A　hybrid　intelligent　control　optimization　method　based　on　multi-objective　evaluation　(MOE)　is　proposed　to　control　the　production　indices　of　a　process　and　maintain　them　in　their　desired　range.　This　method　consists　of　a　loop　control　layer　with　an　air-fuel　ratio　adjustment　model　and　a　setting　layer.　The　set　points　of　the　control　loops　are　produced　by　case-based　reasoning　(CBR)　and　followed　by　the　closed　loop.　These　set　points　are　also　adjusted　through　the　real-time　evaluation　and　online　correction.　The　proposed　approach　has　been　applied　to　the　roasting　process　and　its　effectiveness　has　been　verified　by　the　results　of　practical　application.　(C)　2012　Elsevier　Ltd.　All　rights　reserved.