Under the impetus of the "dual carbon" goals, a large number of distributed generation have been integrated into the distribution network, transforming the fault current direction from unidirectional flow to bidirectional flow, which makes the traditional distribution network fault location method no longer applicable. An active distribution network fault section location method based on an improved sparrow search algorithm was proposed. High-quality initial sparrow populations were generated using statistical parametric mapping (SPM) chaotic mapping. The discoverer's path was enhanced by integrating the sine cosine algorithm and introducing a nonlinear inertia weight to balance local and global search capabilities, thus accelerating the convergence of the algorithm. Additionally, the follower's path incorporated spiral search and Levy flight to enhance exploration of unknown areas and facilitate escape from local optima. A switch status function suitable for active distribution networks was established, and an anti-misjudgment factor was added to the traditional objective function to prevent information distortion. Case study results indicate that the proposed method can adapt to multiple fault sections and the switching of distributed generation at different locations, significantly advantages both localization accuracy and convergence speed, and demonstrating good fault tolerance.