Prediction And Optimization Of Wave Height Based On Machine Learning Algorithm

Ocean wave is an important process of ocean surface movement,and machine learning is a hot research field in the world.This paper combines Marine science and machine learning to explore the feasibility of the application of Marine science and machine learning,laying a foundation for future research.Two different algorithms in machine learning are used to predict and correct the effective wave height.(1)The Support Vector Machine(SVM)was used to establish the prediction model,and the wind field and wave field were selected as the learning elements to compare the influence of different feature vectors on the prediction results of effective wave height.The east sea area of Taiwan island was taken as the experimental area,and the numerical model data of NCEP reanalysis was used as the learning sample.Using the support vector classifier,four sets of models with different feature vectors are established to predict the effective wave height of ocean waves,and the results of the four models are compared and analyzed.Experiments show that too many or too few eigenvectors will have different effects on the prediction results and computational efficiency of the model.When wind field and wave field are used as eigenvectors for learning,the prediction results in this region are closer to the model prediction results,with a correlation coefficient of nearly 99% and a root-mean-square error of about 0.2m.(2)The Neural Network model is used to establish the model of pattern result optimization to optimize the effective wave height of MASNUM wave mode.The wind field and wave field were selected as the learning elements,and the south China sea was selected as the experimental area.According to the seasons,three models were divided into summer,autumn and winter for training.Compared with satellite observations,the corrected results show a 30% reduction in root-mean-square error and absolute mean difference.