The Verification And Optimization Of Algorithm For Vessel Intelligent Collision Avoidance Decision-making

Vessel Intelligent Collision Avoidance Decision-making is of great concern to navigation areas at home and abroad, It aims to reduce human factors in ship collisions. Our school research team has been formed Personifying Intelligent Decision-making for Vessel Collision Avoidance (referred to as PIDVCA) theory after a long period study, and construct a PIDVCA algorithm (hereinafter referred to as algorithms).Because of complex encounter situation, restriction of (referred to as collision avoidance rules)and the custom of seafarer, and so on multiple constraints, there are still some problems on rationality, applicability and stability of the algorithm. In order to solve the problems, the paper integrates with the National Natural Science Foundation research project, carried out the verification and optimization of Wide Waters Intelligent Collision Avoidance Algorithm, on the basis of the preliminary study.Firstly,This paper introduces the PIDVCA theory, based on full awareness and understanding of research objectives, ideas and principles, carefully analyse how to transplante the decision-making process of human mind "perception, information processing, knowledge application, analysis and judgment, decision-making " to the computer, making the machine can achieve the key technologies of "personifying intelligent" effects - PIDVCA mathematical model and reasoning process. Algorithm as the core technology, which consists of a series of mathematical models and reasoning processes, combine qualitative thinking process with quantitative calculations through programs ultimately.Secondly, around the rationality and applicability of the algorithm, the paper adopts the combination of theoretical analysis and experiment method to carry out algorithm verification. use Plane Analytic Geometry verify mathematical model and design of targeted simulation experiment to test reasoning process . First draw the geometric graph for various types of established mathematical models, including the avoidance angle, avoidance time, recovery time, new course forecast parameters, limit time and so on, derive the geometric formula for all mathematical models to verify the correctness of the models. Check whether the conditions of encounter features is satisfied to verify the applicability of mathematical models; then based on "safety of navigation and automatic collision avoidance simulation test platform" embedded PIDVCA programs dynamic link library, design simulation experiments to test the algorithm targeted the correctness of reasoning and decision-making process is reasonable. Thirdly, carry out algorithm optimization, which direct against errors and omissions of the mathematical model, negligence and errors of logic reasoning process, and other bugs found out in verification process. According to the geometric analysis of the specific graphic, amends and adds to the mathematical model errors and omissions, solve the applicability problem of the algorithm; Observing the simulation process, tracing debug combined with Visual studio C + + compiler environment, correct the errors of reasoning process. The main algorithm optimitzation of this paper: First, proposed and implemented " optimization of minimum track offset algorithm " to solve unreasonable avoidance effects of special case of the situation; second,study and realize the intelligent decision-making for close-quarters danger encounter which has not been considered before. And the paper improve the algorithm further, enhance the practicality of the algorithm.Through the paper research, solve problems of the applicability, reasonableness and practicality of the algorithm. Therefore, the subject selection is of a important research value.It is of practical significance in promoting the application of research results.