To　accurately　predict　the　oxygen　content　in　flue　gas　during　the　municipal　solid　waste　incineration　process,　this　paper　uses　mutual　information,　differential　evolution　algorithm　and　stochastic　configuration　network　to　select　the　features　of　the　oxygen　content　in　flue　gas　and　predict　the　modeling.　Firstly,　the　mutual　information　method　is　used　to　eliminate　some　irrelevant　variables.　Secondly,　the　differential　evolution　algorithm　is　combined　with　the　stochastic　configuration　network　to　further　eliminate　redundant　variables　from　the　above　selected　feature　variables,　so　as　to　determine　the　input　variables　for　predicting　the　oxygen　content　in　flue　gas,　and　train　the　stochastic　configuration　network　prediction　model　for　the　oxygen　content　in　flue　gas.　Finally,　the　actual　data　sampled　from　a　solid　waste　incineration　plant　in　Beijing　is　used　to　test　and　verify.　The　results　show　that　the　hybrid　feature　selection　method　and　prediction　modeling　method　in　this　paper　are　effective　and　can　accurately　predict　the　oxygen　content　in　flue　gas　during　the　solid　waste　incineration　process.　©　2023　IEEE.