An abnormal motion condition monitoring method based on the dynamic model and complex network for AUV

Abstract

This paper proposes a method based on the Autonomous Underwater Vehicle (AUV) dynamic model and complex network theory to effectively monitor the abnormal motion conditions of the AUV on the vertical plane. First of all, we simplify the AUV’s dynamic model and use a neural network to identify its parameters. The error between the dynamic model’s output and the AUV measurement is employed to build a weighted complex network, which describes the AUV motion condition pattern. Afterwards, the feature matrix is extracted from the complex network. Finally, the Support Vector Machine (SVM) is introduced to classify the feature matrix, and the three abnormal motion conditions are classified. The experimental results show that the method proposed in this paper can effectively monitor the abnormal motion conditions of AUV. Compared with other advanced methods, the performance of the proposed method is greatly improved.