Abstract: In order to provide the maritime department with the scientific and accurate prediction of the essel traffic flow, this paper proposes an algorithm based on the cultural firefly algorithm (CFA) to optimize the generalized regression neural network (GRNN) to predict vessel traffic flow. This paper introduces the statistial method of fairway traffic flow based on automatic indentity system (AIS). Fast rejection test and cross stand test are used to judge whether the ship track passes through the observation line of a certain cross section of the channel. The latitude and longitude data in the AIS data is converted to the Mercator plane coordinate system. The realization principle of GRNN is studied. CFA takes the output mean square error of GRNN as fitness function. Its encoding takes the weight value in the input and hidden layers, the weight value in the hidden layer and the output layer, the threshold of the hidden layer and the threshold of the output layer. Its evolutionary goal is to get the most suitable and optimal structure of GRNN. With the statistics and pretreatment data collected from AIS, the CFA-GRNN is used to predict the vessel traffic flow, and the experimental results and errors are analyzed. The results show that CFA-GRNN has better generalization performance and higher accuracy of the prediction results than GRNN and firely algorithm-GRNN, and is not easy to fall into local optimum. This study is of theoretical and practical significance for forecasting and analyzing ship traffic flow.

Key words: vessel traffic flow prediction; general regression neural network (GRNN); cultural algorithm; firefly algorithm (CFA)