Unbalance position of aeroengine flexible rotor analysis and identification based on dynamic model and deep learning

Abstract

The rotor system design is gradually becoming lighter as the thrust-weight ratio of the new generation of aeroengines improves, making the power turbine rotor structure more slender and flexible, and it is difficult to judge the axial distribution of unbalance. Meanwhile, the engine rotor’s working speed is above the second critical speed and even exceeds the third-order speed. The small imbalance of the rotating shaft will cause large vibration, coupled with the influence of mode shape, resulting in difficulty in rotor dynamic balance, low efficiency and poor effect of the flexible rotor dynamic balance. Therefore, finding out the unbalance position is of great significance to determine the balance plane during rotor dynamic balance, and improve the balance efficiency and effect of flexible rotor. Firstly, the paper established the dynamic model of power turbine rotor, analyzed the vibration response law of rotor with unbalance at different positions, and found out the sensitive position of rotor unbalance. Then, a power turbine rotor unbalance position identification method, based on deep learning, one-dimensional convolution neural network (1D-CNN) is proposed. The multiple measured response data of the rotor with different unbalance positions are input into 1D-CNN for training. Finally, at subcritical speed, critical speed and supercritical speed, the accuracy of 1D-CNN method for rotor unbalance position identification is 99.5%, 99.5%, and 99.0% respectively, which verifies the effectiveness, feasibility of 1D-CNN method for rotor unbalance position identification, and provides a new solution for intelligent identification of aeroengine rotor unbalance position.