Aiming at the challenges of limited harmonic monitoring equipment deployment and complex harmonic propagation paths in new power systems, this paper proposes a harmonic propagation path and boundary identification method based on voltage morphology comprehensive feature clustering. First, the generalized admittance criterion(GAC) is utilized at edge monitoring points to locally determine the dominant harmonic source, preliminarily defining boundaries and identifying suspicious buses. Second, by integrating piecewise aggregation approximation(PAA) and dynamic time warping(DTW) algorithms, the global distribution and local trend variation features of harmonic voltages are extracted. Comprehensive clustering is then applied to screen buses with voltage morphology highly similar to the target bus, thereby determining propagation paths and tracing ranges. Simulation results based on the IEEE-33 node system show that the proposed method, through edge collaboration, achieves “first orientation, then matching”, effectively reducing data transmission and avoiding misjudgment of distant busbars. This method provides an efficient and low data-dependent solution for harmonic source tracing in complex power grids, offering targeted guidance for subsequent harmonic contribution evaluation and mitigation.