Research On The Fitness Of Intelligent Optimization Algorithms To Some Applications In Ship Engineering

The ship design involves various aspects, such as ship perfornance, strength of structures, resistance and propulsion, and maneuverability, etc. Therefore, it is a typical decision-making procedure of multi-disciplinary and multi-objective optimization. The conventional design process highly depends on trials and makes desions through comparisons on the quantitative analyses. It is hard to achieve the optimum results. By learning the evolutionary mechanism of nature, intelligent optimization algorithms have been developed to conduct optimization. They are acess to a wider range of problems with strong capability of global search and good robustness, particularly for those problems with highly complex search space and multiple conflicting objectives. Thus, these algorithms have attracted extensive attentions in the area of naval architecture and ocean engineering with focus on ship design and performace predictions.In this dissertation, major intelligent optimization algorithms are examined and compared. Some conclusions are drawn after thorough studies over each algorithm. Then, three examples regarding ship initial design, structure and materials and maneuverability are solved by using the corresponding algorithms with expectation to expand their applications in the area. Major works are summaried as following.In Chapter2, five intelligent optimization algorithms are analyzed, including Genetic Algorithm (GA), Particle Swarn Optimization (PSO), Ant Colony Optimization (ACO), Simulated Annealing Algorithm (SA) and Artificial Neural Network (ANN). Through a typical benchmark example of Travleing Salesmans Problem (TSP), these algorithms are compared with each other on time complexity, space complexity, and convergence and so on. Therefore, this chapter lays foundation for further studies in next chapters.In Chapter3, ship initial design is studied under multidisciplinary and multi-objective optimization (MDO) by using GA. A MDO flow is prosped with one system control level and5subsystems. They are (1) buoyancy and stability subsystem;(2) resistance subsystem;(3) propulsion subsystem;(4) maneuverability subsystem, and (5) cost subsystem. The NSGA Ⅱ was employed to obtain the approximate set of Pareto solutions and TOPSIS was adopted to rank these solutions. The good performance ship model has been obtained and its principal dimensions are the global Pareto solutions. Then, the turning motion and zig-zag maneuver of the optimized ship was simulated under the actions of winds, waves and currents so as to verify its feasibility.In Chapter4, ANN was used to substract the conhesive strength and fracture energy of toughened high strength steel for ships. This grade steel is strongly nonlinear fracture with irregular crack surface. Thus discrete cohesive zone model (DCZM) was used to study their fracture behavior. However, the determination of cohesive strength and fracture energy is disturbing. A method was proposed to construct response surface by ANN to obtain the two parameters. The results were comapared to other two approaches. The load vs. displacement curve predicted by using the determined parameters matches the experimental data quite well. This proves that the proposed method can work well for such problems.In Chapter5, GA was used to tune the auto-disturbances-rejection controller (ADRC) parameters. A three-step procedure was established.(1) The proper TD settings were verified through the step response.(2) The parameters of the ESO were designed. A NSGA-Ⅱ method was used to obtain the approximate parameters in line with observer, and it achieved a satisfied solution.(3) The ADRC three modules were combined to decide NLSEF settings. An example of15,000t tanker course controller shows that the proposed method can easily get the optimal available parameters to design the ADRC. Compared to PID, the designed ADRC has less energy consumption, smaller overshoot and stronger robustness. Thus, the proposed was verified.In summary, this dissertation makes efforts on the fitness of intelligent optimization algorithms to some applications in ship engineering. Three quite different problems in the area were studied. This will benefit to expand applications of those algorithms in ship engineering.