Research On Optimization Algorithm Of The Thrust Allocation For Dynamic Positioning Systems Of Ships

Along with social progress and technological development, dynamic positioning system (DPS) has been widely studied and applied.DPS of ships is a kind of closed loop control system which can resist external interferences as wind, wave and ocean current relying on their own power. It also can keeps ship steadily in given position and yaw angle.Thrust allocation is an important part of the Dynamic Positioning system. Its role is allocate the expected force and moment output signals into speed and direction of the thrusters.Thrust allocation problem can be expressed as a constrained nonlinear optimization problem.Thrust allocation optimization, not only can improve the relative performance of DPS, but also play an active role in reducing errors, reducing fuel consumption and equipment wear.In this paper,engineering background is the experimental model boat CyberShipII of the Norwegian University of Science and Technology (NUST). For the Thrust allocation problem of DPS for ships, the mathematical optimization model is established. In the past, sequential quadratic programming (SQP) was adopted to solve this problem. But SQP possesses some defects such as local convergence and poor robustness. According to the algorithm development trends, this paper adopts particle swarm optimization (PSO), which has simple structure, fast convergence and easy to implement, to solve this problem in order to get better results. Specifically, this paper improves basic PSO and proposed a parallel immune particle swarm optimization (PIPSO). The basic idea is making use of two parallel sub-populations, each sub-group separately in accordance with the global version and local version of PSO for the speed and location update. In this way it can combine the advantages of two versions of PSO, i.e. with fast convergence and preventing from premature. Moreover the immune mechanism is introduced into this algorithm for making the exchange of information between sub-populations basesd on immune mechanism. This can further accelerates the convergence of the algorithm and effectively prevent premature. To verify the performance of proposed algorithm, at first it applied to solve the classical function optimization problems which the optimal solutions are known. The results show its feasibility and effectiveness.On this basis, the proposed algorithm is applied to CyberShipII thrust allocation optimization problem further. Two typical environments are selected for making the corresponding simulation tests. Furthermore, two different parameters of the proposed algorithm are selected for comparing and analyzing its optimization performance in different cases.Related work shows that the proposed algorithm is effective for thrust allocation optimization problems and it can obtain better solutions. This study has a certain theoretical significance and application value.