Study On Control Of Ship Anti-rolling On The Basis Of The Hyrid Genetic Algorithem

The rolling control is established as a subject for over 100 years, during which there were more than 350 equipments of anti-rolling or means of the rolling control. Compared with fin stabilizer, the initial outlay upon equipment of rudder roll stabilization is less. Both have the nearly same effect of anti-rolling, moreover, the anti-rolling by rudder will not have underwater noise after the system is closed. One notable character of the anti-rolling by rudder is that the method is sensitive to the math model of ship. The change of the variable of ship model can reduce the effect of anti-rolling or even lead to failure. So the designed controller must have the good robust stability. The common character of fuzzy system and neural network is that, dealing with and solving the problems, they do not need the precise math model of the object. The fuzzy-neural network is a technique combined with the powerful structural knowledge expressing capacity of fuzzy logic reasoning and the powerful self-learning capacity of neural network. The fuzzy control has the controlling experience of people as knowledge model of control and has fuzzy set, fuzzy language variable and fuzzy logic reasoning as math tool of controlling algorithm. Neural network is non-linear dynamics system and has many good character and capacity, such as parallelism, store distribution, high non-linear, self-learning and self-organization, good robustness and learning association. Eventually, we apply the genetic algorithm and the simulated annealing algorithm to optimize the variable of fuzzy membership function and the weights of neural network. It is the problem of this paper to apply the new controlling method to solve the present problem of rudder roll stabilization.