Decision Support Model Based On The Cbr Method Of Ship Collision Avoidance

It is an information time and the intelligent, networking, and ship-bank incorporation are the development trend of ship automation. Highly intelligent ship is the most promising one in the 21 century. The research and application of the Intelligent Decision Support System for Vessel Collision Avoidance (IDSS-VCA) has some meanings for sailing security, avoiding collision, reducing the loss of life and property and protecting the ocean's environment. As an important component of IDSS-VCA, the maritime affairs are an indispensable researching aspect of vessel collision avoidance. In order to reduce or avoid the ocean incidents, we need to make some statistics and analysis on these maritime affairs and learn some lessons from them, as well as improve the ship manipulating method. The validly storage and usage of these maritime affairs not only can help us to find a new way to solute the collision avoidance issue, but also can make us a great step to the intelligent collision avoidance research for the case reasoning is an effective simulation model of human thinking and reasoning.Case-Based Reasoning (CBR) is a technique that emphasizes the importance of past experiences which are collected and stored as cases in a case base. A case documents the relationship between a problem and its solution. A CBR system solves a new problem by retrieving, updating and reusing a similar one from a case base. Therefore, CBR is a useful way for solving the collision avoidance problem, because cases can well document human experiences. As a new subject of ship collision avoidance researches, this thesis focuses on developing a case-based reasoning decision support model for vessel collision avoidance. Our research works involving the basic structure design of the model, which include three parts, that is, modeling of collision case-database, case learning and using. Comparing with the other decision making methods for vessel collision avoidance, this model is easier in acquiring knowledge, simpler in reasoning and more convenient in system structuring. The CBR model will become a new development in vessel collision avoidance making.Firstly, the design process and the basic structure of the case base are introduced. Then, the primary principles for extracting information of cases from maritime affairs or navigation data are proposed. In the following, the case representation method with combination of object-oriented and frame characteristics, and a hierarchical indexing tree are discussed in detail. A case learning method based on an evaluation system for vessel collision avoidance process is presented which can be used to evaluate the sailing operation, create the new case and update the case base. The nearest neighbor case retrieval algorithm is also developed. The experimental results demonstrate the validity of the model. Finally, the conclusions and future work are discussed.So, it is necessary and feasible to create a decision making model for vessel collision avoidance based on the CBR techniques, which can combine the research achievements of maritime affairs and automatic collision avoidance by the case creation, storage, index, retrieval and learning. Although the experiment results showed the usefulness and applicability, we still have to conduct large-scale exclusive experiments for different navigation environments and more complicated encounter simulations in order to evaluate the approach effectively.