Historical　district　is　the　product　of　space　and　social　factors　in　the　process　of　urban　development,　it　owns　inherent　characteristics　and　patterns.　Today,　the　best　way　people　for　the　historic　district　is　through　travel.　Tourism　development　is　an　effective　way　to　further　develop　historical　districts.　How　to　maintain　a　balance　between　historical　district　protection　and　development,　how　to　effectively　evaluate　planning　programs　is　still　an　urgent　problem　to　be　solved.　This　study　uses　space　syntax　and　neural　network　method　to　construct　the　evaluation　model　of　tourism　spatial　planning　congestion.　The　purpose　of　the　model　is　to　establish　an　individual　behavior-oriented　model　to　understand　the　interaction　between　space　and　pedestrians.　Utilizing　the　advantages　of　space　syntax　and　neural　network　method,　Kulangsu　street　accessibility　and　the　traffic　flow　of　multiple　attractions　are　analyzed.　Through　the　model,　the　conflict　between　the　road　choice　and　the　flow　of　people　and　the　pressure　of　the　attractions　attraction　are　obtained,　so　as　to　assess　the　pros　and　cons　of　the　status　quo,　and　make　recommendations　for　future　improvement　based　on　the　obtained　results.　©　2019　IEEE.