To address the issue of inconsistency between the actual primary traveling wave signal and the measured secondary traveling wave signal, an accurate detection method for traveling waves based on complementary ensemble empirical mode decomposition (CEEMD) and least square QR decomposition (LSQR) is proposed. Firstly, the nonlinear amplitude-frequency and phase-frequency response characteristics of the specialized traveling-wave sensor are examined, revealing the distinction between the primary and secondary traveling waves. Secondly, CEEMD is used to decompose the secondary traveling wave into inherent modal function components of different frequency bands. Then the least square method is used to construct the traveling wave inversion model, and the LSQR algorithm is used to solve the inversion components of each inherent modal function component iteratively. Finally, the inversion primary traveling wave signal is synthesized by linear superposition of each inversion component. Simulation and experimental results show that the proposed method is not affected by noise and mode aliasing effects, and the similarity between the frequency-division inversion traveling wave and the real traveling wave can reach 0.99, which realizes the accurate detection of fault traveling wave.