Study On Unmanned Surface Vehicle Modeling And Logic Networks Adaptive Control Method

In recent years, more and more attention has been paid on unmanned surface vehicles (USV) with advantages of small volume, good flexibility, high speed and no casualties. Because of high order, nonlinear, strong coupling and high uncertainties, it is difficult to derive an accurate dynamic model of USV. Furthermore, the six degrees of freedom (DOF) motion model of USV has strong randomicity and nonlinearity under wind, waves and other disturbances, which make it difficult to control using traditional control methods. Efficient control systems are very important to improve the observation effect of reconnaissance equipment and the precision of the weaponry system. The USV control system needs to strengthen its control ability, self-study and self-adaptive abilities for highly dynamic and unpredictable marine environment. Therefore advanced control strategies are kind of need to get satisfactory maneuvering performance. Intelligent control methods are suitable to solve these problems. This dissertation investigates mathematical modeling method and adaptive control methods of USV to slove the above problems. The main work of this study includes the following aspects:The six DOF model of USV is derived in detail by analyzing all the hydrodynamic forces and moments on the USV including environment wind and wave disturbances. This dissertation adopts the MMG modeling method, in which the hydrodynamics are decomposed into three parts that are hull, propeller and rudder, with adequate consideration of mutual interference.Logic-based fuzzy networks (LFN) and their application in USV control system are implemented. A five layer LFN framework is established using AND_U and OR_U unineurons. Combined with the line-of-sight guidance method, a design of USV path following system is presented based on the LFN. With prior domain knowledge, the training results can be easily interpreted and directly translated into a series of logic expressions formed over a collection of information granules. Hybrid learning strategies are used in this study. Firstly genetic algorithm is used to minimize the network structure, and then pruning algorithm is adpoted to adjust the structure. Gradient-based learning and paticle swarm optimizition algorithms are used in parameter optimization. Considering the characteristics and requirements of the ship and ocean engineering, numerical simulations are conducted on Matlab/Simulink software under several ocean environmental conditions. The simulation results demonstrate the effectiveness of the proposed approach.An adaptive controller with combinations of online SVR algorithm and feedback linearization theory is proposed for USV. Starting from adaptive inverse control, an online adaptive SVR algorithm is adopted to identify the control plant’s inverse model. The adaptive USV heading control system is presented. Besides that, using simplified track models, a SVR adaptive control method based on feedback linearization for USV track control system is proposed. The identificated inverse model is used as a controller, and adaptive inverse track control system is established through off-line identification of the inverse dynamic model and compensation of inverse error. Using matlab/simulink as the digital simulation tool, numerical simulations are conducted. The simulation results demonstrate that the proposed algorithm provides an alternate effecitive way for USV motion control.