Study of Submersible Trajectory Prediction Based on Kalman Filter Algorithm

This paper focuses on trajectory prediction for manned submersibles in seafloor exploration and operations, with an emphasis on ensuring rapid search and rescue after loss of contact. We use a position prediction model based on the Kalman filter algorithm to predict the search and rescue difficulty and joint motion of multiple submersibles. First, we present a model to predict the position of a submersible. By establishing 3D coordinate axes of the sea area where the submarine was lost, combined with the Kalman filter algorithm, we realized an accurate prediction of the position of the submarine. Second, the Kalman filter-based prediction model is modified and added according to the initial position of the submarine, seabed depth, current parameters, and search parameters, combined with the Bayesian updating process to calculate the probability distribution, the model determines the location of the submarine in the process of movement of the densely populated points, and marks the rescue circle in order to guide the rescue operation. Over time, the extent of the search and rescue area expanded, but the search and rescue success rate decreased significantly. Finally, we extended the prediction model to accommodate the multi-submersible scenario. By introducing a mutual collision interference prediction model and a joint submersible prediction model, the SAR efficiency and success rate were improved.