Research On Forecasting System Of Qiantang River Tidal Bore Based On Nonlinear Theory

While appreciating the tidal bore in Qiantang River,which is famous for its unique charming,people should neglect the harms brought by it.The great personal and property losses are brought to people on both sides by the tidewater of Qiantang River every year.If capable of providing accurate forecast information about the tidal bore,the harms brought by the tidal water could be reduced.The methods for forecasting tidal bore in Qiantang River are discussed through chaos theory in this paper.The content mainly includes chose identification of tidal bore,tidal time difference sequence and tidal level.Besides,it is forecasted by combining BP neural network and SVM model.The main job contents are as below:(1)Study on chaotic characteristics of tidal bore system of Qiantang River.Firstly,the existing tide predication theory and the methods that have been adopted for forecasting are analyzed.Secondly,the causes for the chaos phenomenon are discussed.It indicates that the hydrologic system existing in the natural world is accompanied with lots of chaotic behaviors.Therefore,the forecast model of tidal bore in Qiantang River is optimized through chaos theory in this paper.However,its success is dependent on the dynamic characteristics of tidal bore system.So,the application of saturated correlation dimension and maximum Lyapunov index method is mainly analyzed and the qualitative and quantitative analysis on chaotic characteristic of tidal bore system of Qiantang River is carried out in this paper.(2)Forecast of tidal hour based on chaos optimization of BP neural network.For the tidal bore and hour every day meets the change law of lunar calendar,the quasi-periodicity is strong and the originally implied chaotic characteristics cannot be easily manifested,the time difference of tidal bore every other day to the tidal hour is built.Through analysis on sequence chaos and reconstruction of its phase space and by taking each phase point as the input parameter of BP neural network,the evolution laws revealed by the reconstructed phase space are studied through BP neural network.The possible errors in tidal hour caused by tidal time difference sequence are ultimately forecasted through BP neural network model based on-chose optimization.Through forecasting four observation stations of Qiantang River,the result shows that RMSE value of four stations is reduced by 83.9% by contrast of traditional method and averagely reduced by 71.18% by contrast of BP neural network.It indicates that the model is reliable and possesses high forecasting accuracy.(3)Tidal level forecasting based on chaos optimization of SVM.By contrast of tidal time difference sequence,the quasi-periodicity of tidal level sequence is strong.The calculation oftracing effect of tidal level sequence through exclusive use of SVM is firstly analyzed.Though the trend of original tidal level sequence could be traced,its precision remains to be increased.Therefore,SVM model is optimized through chaos theory.RMSE value of its forecast result is reduced by 60.28% by contrast of the original model,which proves that the chaos theory could revise the model effectively.