Abstract: In view of the problem that harmonic signals measured at the point of common coupling (PCC) of actual wind farms are often disturbed by noise, a harmonic denoising algorithm combining complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and improved wavelet threshold is proposed. The algorithm adaptively decomposes the noisy harmonic signal into a series of intrinsic mode functions (IMFs) by using CEEMDAN, separating noise and signals according to different frequency ranges. Then, the correlation coefficients between each IMF component and the original harmonic signal are calculated respectively, and high-frequency IMF components with correlation coefficients lower than a set threshold are selected for improved wavelet threshold denoising. After merging with the reserved IMF components, a pure harmonic signal is reconstructed. The experiment builds a grid-connected simulation model of a doubly-fed wind farm based on the MATLAB/Simulink platform to analyze the denoising algorithm. The results show that the denoising effect of the joint denoising algorithm is ideal, which is superior to the other four comparative algorithms in both qualitative and quantitative aspects, laying a foundation for the subsequent accurate analysis of components in harmonic signals.