Based　on　deeply　analysis　for　optimization　process　of　basic　fruit　fly　optimization　algorithm　(FOA),　a　new　improved　FOA　(IFOA)　method　is　proposed,　which　modifies　random　search　direction,　increases　the　adjustment　coefficient　of　search　radius,　and　establishes　a　multi-sub-population　solution　mechanism.　The　proposed　method　can　process　the　nonlinear　objective　function　that　has　non-zero　and　non-negative　extreme　points.　In　the　paper,　IFOA　method　is　applied　to　ill-conditioned　problem　solution　in　the　field　of　surveying　data　processing.　Application　of　the　proposed　method　on　two　practical　examples　show　that　solution　accuracy　of　IFOA　is　superior　to　that　of　three　well-known　intelligent　optimization　algorithms　and　two　existing　improved　FOA　methods,　and　it　is　also　better　than　truncated　singular　value　decomposition　method　and　ridge　estimation　method.　In　addition,　compared　with　intelligent　search　method　represented　by　particle　swarm　optimization　algorithm,　The　IFOA　method　has　the　advantages　of　less　parameter　settings,　simple　optimization　process　and　easy　program　implementation.　So,　IFOA　method　is　feasible,　effective　and　practical　in　solving　ill-conditioned　problems.　©　2019　Walter　de　Gruyter　GmbH,　Berlin/Boston　2019.