Research on Model-aided Dead Reckoning and Path Following Control Methods for Flat UUVs

Recent years, countries all over the world have enhanced the deploitation, utilization and contest of the ocean resources rapidly because of the fast increasing requirement for resources. Unmanned Underwater Vehicles (UUVs), as a special equipment for resources exploration, environment monitoring and maritime reconnaissance, have the advantages such as small in size, friendly in controllability, long duration and great carrying ability that make them suitable for all kinds of military scientific and economic tasks. However, UUV’s own complexity as an integration of high-techs, in addition to the unpredictability and complexity of the ocean environment make it difficult to finish missions safely and successfully. As a result, in order to improve the feasibility and reliability of UUV, it is of great significance to combine relative theories with the dynamic characteristics and control requirement and to investigate the dynamic model of UUV as well as methods concerning about position keeping, space guidance and motion control.Dead reckoning is one of the most popular navigation methods. Because the path following control is relied on the vehicle’s position, the mission could not be realized when the dead reckoning is out of work if the DVL is failed. Without adding any sensor, this paper mainly studies about model-aided dead reckoning and spatial path following control method based on the flat UUV’s dynamic model in order to deal with the failure of dead reckoning using Doppler Velocity Log (DVL).1. Establishing a6-DOF dynamic model for a flat UUV and figuring out main disturbances during path following control.Concerning about the dynamic model, the paper researches about the problems of motion and model and analyzes about the characteristics of a flat UUV. Based on the Newton-Euler equations, motion characteristics are analyzed according to the model, geometric shape and distribution of actuators. Main disturbances during the motion control process are considered, including unstructured uncertainties, current, model uncertainties, wave and the vehicle’s hydrodynamic parameters.2. Proposing a TLS algorithm for dynamics identification and realizing dominant hydrodynamics estimation in the three subsystems.During parameters estimation process, mistakes may be caused by errors of data matrix and observer vector. In order to solve this problem, a total least squares (TLS) algorithm based error criterion is proposed to estimate parameters according to the three essential factors of dynamic model hydrodynamic parameter estimation after a deep exploration of estimation methods. To solve the problem that UUV dynamic model which is presented as Newton-Euler equations is too complex to be suitable for applications in real world engineering, three subsystems, including speed subsystem, steering subsystem and depth dynamics, are established mainly involving the dominant hydrodynamic terms. Finally, identification results are testified by simulations. The identification of UUV dynamics is used for model-aided navigation and spatial path following control.3. Putting forward a model-aided dead reckoning algorithm and validating its correctness by post-processing simulation of lake-trials data.While the UUV is sailing underwater, DVL measured velocity maybe ineffective due to the geographic shape of the bottom, medium status and sometimes temporary failure of DVL. This occasional ineffectiveness will lead to a great error of the dead reckoning. Furthermore, it will also cause security problems such as unknown status of UUV and function issues such as inability to finish the missions. To solve the problems mentioned above, a model-aided dead reckoning algorithm is proposed based on the UUV dynamic model. At the same time, to eliminate parameter bias of the model-aided navigation system, current estimation for velocity compensation is introduced into the system. Finally, the whole system is testified with simulation by post-processing lake-trials data, which is satisfied with navigation accuracy. This proposed algorithm is a solution to estimation error caused by the failure of dead reckoning using DVL.4. Presenting an idea of spatial separation and planar guidance for3-D path following control, where the controllers are designed for the established three subsystems based on Lyapunov theory. Furthermore, simulation results demonstrate the proposed idea with a mission for an offshore platform using the given controller.Firstly, to eliminate the disturbance in surge control system caused by model approximating error, coupled hydrodynamic terms and environment disturbances, an adaptive neural network (NN) controller with parameter correction is proposed to achieve the surge control. Secondly, in order to eliminate the influence in depth control caused by varying parameters and uncertainties due to time varying hydrodynamic damping, RBF neural network is utilized for estimation and its optimal estimation is achieved through an adaptive law. Meanwhile, virtual control input is calculated by backstepping method which would guarantee the tracking error converge to a small neighbourhood zero domain. Thirdly, due to the underactuated characteristics of heading control system, horizontal control issues are decomposed into geometric issue and dynamic issue. Line-of-sight (LOS) guidance laws for straight line path and circle path are proposed respectively under the assumption that surge speed of UUV is controllable. Then the adaptive NN heading controller is designed. The simulation results show that the speed subsystem, steering subsystem and depth subsystem are controlled in expectation respectively. Finally, a simulation for three-dimensional path following experiment is designed according to engineering cases. Results show that the designed spatial path following controller is not only easy to tune the controller parameters, but also can achieve path following of spatial curve with zero error rapidly. The proposed control scheme completes the mission well and has a good robustness to the disturbances. Meanwhile, the same mission is finished pretty well using MADR when the DVL is failed.This paper mainly studies about dynamics identification, model-aided dead reckoning and spatial path following control method. Its results have important practical values to underwater vehicles whose missions need to be finished safely and reliably.