The　differences　of　SARS-CoV-2　variants　brought　the　changes　of　transmission　characteristics　and　clinical　manifestations　during　the　prevalence　of　COVID-19.　In　order　to　explore　the　evolution　mechanisms　of　SARS-CoV-2　variants　and　the　impacts　of　variant　evolution,　the　classic　SIR　(Susceptible-Infected-Recovered)　compartment　model　was　modified　to　a　generalized　SVEIR　(Susceptible-Vaccinated-Exposed-Infected-Recovered)　compartment　model　with　age-group　and　varying　variants　in　this　study.　By　using　of　the　SVEIR　model　and　least　squares　method,　the　optimal　fittings　against　the　surveillance　data　from　Fujian　Provincial　Center　for　Disease　Control　and　Prevention　were　performed　for　the　five　epidemics　of　Fujian　Province.　The　main　epidemiological　characteristics　such　as　basic　reproduction　number,　effective　reproduction　number,　sensitivity　analysis,　and　cross-variant　scenario　investigations　were　extensively　investigated　during　dynamic　zero-COVID　policy.　The　study　results　showed　that　the　infectivities　of　the　variants　became　fast　from　wild　strain　to　the　Delta　variant,　further　to　the　Omicron　variant.　Meanwhile,　the　cross-variant　investigations　showed　that　the　average　incubation　periods　were　shortened,　and　that　the　infection　scales　quickly　enhanced.　Further,　the　risk　estimations　with　the　new　variants　were　performed　without　implements　of　the　non-pharmaceutical　interventions,　based　on　the　dominant　variants　XBB.1.9.1　and　EG.5.　The　results　of　the　risk　estimations　suggested　that　non-pharmaceutical　interventions　were　necessary　on　the　Chinese　mainland　for　controlling　severe　infections　and　deaths,　and　also　that　the　regular　variant　monitors　were　still　workable　against　the　aggressive　variant　evolution　and　the　emergency　of　new　transmission　risks　in　the　future.