A Single-Ended Fault Location Method for Distribution Networks Based on Signal Injection and Similarity of Reflected Wave Energy Distribution

Conventional traveling-wave methods (TMW) for fault location in multi-branch distribution networks typically rely on multi-terminal measurements. To address this limitation without increasing the number of sensors, this paper proposes a single-ended traveling-wave method (SETWM) capable of simultaneously achieving fault distance estimation and branch identification. First, a double-exponential excitation signal is injected into the de-energized line, and the reflected wave signals are acquired. The acquired reflected wave signals are then transformed into the time–frequency domain using the Wigner–Ville distribution (WVD), and energy distribution vectors (EDVs) are extracted within specific time–frequency windows. Subsequently, cosine similarity is introduced as a discriminant metric to quantitatively evaluate the similarity between EDVs. By applying a predefined similarity threshold, three categories of fault-characteristic reflected waves (FCRWs) are identified within the reflected wave group. Finally, the propagation time differences of the FCRWs are used to obtain three distance parameters related to the fault location. Combined with the network topology, this enables precise identification of the faulted branch and accurate estimation of the fault distance. The proposed method is validated through both EMTP/ATP simulations and experimental tests. Results demonstrate that the method effectively identifies FCRWs and accurately determines the faulted branch for fault resistances ranging from 0.1 Ω to 5000 Ω, achieving a location error of less than 40 m.