Municipal　solid　waste　incineration　(MSWI)　is　an　effective　method　for　waste　to　energy　in　the　developed　and　developing　countries.　However,　it　also　produces　multiple　flue　gas　pollutants　such　as　NOx,　SO2,　HCl,　CO,　and　CO2.　Due　to　differences　in　MSW　components,　seasonal　and　regional　factors,　the　operational　control　mode　and　pollution　emission　level　in　developed　and　developing　countries　are　different.　In　China,　manual　operational　mode　is　usually　used.　To　reduce　emission　concentrations　of　multiple　flue　gas　pollutants　often　resort　to　injecting　large　quantities　of　cleaning　material　such　as　urea,　lime　water　and　activated　carbon　without　optimizing　the　manipulated　variable　value.　Our　objective　is　to　obtain　the　optimal　＂air　and　material　distribution＂　values　in　terms　of　minimizing　pollution　emissions　and　to　replace　the　empirical　given　values　with　the　manual　control　mode.　An　optimization　method　for　multiple　flue　gas　pollutants　emission　reduction　is　proposed.　Firstly,　based　on　the　experience　of　domain　experts,　the　pollution　model　inputs　dominated　by　manipulated　variables　are　determined.　Then,　considering　the　attributes　of　various　flue　gas　pollutants,　a　novel　hierarchical　incremental　learning　strategy　for　the　interval　type-2　fuzzy　broad　learning　system　is　devised　to　establish　a　multi-input　multi-output　model.　Finally,　a　new　fuzzy　adaptive　particle　swarm　optimization　(FAPSO)　algorithm,　incorporating　the　elite　particle　splitting　(EPS)　strategy,　i.e.,　EPS-FAPSO,　is　introduced　to　determine　the　optimal　values　for　primary/secondary　air　volume.　By　using　a　relatively　stable　operating　condition　data　from　an　MSWI　power　plant　in　Beijing,　the　effectiveness　of　the　proposed　method　is　validated.　And　a　software　system　is　developed　and　realized　on　a　hardware-inloop　simulation　platform,　laying　a　foundation　for　industrial　application.