Calculation Method of Ship’s External Magnetic Field Based on Neural Network

Abstract

In view of the difficulties in the calculation of the external magnetic field in the closed-loop degaussing of the ship, this paper analyzes the relationship between the internal and external magnetic field of the ship, and builds a neural network model accordingly to realize the calculating model between internal and external magnetic field. Aiming at the problem of insufficient training set, a data augmentation method based on the linear nature of the induced magnetic field is proposed. By using this method to expand the training set, the training speed and accuracy of the model can be improved. Through experimental verification of the steel double-layer shell model, the proposed magnetic field estimation method can accurately predict the external magnetic anomaly when the induced magnetic field and the fixed magnetic field are changed. This method not only has the advantages of fast calculation speed and high real-time performance, but also solves the long-standing problem that the fixed magnetic field changes are difficult to predict.