Aiming at the difficulty of feature extraction caused by random noise and interference signals often accompanying polyethylene pipelines, this paper proposes a feature extraction method based on refined composite multiscale entropy of acoustic emission from polyethylene pipeline leakage, and analyses the attenuation characteristics of the acoustic emission signal of PE pipeline leakage. Firstly, according to the acoustic wave attenuation theory, it is proposed to use the attenuation coefficient to describe the attenuation characteristics of PE pipeline leakage AE signals under different pressure conditions, and then the leakage acoustic emission signals of polyethylene pipelines under different pressure conditions are extracted, and it is found that the acoustic emission parameters such as amplitude, energy, RMS, and ASL can reflect the changes of the leakage acoustic emission signals with the pipeline pressure and the detecting distance by using the least-squares simulation method. Finally, a refined composite multi-scale sample entropy based AE feature extraction method for polyethylene pipeline leakage is proposed, which compared with the traditional multi-scale sample entropy, has a certain degree of anti-interference ability for noise signals and irrelevant signals in the leakage signals. And this paper provides a way of thinking for the identification of the acoustic emission from polyethylene pipeline leakage in the next step.