As a key component of machine tool spindle,the remaining useful life prediction of rolling bearings directly determines the remaining life of the whole mechanical equipment.If the health status or damage of rolling bearings cannot be predicted in time,it will not only affect the formulation of maintenance strategies,but also cause cascading failures,which is likely to cause catastrophic accidents of mechanical equipment.Aiming at the problem of adaptive fault feature extraction and intelligent diagnosis of rolling bearing vibration signals under big data,a life prediction model combining convolutional neural network and long short-term memory network (CNN-LSTM) was constructed,which could avoid the influence of manual participation and realize the complementary advantages of the network.The degradation state and residual life of rolling bearings were predicted,and compared with convolutional neural network (CNN) and long short-term memory neural network (LSTM).The experimental results show that CNN-LSTM has higher prediction accuracy.