10 th International Research/Expert Conference ”Trends in the Development of Machinery and Associated Technology” TMT 2006, Barcelona-Lloret de Mar, Spain, 11-15 September, 2006 APPLICATION OF GENETIC ALGORITHMS ON SHIP CONTROL SYSTEMS Fuat ALARÇIN Berna BOLAT Faculty of Mech. Eng. Faculty of Naval Architecture and Marine Eng. Yildiz Technical University, Besiktas, 34349 Istanbul Yildiz Technical University, Turkey Besiktas, 34349 Istanbul Turkey ABSTRACT Artificial intelligence based ship control systems such as fuzzy logic, neural networks and genetic algorithms. Nowadays, Genetic algorithms (GAs) are being used in a widespread way in optimization problems. In this paper, the course keeping autopilot using genetic algorithm is applied to stabilize the yaw motion of the containership. The simulation and optimization of the ship control systems with genetic algorithm have been obtained by running computer software programmed. Simulation results indicate that the autopilot system with genetic algorithm is able to obtain good performance in cases of rough sea conditions. Keywords: Ship Simulation, Control System, Genetic Algorithms. 1. INTRODUCTION The problem of the course keeping has been widely studied by numerous authors [1, 2, 3]. Most work has focused on the system using classical Proportional-Integrative-Derivative (PID) controllers. This type of PID autopilots was first presented by Minorsky [4]. This controller is widely used in marine engineering control systems. A ship autopilot designed based on the PID control is simple, reliable and easy to construct. However, the autopilot performance in various environmental conditions is not satisfactory [5]. In order to improve the performance of the controller, Genetic algorithms [6, 7] which are the optimization technique based on the principles of natural evaluation and population genetics have been employed in the ship motion control research to minimize output error. The first step in the genetic algorithms is to generate a random population, where each individual is represented as a set of parameters. The next step of the algorithm involves evaluating the fitness function, which indicates the quality of each individual. After the fitness function is evaluated in every generation of the genetic algorithms, three main operators which called selection, crossover and mutation are applied. Selection is used to select the individuals from the population, according to their fitness value. The Crossover operator mixes randomly the features of two individuals by the combination of their genes. There are several crossover techniques such as one-point crossover, two-point crossover, uniform crossover, etc. The main aim of mutation operator is replaced the character of genes randomly. Genetic algorithms have also found widespread used in controller optimization particularly in the fields of fuzzy logic [8] and sliding mode controllers [9]. The application of genetic algorithms to PID parameter optimization is better tuning technique than Zegler-Nichols technique [10]. The objective of this paper is to accomplish changes of yaw motion with minimal overshoot and with minimal oscillation. 1475