Ship Roll And Yaw Stability Based On Intelligent Model

Ocean covers two of third earth surface, which has a strategic military value. Ocean is significant for a nation's safety and development, an important part of world economy, as about 90% volume and 70% value of the international trading goods are transported by ship. Improving navigation safety is a forever critical topic in this field. Disturbed by wind and waves, ship has a coupled motion in six degree of freedom, which is harmful for safe navigation. As roll and yaw are most dangerous motions in all these six motions, stabilization of roll and yaw is a main solution to improve navigation safety. However, the uncertainty of ship model in various ocean conditions is a challenging problem in ship motion control. Lack of knowledge about navigation environment, the uncertainty weakened the stabilization effects. Based the National Defense Foundation project in the "11th five years plan", the paper focuses on the roll and yaw stabilization for ship with uncertain model.In this thesis, two classes of uncertain ship models are analyzed for ship motion control. Adaptive LQG (Linear Quadratic Gaussian) control base on intelligent model is proposed for one class of ships which can obtain parts of model information by hydrodynamic experiment. The key idea is to propos a full information model database by mining from the available information. Based the current navigation environment, a corresponding model is selected form the model database, and control rate for this model was calculated by LQG algorithm online. First, a hydrodynamic parameter database in three dimensions is proposed by RBF neural network and experiment data. The three dimensions are navigating speed, sea state number and the encounter angle between wave and navigating direction. Then, we proposed the disturbance model and extended the ship motion model to include disturbance information. Based Kalman Filter and Separation Theory, optimal estimation and control are deduced. Having the current navigating speed, sea state number and encounter angle from sensors, the corresponding parameter model was selected from database for online control.For another type of ship model which is unable to be applied a hydrodynamic experiment and no ready model information can be referenced, self-organizing control system base on intelligent model is proposed. Modeling, training controller and real time control are separated by two independent close loops, called real world system and imagine world system. Independent roll motion and couple motion of roll and yaw are controlled by the self-organizing control system. RTRL, nprKF and EKF algorithm are proposed to train ship forward model and controller respectively.The results of simulation indicate that adaptive LQG control system and self-organizing control system base on intelligent model have different advantages, and both of them are able to effectively stabilize roll and yaw for ship with uncertain model. The two control systems can be applied to control other dynamic plants with uncertain model.