To address the difficulty in accurately detecting and locating single-phase-to-ground faults in active distribution networks under random fluctuations of distributed energy resources and disturbances, this paper proposes a fault sensing and section location method based on the linear eigenvalue statistics of zero-sequence voltage. Firstly, from the perspective of high-dimensional space, the advantages of zero-sequence voltage in constructing fault sensing and location methods are analyzed. Secondly, a sliding window data matrix of node zero-sequence voltage and its covariance matrix are constructed, and the linear eigenvalue statistics are used to sense the fault occurrence time. Faulted-section localization is achieved based on the node contribution of abnormal eigenvectors and the difference index between adjacent nodes. Simulation verification is carried out on a 33-node distribution network under complex disturbance scenarios including no access of distributed generation, access of distributed generation, and neutral grounding via an arc suppression coil. The results show that the proposed method can effectively identify the occurrence time of single-phase-to-ground faults and achieve high-precision faulted-section localization under complex disturbance conditions.