Fatigue Life Prediction Method of Engine Rotor Based on GA-BP Neural Network

Due to the combined influence of temperature and stress on rotor fatigue life, ensuring accurate prediction is challenging. Therefore, this study proposes a method for predicting engine rotor fatigue life based on a GA-BP neural network. The temperature distribution of the rotor during high-speed rotation in high-temperature and high-pressure environments is analyzed using energy conservation, mass conservation, and momentum conservation equations. Additionally, the stress distribution within the rotor is determined by considering mechanical balance and geometric equations. These parameters serve as inputs for the GA-BP neural network to predict engine rotor fatigue life through forward propagation and back propagation. The test results indicate that the method's prediction error for rotor fatigue life is generally within 60.0 h.