Ensuring Vibration Reliability of Turbopump Units Using Artificial Neural Networks

Abstract

This paper is devoted to developing the scientific approach of using artificial neural networks for solving a significant problem of vibration reliability of rotary machines that is urgently needed to improve the quality of their diagnosis and manufacturing. The proposed methodology integrates analytical dependencies, recent techniques of numerical simulations and artificial neural networks. The design schemes for realizing the related approach are presented on the example of the turbopump unit for liquid rocket engine. The main advantage of this approach in comparison with the traditional regression analysis and other existing techniques is absence of necessity for setting trial imbalances and carrying out additional initial starts of the turbopump unit. The mathematical model for identification nonlinear parameters of the dependence between bearing stiffness, deflection of the rotary axis, and rotor speed is presented. The proposed methodology is proved by the research of rotor dynamics on the example of turbopump units for liquid rocket engines and allows refining parameters of the nonlinear mathematical models describing forced oscillations of the rotor as a complicated mechanical system with nonlinearities. The results of the research can be used for carrying out the virtual balancing procedure for identification the system of imbalances by the reliable model of forced oscillation of the system “rotor – bearing supports”.