The Mathematical Models Research For Analysis And Prediction Of Sea Wave Data

The investigation of sea wave is one of the important fields in oceanography. It provides environmental information for sea transportation, fishery, oil development and military activities and so on, furthermore guarantees the safety of maritime activities.For a lot of random characters and undetermined factors, the prediction of sea waves is the hot point and difficult point in oceanography, the difficulty and accuracy of the prediction are still key issues to solve.On the basis of real history data of significant wave height, we researched that the suitable range of the models, GM(1,1), steep GM(1,1), moving average model, cubic exponential smoothing model and ARMA, when used to do prediction for sea wave, and the improvements are applied according to prediction condition analysis.Firstly, grey GM(1,1) model is established to predict the significant wave height. Through the simulation experiment and analysis, the grey GM(1,1) model is suitable for the prediction of significant wave height with monotonous smooth data.Secondly, steep GM(1,1) model is established to predict the significant wave height with the oscillation characters.Thirdly, the paper take the significant wave height data as time series, and establish the prediction models of moving average. For the data with the experimental analysis, it is found that the model is disadvanced for the determination of moving coefficients and the averaging of the data. The paper do some improvement to build right moving average prediction model pointing to the disadvantages of the moving average model with the experimental analysis, it is pointed out that its disadvantage is experienced determination of the right coefficients. Thus, the paper build the cubic exponential smoothing model, which follows the rule of "light weight from the past" and the prediction condition is good.Finally, for unstable character of sea wave, the paper do stabilized processing to the significant wave height data, then establish Auto-Regressive Moving Average model(ARMA) for the processed data, it is pointed out that it is suitable for the prediction of significant wave height, and confirmed that through massive simulation experiments.