In　order　to　realize　the　intelligent　evaluation　of　moisture　content　of　transformer　oil-paper　insulation,　a　comprehensive　evaluation　model　combining　multi-frequency　domain　characteristic　parameters　and　random　forest　algorithm　was　proposed.　Firstly,　on　the　basis　of　in-depth　analysis　of　the　variation　of　dielectric　properties　in　frequency　domain　of　oil-paper　insulation　samples　under　different　moisture　conditions,　nine　characteristic　parameters　with　significant　correlation　with　moisture　content　were　extracted.　Secondly,　multiple　groups　of　measured　characteristic　data　in　the　frequency　domain　were　normalized,　and　the　water　content　of　the　measured　sample　data　was　evaluated　by　random　forest　algorithm　under　the　condition　of　multiple　characteristic　data.　The　experimental　results　show　that　the　root-mean-square　error　and　mean　absolute　error　of　the　evaluation　model　constructed　in　this　paper　are　significantly　reduced　compared　with　the　traditional　model,　which　can　provide　an　effective　scientific　basis　for　the　subsequent　quantitative　evaluation　of　the　transformer　oil-paper　insulation　state.　©　2022,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.