Aiming　at　the　problem　that　the　accuracy　of　traditional　harmonic　responsibility　division　is　affected　due　to　the　fluctuation　of　background　harmonic　voltage　and　the　variation　of　system　harmonic　impedance　in　the　real　power　system,　a　multi-harmonic　source　responsibility　division　method　based　on　the　pre-screening　of　user-side　data　and　the　classification　estimation　of　system　harmonic　impedance　is　proposed.　First,　user-side　dominant　fluctuation　data　screening　is　performed　based　on　correlation　analysis;　second,　system　harmonic　impedance　data　are　classified　based　on　the　DBSCAN　density　clustering　method;　finally,　system-side　harmonic　impedance　is　estimated　based　on　the　classified　iteration　of　the　screened　dataset　and　harmonic　liability　delineation　is　performed.　The　simulation　results　demonstrate　that　the　method　presented　in　this　paper　effectively　mitigates　the　impact　of　background　harmonic　voltage　fluctuations　and　variations　in　system-side　impedance,　leading　to　more　accurate　classifications　of　harmonic　responsibility.　©　2025　SPIE.