Prediction of thrust bearing parameters in shafting-shell coupling system from frequency response function with artificial neural networks

Abstract

The thrust bearing is a critical component for power transmission and vibration coupling between the propulsion shafting and the shell of the underwater vehicle. Identifying thrust bearing parameters is crucial for studying the vibration and diagnosing bearing faults of the underwater vehicle. In this study, key information was extracted from the frequency response function (FRF) of the shafting-shell coupling system, and then the artificial neural network (ANN) was used to predict the stiffness and damping of the thrust bearing. The dataset used to train the ANN came from the analytical dynamic model of the shafting-shell system. This analytical approach offers high computational efficiency, making it feasible to generate a substantial amount of training data within a reasonable timeframe. In the analytical dynamic model, the shell was modeled using the Flügge theory, while the shafting system was modeled using the Euler-Bernoulli beam theory. The bearings were simplified as a spring-damping system to represent the connection between the shafting system and the shell. This study employed two ANN algorithms: Backpropagation Neural Network (BP) and Genetic Algorithm-optimized Backpropagation Neural Network (GABP). The results indicate that both BP and GABP effectively predict the stiffness and damping of thrust bearings. Moreover, GABP demonstrates more stable prediction results with smaller prediction errors. The proposed method for predicting thrust bearing parameters leverages features from the FRF to train the ANN, which provides good robustness, maintaining effective results even when the signal-to-noise ratio of the FRF is reduced. The thrust bearing parameter prediction model was validated through experiment, confirming the effectiveness of using ANNs to predict bearing parameters in shafting-shell coupling systems from FRF. This study realizes efficient prediction of bearing parameters, providing a reference for vibration reduction, operational state monitoring, and fault diagnosis of underwater vehicles.