In the process of modern industrial production, the motor has an extremely important role, and the bearing failure is the most common of all faults of the motor fault. To capture the vibration signal information of motor rolling bearing fault effectively and enhance the precision of fault identification, a method for diagnosing faults in rolling bearings combining signal processing and intelligent algorithm is proposed in this paper. Firstly, the ant colony optimization algorithm is emplyed to optimize the two parameters of variational mode decomposition, namely, the number of modes and the penalty factor, in order to get the optimal parameters. Then, using the minimum envelope entropy as fitness function, the optimal parameter feature function (IMF) component is obtained, its index is calculated, and the sample feature vector of the fault model is constructed for testing and training. Finally, the existing experimental data of bearing faults are used for simulation experiments to verify the accuracy of the proposed method.