Study On Modeling And Nonlinear Adaptive Control Of Underwater Vehicle

Unmanned Underwater Vehicle?UUV?is an effective tool for the human to explore and exploit the deep ocean,which proposes more accurate and stable requirements for the develop-ment of the vehicle system when vehicles are utilized in many activities of the human being to discover ocean like inspecting underwater infrastructures or deploying submarine equipment.It is not easy to understand the underwater vehicle,especially the nonlinear unstable underwater vehicle,in nature or manipulate it precisely in the exploration underwater.Models for the un-derwater vehicle are utilized to explain the relationship between movements and forces exerting on the robot.Yet currently there are no general methods can arrive a better model with structure and parameters for vehicle automatically in the discovery of model datasets.And considering the thruster configuration,the designed shape,and weight as well as the unknown underwater environments,the system described in mathematics or dynamic shows time-varying evolution in the model.Moreover,changes in the underwater environment,sensor instruments or actua-tors carried as well as some unexpected disturbances and obstacles in the operation will make the control system become unstable.Therefore,it is necessary to discover the modeling of a vehicle,perceive underwater surroundings,and develop a suitable controller that can adapt to nonlinearities,uncertainties in the vehicle system from the perspective of theoretical research and control applications.In this thesis,the modeling and control of the underwater vehicle are investigated.CFD numerical computation and the system identification are proposed to modeling the vehicle for control scheme.In order to improve the adaptation of the vehicle to the coupling disturbances caused by the system and the environment,the nonlinear control of the unstable underwater vehicle system is carried out.In this paper,some interesting approaches that can model the underwater vehicle via discovering datasets and controllers are proposed.Results from numer-ical simulations and experiments are both demonstrated and discussed,which proves to be an appropriate solution in theory and technique for further application in water.Contributions to the modeling and control of a vehicle in this thesis are listed as the following:?1?Aiming at the difficulties of determining the structure and parameters of an underwa-ter vehicle,the discovery and the parametric identification problems of the underwater vehicle model are studied utilizing the experimental data.Gene trees that can describe the movement mathematical equation are introduced.The evolution process of the model expressed in gene trees using symbolic regression method is also given out.Discarding the human bias in the study of the modeling for a vehicle and the limitations of knowledge in the field of expertise,the intrinsic relationship hidden in the measured datasets of a vehicle is revealed intelligently,which empowers the data the ability of self-discovery.Exact mathematical equation with a new model structure is obtained using experimental data via investigating the structure and parame-ters of the model,which provides a new approach to re-understand the vehicle system and the basis for future controller design.?2?Considering the effects of the water flow situation on an underwater vehicle,a prediction method that can effectively identify the water flow pattern is proposed the mechanism of the fish lateral line that can perceive water flow.The computed fluid dynamic?CFD?technique is carried out to extract sensor data.Linear discriminant analysis?LDA?method in the view of the concept of compressed sensing is employed to reduce the dimensionality of data.Support vector machine approach is utilized to train the water flow direction classifiers and the framework that fusing water flow forecasting and an inertial navigation system is proposed.In the thesis,the ability of vehicle identifying current like the fish lateral line is investigated in theory,which also provides a new point of view for an underwater vehicle to exploit ocean.?3?Aiming at the difficulty of developing more accurate vehicle model for control scheme,the method for determining the control model parameters is given based on the geometric model and the nonlinear model of a vehicle.The model uncertainties considered by the robust adaptive control are neglected,and the vehicles with complex shape and streamlined bodies are both modeled.CFD software is used to calculate the key terms of the underwater vehicle dynamics model:the additional mass and the damping matrix.In order to facilitate the adaptive control application when the theoretical vehicle model is described clearly,the nonlinear formula of the vehicle is simplified by using Taylor expansion method.?4?Aiming at the nonlinear control problem of an unstable underwater vehicle system sub-ject to disturbances,the problem of the general nonlinearity on the control performance of un-derwater vehicle is studied based on the robust adaptive control framework.Considering the uncertainties of the model,nonlinearities,the coupling relationship of each DOF and the in-terferences from the environment?salinity,mechanical shock?,the basic diagram of the robust adaptive control scheme as well as the stability criterion is introduced.Aiming at the pitch chan-nel control of a 6 DOF underwater vehicle system with static instability in water,the projection based model reference adaptive control?MRAC?and L₁adaptive control scheme are both used to deal with this problem.Performance results of control are demonstrated and it is found that the L₁adaptive control surpass MRAC whether in the response speed or in the adaptability of control approach to stabilize the vehicle system.?5?Aiming at the nonlinear control problem of an unstable underwater vehicle system sub-ject to disturbances,the problem of the actuator nonlinearity on the control performance of underwater vehicle is studied based on the robust adaptive control framework.With the pur-pose of coping with the time-varying model parameters of a vehicle as well as the nonlinearities of actuators like input saturations,dead-zone,and time-varying delay,the anti-windup com-pensator is extended into the structure of MRAC and L₁adaptive control to optimize the con-troller outputs.The attitude and position control of a 6 DOF underwater vehicle in the pitch channel and the depth mode are studied and simulated,which proves the effectiveness of the proposed control scheme.The robust and adaptive performances of L₁adaptive control with anti-windup compensator are demonstrated and compared with the controller utilized in the Chapter 5,where measurements noise,unexpected disturbance and limitations of actuators are considered.Results show that the L₁adaptive control with anti-windup compensator can adapt to the disturbance quickly.The study on modeling and nonlinear adaptive control of an underwater vehicle is presented in this work.The methods proposed can be used in the research of various systems,and can provide a new perspective for the human being to investigate an underwater vehicle in theory and apply vehicle in the future exploration.