Study Of Regional Tide Forecasting Based On Support Vector Machine

Tide is one of the most important components in the marine environment.Tide forecasting plays an important role in ship transportation at ports.In actual production and life,in order to ensure the safety of maritime transportation and improve the efficiency of shipping,port terminals usually need to set up as many tide observation stations along the coast to obtain accurate tide water level information and avoid ship stranding accidents,and make full use of Port water depth resources to improve ship turnover rate and increase ship cargo load.The creation of more tide observation stations requires more manpower,material resources,and financial resources.In order to reduce costs and ensure the accuracy of tidal water level forecasting,we introduce the concept of regional tidal forecasting and predict the tidal water level information of other tidal testing stations in the area based on the water level information of limited tidal tidal testing stations in a specific area.Then,reduce unnecessary tidal measuring stations in the area and saving costs.Firstly,tides are periodic movements caused by the gravity of celestial bodies on the sea water.The main method of tide forecasting is the harmonic analysis method,and the average error of the forecasting is 20?30cm.This method expresses the tide as a combination of several sub-tides,and uses the harmonic analysis method to determine the parameters of each sub-tide.However,the harmonic analysis method only considers the influence of the tidal astronomical tide in tidal prediction,and ignores the influence of time-varying factors such as wind,air pressure,water temperature,and precipitation on the tide.In practical applications,the use of machine learning algorithms in the field of prediction is a new research direction that has emerged in recent years.Among them,support vector machines have a strong ability to learn and express complex nonlinear mapping relationships.This article will use support vector machine for tide forecasting and improving the accuracy of tide forecasting.Secondly,a regional tidal forecast model is established.Support vector machine is used to establish a map,and the water level information of a limited number of tide gauge stations in a specific area is selected as input.The water level information of other tide gauge stations in the same area is selected as the prediction target output.By training the support vector machine model.The model can fully learn and master the tidal water level relationship between the tide gauge station and the target tide gauge station in the selected area.and select the water level information of several measured tidal gauge stations in the specific area for pordiciton,the fitting result is compared with the fitting result of the traditional Autoregression(Autoregressive Models,AR)linear model,which proves that the fitting precision of the regional tide forecasting model of the support vector machine is higher than that of the traditional AR linear model.Finally,further expand the scope of regional tide forecasting and compare the accuracy of tide forecasting at different sites.Through simulation results,as the scope of regional tide forecasting expands,the accuracy of tide forecasting gradually declines.Therefore,it is still necessary to ensure enough measurement points to obtain higher-precision prediction results in large water scale.In addition,in order to meet the actual tide forecasting requirements,multi-step forecasting at different sites is carried out.The experimental results prove that the regional tide forecasting model based on support vector machine proposed in this paper has stability and reliability in tide forecasting and has practical application value.