the　signal　components　of　each　frequency　band　are　used　to　generate　energy　feature　vectors;　further,　the　energy　feature　vectors　generated　from　more　than　50　transformers　with　different　insulation　states　are　used　to　build　a　training　sample　set　to　train　a　probabilistic　neural　network　and　construct　a　PNN-based　evaluation　method.　Finally,　the　method　is　used　to　evaluate　the　insulation　of　the　testing　sample　set.　The　results　show　that　the　proposed　method　based　on　wavelet　analysis　and　PNN　can　quickly　and　effectively　evaluate　the　aging　state　of　transformer　oil-paper　insulation,　and　can　provide　new　ideas　to　evaluate　the　insulation　of　transformers　using　depolarization　current.　©　2022　SPIE