The　water　crisis　has　become　one　of　the　major　challenges　facing　mankind　in　the　21st　century.　As　a　water-scarce　country,　China＇s　sustainable　socio-economic　development　has　been　affected.　Therefore,　good　water　resources　planning　is　of　great　importance　to　social　and　economic　development.　The　water　consumption　forecast　is　a　prerequisite　for　the　smooth　implementation　of　water　resources　planning.　Water　use　forecasting　is　often　difficult　due　to　problems　such　as　incomplete　consideration　of　influencing　factors　in　the　forecasting　process.　In　this　study,　we　propose　a　water　consumption　prediction　model　combining　Spatial　Clustering　and　Multi-layer　Perceptron　neural　network.　This　method　not　only　strengthens　the　spatial　characteristics,　but　also　further　improves　the　accuracy　of　prediction.　©　2022　IEEE.