The　Middle　Route　of　the　South-to-North　Water　Diversion　project　(MRP)　effectively　alleviates　the　problem　of　serious　shortage　of　water　resources　in　North　China.　However,　owing　to　the　long-term　operation　of　this　project,　abnormal　algal　proliferation　has　occurred　in　the　upper　reaches　of　the　main　canal,　and　the　water　diversion　outlet　is　easily　blocked　by　the　large　colonies　of　algae　falling　into　the　canal,　which　has　seriously　threatened　the　water　quality　and　water　supply　safety　of　the　canal.　In　light　of　this　issue,　this　study　coupled　the　reinforcement　learning　model　with　the　one-dimensional　hydrodynamic　model　of　the　series　sluice　group　to　build　an　ecological　scheduling　model　for　prevention　and　control　of　algal　proliferation　in　the　MRP.　The　model　was　applied　to　the　canal　section　(approximately　300　km　long,　from　No.　G2　to　No.　G17)　where　in　algae　grows　easily.　The　results　show　that:　(1)　The　model　can　achieve　the　desired　reciprocating　adjustment　of　the　canal　water　level　in　the　process　of　decline　stability　rise　within　0.8　m　and　stably　maintain　the　low　water　level　for　more　than　72　h;　(2)　during　the　process　of　ecological　scheduling,　the　maximum　drawdown　of　water　level　satisfying　the　safety　regulation　and　control　constraints　of　the　MRP　is　1.0　m;　(3)　for　an　ecological　regulation　water　level　drop　greater　than　0.3　m,　it　is　more　difficult　to　restore　the　initial　water　level,　and　the　water　level　rise　range　should　therefore　be　greater　than　the　decrease　range;　(4)　the　constructed　model　exhibits　high　robustness.　Even　for　an　observation　error　level　less　than　0.08　m,　the　success　rate　of　the　model　can　still　exceed　80%.　The　ecological　scheduling　scheme　generated　by　the　model　can　be　used　to　destroy　the　algal　habitat　conditions　through　hydrodynamic　dynamic　regulation,　whereby　decisional　support　can　be　provided　for　the　prevention　and　control　of　abnormal　proliferation　of　algae　in　the　main　canal.　©　2021　Elsevier　B.V.