With the construction and development of golden waterway of the Yangtze river, higher requirements are put forward for the accuracy of the prediction of vessel traffic volume in the waterway. Based on the analysis of the characteristics of vessel traffic volume over the years, the control of traffic volume is the ultimate goal. The netural network has a wide range of applications in this regard. The problem with RBF (Radial Basis Function) is the difficulties in determining parameters, such as the weights. The genetic algorithm has the advantages of fast global searching. Therefore, is good for finding the ergodic initial values of weights for the RBF neural network. In this paper, the ship traffic volume prediction model based on genetic algorithm optimization RBF neural network is established for simulation, and the traffic volume data of anqing Yangtze river bridge in anqing section of the Yangtze river is used for verification. The prediction results show that, compared with the ordinary RBF neural network, the optimized RBF neural network has a smaller prediction error. This indicates that the optimized RBF neural network has a smaller calculation quantity, a faster recognition speed, and a smaller prediction error; and that it has a broad application prospect in the prediction of ship traffic volume in the Yangtze river waterway.
With the development of economy, the shipping industry plays an increasingly important role, and the contradiction between advanced navigation technology and backward port transportation management has become a prominent contradiction in the development of waterway transportation in China at present. In order to solve this contradiction, it is necessary to strengthen the management of port ships entering and leaving Port. The prediction of ship traffic volume can provide the basic basis for the planning, design and ship navigation management of waterway, and maximize the navigation ability of waterway. Accurately predicting the traffic volume of heavy waterway ships can provide reliable data for waterway management department, which is of great significance for relieving waterway congestion, improving navigation efficiency and reducing ship traffic accidents. At present, there are many methods used for ship traffic volume prediction, including regression prediction method, time series prediction method and Grey theory prediction method. (Zhang, 2008) However, these methods have some limitations in the prediction of ship traffic volume. Neural network because of its simple network structure, fast learning methods, good promotion ability, but also has a strong information comprehensive ability, good fault tolerance, it can properly coordinate the input information. It is widely used in the field of ship traffic volume prediction and shows its advantages. Some scholars (Fu, Li, Zhang, 2009; Li, Li, Yao, 2006) use the neural network of reverse propagation model (Back Propagation, BP) and Radial Base Function to predict the traffic volume of port ships respectively, and the results show that compared with BP neural network, RBF neural network structure is simpler, learning speed is also fast. The prediction error is smaller, but it also has some drawbacks. For example, the training speed is slow, easy to fall into the local optimization, while the neural network model construction process also has some parameters, such as the number of hidden layer nodes, the center width needs to be repeatedly tested to determine. Genetic algorithm is simple to operate, the parallel search ability is combined into the network weight training of neural network, which can search for the initial point with global traversal, ensure the convergence of network training and reduce the training time. (Qian, Wang, 2009) The application of RBF neural network optimized by genetic algorithm in the field of highway short-time traffic volume prediction shows its advantages. (Lang, Yang, 2010) In this paper, RBF Neural network based on genetic algorithm is applied to the prediction of ship traffic volume in port, and the ideal effect can be obtained by using Anqing port to prove that the optimized RBF neural network is applied to the field of ship traffic volume prediction.
