Design And Implementation Of A Tidal Forecast System Based On Improved RBF Neural Network

Tide is one of the most important parts of marine environment.Tide prediction is of great significance in maritime transportation,port construction,the use of tidal energy,etc.With the continuous development of the shipping industry,as well as the requirements of safety and shipping efficiency,the accuracy of tidal numerical prediction is highly required.Nowadays,the most used method of predicting tide wave is harmonic constants.Historical data of the tide also was used to predict the tides by means of nonlinear mathematical forecasting,such as chaos theory,neural network and vector supporting machine.In recent years,the neural network has emerged as a new method in tide forecasts fields.In particular,radial basis function(RBF)neural network has been widely applied in the field of pattern recognition and system prediction.In this article,RBF neural networks has been used for tide prediction and the results have been fully discussed.When we use traditional analysis methods for tidal prediction,sea forecast accuracy decrease significantly in complex factors because we only consider linear effects.In addition,traditional RBF neural networks lack the necessary reasoning and evidence,and some parameters need to be determined based on the specific issue.In this paper,particle swarm optimization(PSO)algorithm is used for optimizing the weights RBF neural network,radial basis function center and width values,and a PSO-RBF neural network model has been established for tide predictions.Our model,basing on the measured data in real-time tidal wave,reflects higher prediction accuracy compared with other major optimization algorithms.The specific contents and conclusions of this paper are as follows:1.RBF neural network prediction model was established by PSO optimization,and the most important ten parameters of effecting tide are selected,which are from the two celestial bodies:the sun and the moon;2.The tide forecast and display system is built up for real-time forecasting and viewing of tide;3.The whole point of monitoring of tidal data in three months of the port was selected as historical data for analysis and prediction.The forecast value of the whole point in the next month has been predicted with relatively accurate results.