The thermal effect of current is the main cause of cable service life and aging failure. It is very important to establish the cable temperature characteristic model. Power enterprises should be able to correctly estimate the related aging failure rate of distribution network cables. However, the existing cable failure rate estimations are calculated at the rated temperature and do not take into account the effect of actual operating temperature changes. Therefore, a method based on artificial neural network is used to estimate the maximum temperature of the cable. This temperature variation is in good agreement with the daily load curve in some extent. The artificial neural network only needs four easily obtained input variables, and the life loss at each stage of the predicted temperature curve is estimated by using the combined electric heating life model of cable insulation. Finally, the life model and probability failure model are used to predict the failure rate of power cables in the future. The results show that the estimation of failure probability is in good agreement with the actual results, which indicates that the three-stage gradual change curve of cable temperature can truly reflect the cable transient temperature change.