This　paper　tries　to　apply　particle　swarm　optimization　(pso)　algorithm　to　improve　the　BP-neural　network,　and　the　second　water　source,　three　water,　four　water　XAJ　parameter　calibration,　the　predicted　results　are　compared.　The　results　of　different　models　of　river　basin　water　right　choice.　This　paper　mainly　studies　the　BP　neural　network　based　on　PSO　algorithm　of　distributed　four　water　xin　an　river　model　calculation,　this　paper　did　research　work　includes　the　following　aspects:　(1)　based　on　the　research　of　the　common　water　level　model,　select　the　appropriate　parameters,　establish　proper　data　model　(2)　based　on　the　research　of　the　common　prediction　algorithm,　BP　neural　network　as　the　main　algorithm　to　parameter　calibration,　and　apply　the　PSO　algorithm　to　optimize　the　BP　neural　network.　©　2018　ACM.