The failure caused by multiaxial fatigue can have negative influences on the life of aircraft engines. Thus, life-prediction models on multiaxial fatigue problems are of vital importance in maintaining the safety and reliability of engines. In this passage, experiment data of five different materials are collected for further research. Then we identify abnormal data and clean them out to for normalization process. After that, we introduce a Multiple Layer perception model (MLP) and a Convolutional Neural Networks (CNN) model to use processed dataset for training and test. We make a comparison between predicted values and real ones to analyze its fitting condition, then we calculate the mean-square error and relative error to estimate the accuracy of our models. The result shows that MLP has a better performance in two materials while CNN has a higher prediction accuracy in some dataset. Finally, we testify that there are no overfitting problems and prove that both models have good generalization ability.