Previous　air　pollution　control　strategies　didn＇t　pay　enough　attention　to　regional　collaboration　and　the　spatial　response　sensitivities,　resulting　in　limited　control　effects　in　China.　This　study　proposed　an　effective　PM2.5　and　O3　control　strategy　scheme　with　the　integration　of　Self-Organizing　Map　(SOM),　Genetic　Algorithm　(GA)　and　WRF-CAMx,　emphasizing　regional　collaborative　control　and　the　strengthening　of　control　in　sensitive　areas.　This　scheme　embodies　the　idea　of　hierarchical　management　and　spatial-temporally　differentiated　management,　with　SOM　identifying　the　collaborative　subregions,　GA　providing　the　optimized　subregion-level　priority　of　precursor　emission　reductions,　and　WRF-CAMx　providing　response　sensitivities　for　grid-level　priority　of　precursor　emission　reductions.　With　Beijing-Tianjin-Hebei　and　the　surrounding　area　(BTHSA,　“2　+　26”　cities)　as　the　case　study　area,　the　optimized　strategy　required　that　regions　along　Taihang　Mountains　strengthen　the　emission　reductions　of　all　precursors　in　PM2.5-dominant　seasons,　and　strengthen　VOCs　reductions　but　moderate　NOx　reductions　in　O3-dominant　season.　The　spatiotemporally　differentiated　control　strategy,　without　additional　emission　reduction　burdens　than　the　14th　Five-Year　Plan　proposed,　reduced　the　average　annual　PM2.5　and　MDA8　O3　concentrations　in　28　cities　by　3.2%–8.2%　and　3.9%–9.7%　respectively　in　comparison　with　non-differential　control　strategies,　with　the　most　prominent　optimization　effects　occurring　in　the　heavily　polluted　seasons　(6.9%–18.0%　for　PM2.5　and　3.3%–14.2%　for　MDA8　O3,　respectively).　This　study　proposed　an　effective　scheme　for　the　collaborative　control　of　PM2.5　and　O3　in　BTHSA,　and　shows　important　methodological　implications　for　other　regions　suffering　from　similar　air　quality　problems.　©　2023