In　this　study,　distributed　lag　nonlinear　models　(DLNM)　were　built　to　characterize　the　non-linear　exposure-lag-response　relationship　between　the　concentration　of　PM2.5　and　O-3　and　multiple　influencing　factors,　including　basic　meteorological　elements　and　precursors.　Then,　a　stratified　analysis　of　different　years,　seasons,　pollution　levels,　and　wind　direction　was　conducted.　DLNMs　and　coupled　Weather　Research　and　Forecasting　Model-Community　Multi-scale　Air　Quality　Model　(WRF-CMAQ)　were　used　to　evaluate　PM2.5　and　O-3　changes　attributed　to　meteorological　conditions　and　anthropogenic　emissions　comparing　2020　with　2016.　As　DLNMs　showed,　PM2.5　pollution　was　promoted　by　low　wind　speed,　high　temperature,　low　humidity,　and　high　concentrations　of　SO2,　NO2,　and　O-3,　among　which　NO2　tended　to　be　the　dominant　influencing　factor.　O-3　pollution　was　promoted　by　low　wind　speed,　high　temperature,　low　humidity,　high　concentration　of　PM2.5　and　low　concentration　of　NO2,　among　which　temperature　tended　to　be　the　dominant　influencing　factor.　Moreover,　north-south　and　easterly　winds　showed　the　greatest　contribution　to　PM2.5　and　O-3,　respectively.　Both　DLNMs　and　CMAQ　showed　that　anthropogenic　factors　alleviated　PM2.5　pollution　but　aggravated　O-3　pollution　in　2020　in　comparison　with　2016,　so　did　meteorological　factors,　but　with　smaller　impacts.　And　anthropogenic　influences　were　more　evident　in　heavily　polluted　seasons　for　both　PM2.5　and　O-3.　This　research　may　help　understand　the　influencing　factors　of　PM2.5　and　O-3　and　provide　scientific　guide　for　abatement　policies.　Moreover,　the　good　consistency　in　the　results　obtained　from　DLNMs　and　CMAQ　indicated　the　reliability　of　the　two　models.