Acoustic emission (AE) techniques have been rapidly developed for use as powerful fault-diagnosis tools, especially in fault localization. Traditionally, AE signals were directly used for localization on simple models such as pipes and plates without subdivision; however, this effect is usually not ideal for complex structures. In this study, a new method capable of obtaining the precise time of arrival for AE signals is proposed to localize the faulty planet gear in a wind turbine gearbox; the core localization concept is to determine a valid physical parameter that presents the linear correlation with distance. Because compressive waves present a constant speed on any given frequency band, continuous wavelet transform based on the Morlet wavelet is used to extract those waves from other components of AE waves. Results obtained after applying this method to actual planet gear fault localization are satisfactory, indicating that this method offers several advantages to fault localization in complex structures. (C) 2017 Published by Elsevier Ltd.