Conducting a study for extracting the fault feature information of the rolling element bearing under variable conditions. Three typical state components of cylindrical roller bearings（normal bearing，outer ring wear，rolling element wear）are selected as the research targets. An experimental bench for a cylindrical roller bearing of a special vehicle gearbox is built. The vibration signals of cylindrical roller bearing failures under different input speeds are collected through the experimental bench. The generalized Stockwell transform（GST）is used to transform the raw vibration signal into the time-frequency domain，and the obtained two-dimensional time-frequency matrix is used as the feature matrix. The characteristic parameters are obtained by performing the singular value decomposition（SVD）on the feature matrix. The extracted characteristic parameters are input into a Support Vector Machine（SVM），and the SVM is used to realize the identification of different states of the rolling bearing. The results show that the proposed method can effectively achieve the vibration signal feature information extraction and state recognition under variable operating conditions. It can provide an effective mean for the online monitoring of rotating machinery equipment.