Study On Vessel Preventing Collision Decision-making Based On Genetic Algorithm

With the development of computer technology and information technology, the study on automatic and intelligent navigation becomes a promising research direction. The study of vessels' intelligent collision avoidance system is usually to search for the optimal evasion strategy in collision situation. Genetic Algorithm (GA) is one kind of optimization algorithms. When solving optimal object problems, it can get the tradeoff between the solving speed ant the optimal key. Using the decision-making variables' some kinds of coding, GA begins to search for the optimal result in the population which consists of many initialized schemes. It obtains and accumulates information of the searching space automatically. And it can control the searching process auto-adaptively. GA uses operation to produce new population, and by controlling corresponded parameters, using the fitness function to evaluate individuals in each generation. By evolving continually, it can get the optimal or relative optimal result. This paper is an attempt to study vessel's avoidance collision based on GA. It can find the better avoidance collision scheme in many schemes. The best preventing collision scheme can be obtained by considering the safety and economy and the international regulations for preventing collisions.In the introduction, the domestic and international research work on vessels' preventing collision is briefly reviewed and the purpose of this paper is given out. In the second chapter ,the author analyses keystone of vessels' preventing collision and gives out the parameters' calculation methods of vessels' movement, and analyses the aims and request of preventing collisions decision-making. In the third chapter, after expounding the GA's keystone and characteristic, by combining the keystone of vessels' preventing collision, a preventing collisions model is set up. The model not only includes the selected line's safety and economy, but also takes account the international regulations forpreventing collisions. Three genetic operators are designed to operate the preventing collision lines, which are randomly produced in initialization procedure. After a certainty generations' evolution, the relative optimal scheme can be achieved. The measure to evaluate the algorithm's performance is designed. And we can optimize the designed algorithm by regulating the operator parameters. In the fourth chapter, the author uses Visual C++ to set up the emulator based on GA .When using the emulator, we can check the results' validity and evaluate the results. In the last chapter, the author summarizes the results and points out the deficiency of the study. Finally, the possible further study of this field is discussed. From the performance result, we can conclude that the preventing collisions simulation system set up in this paper is correct, and it is efficient to provide the decision-making scheme. By improving the preventing collisions model and the genetic operators, the system can work as an assistant decision-making system. And it can provide important information for vessels' preventing collision decision-making.Wang Ze-sheng (Transportation Information Engineering & Control)Directed by Associate Prof .Cheng Jina-biaoProf. Shi Chao-iian...