Intelligent Optimization Of Neural Network And Its Application In Tidal Level Forecast

The tidal water level information plays an increasingly important role in human activities.Accurate and real-time ocean tide information is of great significance for the safe navigation of marine ships,the design of coastal docks and ports.With the development of shipping industry,the requirements for the accuracy of the tide information are also increasing.In addition,with the large-scale development trend of ships,providing real-time and accurate tide water level information can greatly reduce accidents such as stranding or collision with bridges.The most traditional and widely used prediction method for tidal level prediction is harmonic analysis.However,the traditional harmonic analysis often needs to obtain a large number of tidal observation records,which can lead to a high cost of money and time.Therefore,it is less feasible to obtain such a large amount of long-term observations for some ocean ports.In addition,the traditional harmonic analysis method mainly considers the role of movement force of celestial bodies.However,the ocean tide is also affected by many time-varying nonlinear factors,such as temperature,air pressure,precipitation and so on.Therefore,the variation trend of tides shows some uncertainties and time-varying.It is difficult to obtain accurate real-time tidal level forecast information using the traditional static model.With the rapid development of artificial neural network,neural network model is widely used in the field of system prediction.In this paper,the most widely used(Back Propagation,BP)error back propagation neural networks is applied to forecast tidal water level so as to improve the prediction accuracy.However,the traditional BP neural network model has some shortcomings:the weights and thresholds of the model may affect the convergence speed and prediction accuracy of the whole system to some extent.If the parameters of the network model are not chosen properly,the simulation model may fall into local optimum,and it is also easily affected by the network topology.In order to further improve the traditional BP network model,a BP neural network tidal water level prediction model based on Self-adaptive Particle Swarm Optimization(SAPSO)is established in this paper.Base on the real measured tidal level data of Isabel tidal gauge,the traditional harmonic analysis model and BP prediction model are compared with the model proposed.In addition,the SAPSO-BP forecast model established in this paper is applied to the regional tidal water level forecast,and a regional tide level forecasting model is established.This paper uses the measured tidal level information of Isabel tide gauge station to do the model simulation experiment.The final experimental results show that the prediction accuracy of SAPSO-BP model issignificantly improved compared with the traditional harmonic analysis method and BP prediction model.