: The increasing penetration of distributed generation has reshaped the operating characteristics of distribution networks, driving their transition from single-ended radial systems to active networks with multiple power injections. The location and capacity of distributed generators may alter the magnitude, direction, and spatial distribution of fault currents, which weakens the applicability of conventional fault-location approaches under complex operating conditions and may reduce the reliability and accuracy of fault-section identification. To address this issue, an improved harmony search-based method is proposed for fault-section identification in distribution networks with distributed generation. First, a fault-information coding scheme and a switching function are formulated according to the fault characteristics of the network, and a fitness evaluation model is developed based on minimum set theory. Then, a Sobol sequence is introduced during harmony memory initialization to improve the uniformity and diversity of the initial harmonies. Meanwhile, a grouped co-evolution strategy based on memeplex partitioning is incorporated to enhance global optimization performance and convergence robustness. Finally, a distribution network model is established in PSCAD, and the proposed method is compared with the standard harmony search algorithm and genetic algorithm through MATLAB-based simulations. The results demonstrate that the proposed method achieves high fault-location accuracy and requires fewer average iterations under single-section faults, multiple-section faults, and FTU information distortion scenarios. Therefore, the proposed method can effectively improve the accuracy and robustness of fault-section identification in distribution networks with distributed generation.