The　prediction　of　the　development　trend　of　internet　public　opinion　is　a　very　significant　reference　for　monitoring　and　control　of　the　network　public　opinion　by　the　relevant　government　departments.　On　account　of　small　sample　characteristics　of　online　public　opinion　and　the　needs　for　both　accuracy　and　stability　in　the　prediction　model,　in　this　paper,　an　improved　grey　wolf　optimization　algorithm　(IGWO)　based　on　the　initialization　of　the　good-point　set　method,　nonlinear　parameter　control　and　the　weighting　of　the　leading　wolf　is　proposed.　Using　IGWO　to　optimize　the　super　parameters　of　SVM　regression　model,　a　network　public　opinion　prediction　model　based　on　improved　grey　wolf　optimized　support　vector　machine　regression　(IGWO-SVR)　is　established.　Empirical　research　is　carried　out　with　Baidu　indexes　such　as　COVID-19　as　public　opinion　data　samples.　The　experimental　results　of　12　test　functions　show　that　the　improved　grey　wolf　optimization　algorithm　has　relatively　strong　global　search　ability,　faster　convergence　speed　and　better　stability.　The　IGWO-SVR　model　has　relatively　outstanding　accuracy　and　stability　in　the　prediction　of　the　development　trend　of　public　opinion,　which　can　provide　better　decision-making　support　for　public　opinion　supervision　department　of　government.　©　2022,　Editorial　Board　of　Journal　of　Systems　Engineering　Society　of　China.　All　right　reserved.