基于支持向量机的模块化潮汐预报

发布时间：2018-03-17 13:39

本文选题：潮汐预报　切入点：模块化　出处：《大连海事大学》2017年硕士论文　论文类型：学位论文

【摘要】：潮汐预报在船舶航行、港口建设、沿岸生产等领域应用广泛。随着船舶的大型化以及对航运效率要求的提高,对潮汐预报精度的要求越来越高,提高潮汐预报精度愈发成为相关领域的热点。而支持向量机作为近20年来最受关注的机器学习算法之一,在处理非线性问题中具有其独特的优势,支持向量机在潮汐预报中的应用是该领域的一个新的研究方向。本文将支持向量机作为工具,对支持向量机在潮汐预报中的应用进行了探讨。目前,传统的潮汐预报方法为调和分析法,但由于调和分析法在预报过程中主要考虑了天文潮因素的影响,导致在环境变化剧烈的情况下预测精度较差,难以满足部分场合对预报精度的要求。本文针对调和分析法对非天文潮部分无法精确预测的问题,提出了一种基于支持向量机的模块化潮汐预报模型。其中,支持向量机用于对潮汐的非天文潮部分进行较为精确的预测,并结合调和分析法稳定的优点,有利于预测精度的提高。首先,按照潮汐的成因,将其分为主要受天体引潮力影响的天文潮部分和受气象、水文以及其它因素影响的非天文潮部分。其中天文潮部分通过调和分析法进行预测,而针对非天文潮部分,由于其受到多因素的影响,具有很强的非线性特征,本文使用具有良好非线性回归能力的支持向量机来预测该部分,这种模块化处理方式有效综合了两种方法的优势,提高了预测精度。随后,为其设计了图形用户界面,该界面直观地表示出了潮汐预报结果和仿真过程。最后,利用Honolulu港的实测数据对预报模型以及界面进行相关仿真试验,试验结果证明,相较于传统潮汐预报方法,本文提出的模块化方法准确、便捷、易操作。
[Abstract]:Tidal forecasting is widely used in the fields of ship navigation, port construction, coastal production and so on. As one of the most concerned machine learning algorithms in recent 20 years, support vector machine (SVM) has its unique advantages in dealing with nonlinear problems. The application of support vector machine in tidal prediction is a new research direction in this field. In this paper, the application of support vector machine in tidal prediction is discussed. The traditional tidal forecasting method is harmonic analysis method. However, due to the influence of astronomical tide factors in the forecasting process, the prediction accuracy is poor when the environment changes dramatically. It is difficult to meet the requirement of forecasting accuracy in some occasions. In this paper, a modular tidal forecasting model based on support vector machine (SVM) is proposed to solve the problem that harmonic analysis can not accurately predict the non-astronomical tide. Support vector machine (SVM) is used to predict the non-astronomical tidal part accurately and combine the advantages of harmonic analysis method to improve the prediction accuracy. Firstly, according to the origin of tide, It is divided into astronomical tide parts which are mainly affected by celestial body tidal force and non-astronomical tide parts which are influenced by meteorology, hydrology and other factors. The astronomical tide part is predicted by harmonic analysis method, but not astronomical tide part. Because it is influenced by many factors and has strong nonlinear characteristics, this paper uses support vector machine (SVM) with good nonlinear regression ability to predict this part. This modularization method effectively synthesizes the advantages of the two methods. Then, the graphical user interface is designed for it, which shows the tidal forecast result and simulation process intuitively. Finally, the prediction model and interface are simulated by using the measured data of Honolulu port. The experimental results show that compared with the traditional tidal prediction method, the modular method proposed in this paper is accurate, convenient and easy to operate.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P731.34;TP18

【参考文献】