Aiming at the problem of excessively low signal-to-noise ratio(SNR) in distributed optical fiber acoustic sensing systems, an optimized noise reduction method is proposed, which is based on the combination of time-domain localized generalized maximum correntropy criterion (TLGMCC), adaptive noise complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN), and lifting wavelet transform(LWT). This method first uses adaptive noise CEEMDAN to decompose the original signal, obtaining modal components. Then, the original signal and these modal components are segmented into multiple time-localized segments, and their corresponding local time segment correntropy values are calculated. Afterwards, the LWT algorithm is applied to process weakly correlated components, and finally, the remaining components are reconstructed to complete the denoising process. Experimental results show that under the conditions of a 5 km sensing distance and 10 m spatial resolution, the system's SNR reached 54.36 dB, while the root mean square error was reduced to 0.091.