Based on artificial intelligence technology and neural network calculation method, this paper summarizes the important factors that affect the ship domain model, and reasonably introduces them into the domain model as the input, and the ship domain size as the output, so as to obtain the scientific and reasonable quantitative boundary or the dynamic boundary adjusted immediately according to the function model. This paper systematically summarizes the risk quantification model and research results of ship collision avoidance decision optimization, comprehensively analyzes the results of collision avoidance decision model based on knowledge base, and constructs the evaluation model of ship collision avoidance decision optimization according to multi-objective genetic algorithm. Through the design of ship collision avoidance algorithm, we can plan a safe and short collision avoidance path, which not only ensures the safe navigation of the ship, but also reflects the ship's automatic collision avoidance intelligence. It has a certain practical significance for the research of ship collision avoidance system and algorithm.