Wind　energy,　as　one　of　the　renewable　energies　with　the　most　potential　for　development,　has　been　widely　concerned.　At　the　same　time,　the　medium　and　long　term　wind　power　prediction　is　easily　affected　by　many　factors.　In　order　to　avoid　the　instability　of　a　single　model,　this　paper　first　builds　a　self-adaptive　filtering　model　and　a　gray　model　with　parameter　optimized　by　teaching　learning-based　optimization　(TLBO),　then　uses　ordered　weighted　averaging　(OWA)　to　assign　weights　to　two　single　models,　and　finally　uses　Markov　chain　to　modify　the　prediction　results　to　further　improve　the　prediction　accuracy.　©　2021　Institute　of　Physics　Publishing.　All　rights　reserved.