In　order　to　obtain　the　nodal　flow　pattern　of　sanitary　sewer　networks　for　the　unsteady　flow　simulation,　the　main　element　and　the　periodic　element　of　the　measured　flow　series　were　identified　according　to　the　non-stationary　time　series　analysis,　and　the　sum　of　the　main　element　and　the　periodic　element　was　the　nodal　flow　pattern.　The　moving-weighted　least-square　(MWLS)　method　and　the　parametric　model　were　used　for　the　main　element　identification　respectively.　The　accumulative　periodic　diagram　was　used　for　the　periodic　element　identification.　Finally,　the　identified　flow　series　were　compared　on　the　basis　of　a　group　of　daily,　hourly　sewage　flow　series　of　a　wastewater　treatment　plant　in　Beijing,　and　the　correlation　coefficient　and　the　residual　mean　square　between　the　identified　flow　series　and　the　measured　flow　series　were　taken　as　the　evaluation　criteria.　Results　show　that　MWLS　method　is　superior　to　the　parametric　model　in　identifying　the　main　element　and　MWLS　method　combined　with　the　accumulative　periodic　diagram　is　superior　to　the　parametric　model　combined　with　the　accumulative　periodic　diagram　in　identifying　the　nodal　flow　pattern.