Actual industrial robots are prone to failures in harsh working environments.Traditionally,vibration signals are used for fault diagnosis.However,vibration data is difficult to collect in actual factories,which causes great trouble in the fault diagnosis of industrial robots.To solve this problem,an intelligent fault diagnosis model of industrial robots current data based on wavelet packet energy spectrum (WPES) and convolutional neural network (CNN) was proposed.The original current signal was decomposed into multiple sub-bands by using wavelet packets,and the corresponding energy characteristics of each sub-band were calculated.When an industrial robot failed,the energy characteristics would change,and the energy spectrum characteristics were converted into a two-dimensional matrix for the design,training and testing of the proposed model.The experimental results show that using WPES-CNN model for fault diagnosis,the fault recognition rate reaches over 99.9%.