선박의 자동 항행을 위한 최적 항로 결정과 LOS 가이던스 시스템

Abstract

Recently, satellite navigation systems with low cost are supplied as the result of technological development related to computer and communication. With the aid of such a navigation system and sensors necessary to navigation, a lot of information about ship navigation can be on-line collected and real-time processed on an electronic navigational chart. While, as a large scale of ships are constructed and the total number of operators is decreased in economical aspect, ship navigation circumstance demands more severe requirements in navigation system such as accurate path tracking and collision avoidance. To supplement requirements related to automatic ship navigation, a guidance and control system with a new concept must be composed.

This thesis deals with three main algorithms required to comprise an automatic guidance and control system for general ships equipped with a main propeller as an actuator and a rudder as a yaw angle controller. One of them is an algorithm to generate optimal way points and/or an optimal route based on the ECDIS equipped with the electronic navigational chart. The second algorithm is to automatically generate line-of-sight guidance law for a ship to follow the navigational route composed of way points. The last algorithm is an autopilot control algorithm to accurately track the yaw angle command generated from the guidance system.

At first, this thesis suggests an optimal route decision algorithm based on the electronic navigational chart. The algorithm comprises two steps which are to generate a variety of routes and to search an optimal route to satisfy navigational costraints. In the first step, in order to generate a variety of routes, the closing and thinning technique to deal with the image information of the electronic navigational chart are adopted. The closing technique makes rough coastal line smooth, the thinning technique offers outline of cruisable area and then a variety of routes are decided based on way points. In the second step, a dividing technique of adjacent convex polygons is discussed in order to transform an optimal route decision problem into boadline decision problem. And then, a real-coded genetic algorithm is applied to decide an optimal route or optimal way points by trading off between navigational distance and propulsive energy loss. Through a lot of simulation examples the effectiveness of the suggested algorithm is assured.

At second, a method to comprise a guidance system is suggested. To do this, a guidance law for straight-line routes is derived as a function of advanced ship speed and perpendicular position error. And also, a guidance law for generating turning routes between straight-line routes is derived as a function of yawing angular velocity. Therefore, it is possible to comprise a guidance system for an optimal route composed of straight-line routes and turning routes for automatic ship navigation. The parameters of the suggested guidance system are regulated by the genetic algorithm. While they are optimized under a position tracking error constraint in the coastal navigation where exist unknown currents, they are optimized under a propulsive energy loss constraint in the oceanic navigation. Many simulations for a linear ship model assures the effectiveness of the suggested guidance system realtive to the conventional guidance system.

At last, a comprising method of autopilot system is discussed in order to compensate the tracking error for the yaw angle command. Most of the control algorithms for the conventional autopilot systems are the types of proportional derivative control because they do not posess nonlinearity compensation ability for the real nonlinear ships. In this thesis, a fuzzy PID control algorithm is discussed which can compensate ship nonlinear dynamics and eliminate the steady state error of the yaw angle. A lot of simulations for nonlinear ship models are executed in connection with the suggested guidance system using various navigational routes. In the conclusion, the suggested algorithms in this thesis are turned to be effective through various route tracking simulation studies.