Research On Vessel Intelligent Steering Control And Its Applications

Because the unmodeled dynamics such as the nonlinearity and uncertainty of ship motion model and external disturbance and so on, vessel steering control is a complicated controlling problem, and the traditional ship course controller is difficult to achieve ideal control effect. This paper proposes a dynamic fuzzy neural network algorithm based on T-S fuzzy model, which is able to generate accurate and parsimonious dynamic neural network on-line quickly and realize self-organization and parameters’identification of fuzzy system. And the algorithm is applied to Hermite function approximation, dynamic system identification and Mackey-Glass chaotic time-series prediction to confirm the effectiveness. Finally the algorithm is applied to design of intelligent ship course controller, and the main research work is as follows:First, based on the analysis of modeling mechanism of Abkowitz model and MMG model, the simple responsive model (Nomoto model) is achieved. And parametric process is conducted in the model. The simple model obtained is used as the model basis in the design of intelligent ship course controller.Second, a kind of self-organizing fuzzy neural network which extracts T-S fuzzy rules in samples is proposed. Based on the training data on-line or off-line, the method incorporates the pruning strategy into growth criteria to learn fuzzy rules, and grow into a compact network structure with restriction. In addition, the linear least squares (LLS) method is used to identify the parameters. To prove the superiority of the algorithm, it is applied to static function approximation, dynamic system identification and the chaotic Mackey-Glass time series prediction, and compared with some other famous learning algorithms. And the result show that the method achieves parsimonious structure with fast learning speed, perfect approximation performance with high accuracy.Finally, we apply the method based on dynamic fuzzy neural network in intelligent vessel steering controller design. The data samples are generated by nonlinear PID ship steering controller and used to train the intelligent vessel steering on-line or off-line. Furthermore, the performance of ship maneuvering prediction and course-keeping is proved by a series of typical steering commands. The simulation study demonstrates the effectiveness and superiority of the intelligent ship course controller.