At　present,　there　are　serious　air　pollution　problems　in　most　cities　in　China.　As　one　of　the　main　atmospheric　pollutants,　PM2.5　has　caused　serious　harm　to　people＇s　health.　In　order　to　improve　the　accuracy　of　PM2.5　concentration　prediction,　this　paper　proposes　a　new　hybrid　model　based　on　complementary　ensemble　empirical　mode　decomposition　(CEEMD)　and　Long　Short-Term　Memory　(LSTM)　to　predict　daily　PM2.5　concentration.　The　daily　PM2.5　concentration　and　meteorological　data　from　January　2010　to　December　2014　released　by　the　US　Embassy　are　selected　as　experimental　data.　Compared　with　extreme　learning　machine　(ELM),　Support　Vector　Regression　(SVR)　and　Long　Short-Term　Memory　(LSTM),　the　CEEMD-LSTM　model　shows　a　higher　prediction　ability.　©　2019　Technical　Committee　on　Control　Theory,　Chinese　Association　of　Automation.