Generalized Predictive Control Of Ship Orientation Based On SVM

The orientation control is a basic part in ship manipulation and control with an aim to maintain and change the ship orientation, as it has direct links with the serviceability and economy of ship navigation and it is also closely related with the sailing safety, energy saving and the labor saving for ship pilots. Due to the defects of the regular automatic helm featuring slow response, the unstable operation of the helm angle and the substantial operation difference, this paper proposes a solution to the defects by combining the system recognition of support vector machine with the basic algorithm of the generalized predictive control to manipulate the ship navigation and orientation in order to achieve the expected result. The research and design of this thesis contain the following aspects:1. The ship's mathematical model is researched, in consideration of the fores and moments from the environmental disturbances, such as winds, waves, currents,etc, it has established a nonlinear ship of mathematic.2. The exposition of the basic algorithm of the generalized predictive control and the introduction to the effects of the generalized predictive control parameter on the systematic properties and the systematic analysis of the basic algorithm of the generalized predictive control. Both the exposition and the introduction have laid a theoretical foundation for this paper's research work.3. The analysis of the related theory of the recognition of the support vector machine. Since the ship performances during its motion have the complex features of nonlinearity and big inertia, this paper proposes to use three different types of the support vector machine to recognize the ship model. The model experiments show that the support vector machine enjoys a sound nonlinearity and identification accuracy. Through comparisons, the author of this paper has selected the v-SVM as the predictive model as it features relatively simpler programming, easier parameter adjustment and easier to achieve expected result.4. As the generalized predictive control is based on the predictive model, this paper also proposes that the model recognition of the support vector machine serve as the predictive model of the generalized predictive control to achieve the expected ship orientation control with the help of the basic GPC algorithm. Compared with the artificial intelligence automatic helm based on the BP nerve network, the more sophisticated generalized predictive control possesses greater robustness and greater ability to overcome nonlinearity and it can provide guiding instructions for the design of intelligence automatic helm of higher performance.