Research on a Support System for Ship's Berthing Maneuvering Using Optimal and Neural Network Theory

在本研究中，作者成功地进行了以下几方面的开发和实验：1.利用方差理论将船舶最短时间操纵问题转化为两点边值问题，并利用共轭梯度恢复法求解。利用神经网络理论，对无最优解的机动情况，求出近似插值解。利用动态GPS和卡尔曼滤波理论，在东京湾立山开发高精度船舶定位系统。开发全自动靠泊系统，控制船舵、可调螺距螺旋桨、船首和船尾推进器。开发以下三个阶段的系统并进行实际实验：（1）第1阶段，没有用于跟踪最优解的支持系统;（2）第2阶段，允许在危险情况下进行小规模手动操作;（3）第3阶段，具有完整的独立系统，以实现考虑最短时间最优解的自动和安全的停泊操纵系统。6开发视觉模拟器，用于在立山和东京港使用盐地丸进行实际海试的初步实验和验证。

In this research, the authors succeeded the following developments and experiments :1. To formulate minimum time maneuvering of ships as two points boundary value problem in theory of variance and gain the numerical solutions using Conjugate Gradient Restoration Algorithm2. To gain approximate interpolation solution for maneuvering case without optimal solution, using a neural network theory3. To develop a high precision ship's position system using a kinematic GPS and Kalman Filtering theory in Tateyama in Tokyo Bay4. To develop full automatic berthing system for controlling the rudder, controllable pitch propeller, bow and stern thrusters in Shioji Maru5. To develop the following systems with three phases and execute the actual experiments :(1) PHASE 1 without supporting system for tracing the optimal solutions,(2) PHASE 2 with permitting small manual operation in the case of dangerous situation(3) PHASE3 with complete stand alone system to realize an automatic and safe berthing maneuvering system considering the minimum time optimal solutions.6 To develop a visual simulator for preliminary experiments and verification of the actual sea trials in Tateyama and Tokyo Port using Shioji Maru.