Localized　attack　model　studies　the　process　where　damage　diffuses　from　a　node　to　its　neighbors　and　the　neighbors　of　its　neighbors　and　has　a　great　practical　significance.　In　this　paper,　we　propose　an　adjusted　localized　attack　process　that　breaks　through　the　restriction　of　the　structure　of　shells　and　considers　node　differentiation　protection.　On　the　basis　of　this　process,　we　construct　a　modified　theoretical　model　and　apply　it　to　computer-generated　and　practical　networks　to　observe　its　accuracy,　the　resilience　of　the　networks,　and　the　validity　of　our　node　protection　pattern.　Results　show　that　our　modified　model　works　well　in　several　of　these　networks,　that　is,　those　networks　with　an　even　degree　of　distribution　tend　to　be　robust　to　that　adjusted　localized　attack,　whereas　our　node　protection　pattern　is　more　valid　on　networks　with　hub　nodes.　(C)　2019　Elsevier　B.V.　All　rights　reserved.