Abstract: Traditional detection methods exhibit significant performance degradation when processing data containing transient signals with substantial energy disparities or operating in colored noise environments. To address this issue, this study proposes an adaptive detection algorithm that combines the robustness of power-law detection with the edge extraction precision of variable-threshold cumulative sum detection. The methodology involves three key steps: First, optimal detection sequences are extracted on a frame-by-frame basis through power-law detection, guided by the maximum signal-to-noise ratio criterion derived from normalized Rayleigh coefficients. Second, the near-zero rate serves as a feature indicator to determine the presence of transient signals within each frame. Finally, variable-threshold cumulative sum detection precisely extracts transient signal edges. Simulation results demonstrate that compared to conventional power-law detection and variable-threshold cumulative sum methods, the proposed approach achieves performance improvements of approximately 2 dB and 5 dB respectively in white noise environments, and nearly 7 dB in colored noise conditions, while maintaining a transient signal leading-edge detection accuracy of 0.0073 seconds.