Underwater propeller fault diagnosis based on deep learning

Abstract

When an underwater robot performs a task, the propeller is most likely to malfunction, such as being entangled by foreign objects or the blades are damaged. At present, its fault diagnosis methods have problems such as relying on manual feature extraction and using neural networks with low accuracy. Therefore, this paper proposes an integration based on an improved one-dimensional convolutional neural network (1D-CNN) and a long short-term memory network (LSTM). Thruster fault diagnosis method. By analyzing thruster data, accurate diagnosis of four different thruster faults can be achieved. A comparative experiment was conducted between the proposed model and some traditional algorithm models. The results show that the proposed method has greatly improved the test accuracy, and this method can effectively diagnose underwater robot thruster faults.