The　C02　emissions　from　municipal　solid　waste　incineration　(MSWI)　processes　significantly　influence　dual-carbon　goals　and　ecological　environmental　protection.　This　article　proposes　a　method　for　modeling　C02　emission　concentrations　based　on　long　short-term　memory　(LSTM).　Initially,　the　Z-Score　method　is　applied　to　handle　outliers.　Subsequently,　the　mutual　information　(MI)　algorithm　assesses　the　correlation　between　features　and　C02　emissions,　facilitating　feature　selection.　Finally,　the　LSTM　algorithm　constructs　a　predictive　model　for　C02　emission　concentrations.　The　effectiveness　of　the　proposed　method　is　validated　using　real　C02　data　from　the　MSWI　process　in　Beijing.　©　2024　IEEE.