Based　on　deeply　analysis　for　optimization　process　of　basic　fruit　fly　optimization　algorithm,　an　improved　fruit　fly　optimization　(IFOA)　algorithm　is　proposed　via　changing　random　search　direction　and　adding　to　a　tuning　coefficient　of　search　radius.　Moreover,　through　introducing　the　regularization　term　of　objective　function　in　IFOA　algorithm,　a　new　method　that　IFOA　algorithm　is　combined　with　Tikhonov　regularization　method　is　put　forward　in　order　to　resolving　ill-conditioned　problems.　Analysis　results　of　practical　example　show　that　solution　accuracy　of　new　method　is　superior　to　genetic　algorithm　and　single　Tikhonov　regularization　method.　When　observation　contains　gross　errors,　the　deviation　between　the　results　and　the　true　value　will　increase　rapidly　using　least　square　method　to　solve　ill-conditioned　problems.　At　this　time,　the　new　method　has　strong　robustness.　Compared　with　intelligent　search　method　represented　by　genetic　algorithm,　new　method　has　the　characteristics　of　less　parameter,　fast　calculation　speed,　simple　optimization　process.　It　is　more　practical　in　ill-conditioned　problems　solution.　©　2016,　Surveying　and　Mapping　Press.　All　right　reserved.