干散货海运量模糊时间序列预测研究

【作者】 陈罡；

【导师】 顾亚竹；

【作者基本信息】 上海海事大学 ， 企业管理， 2006， 硕士

【摘要】 干散货航运市场是整个航运市场的重要组成部分。为满足干散货国际贸易中海运需求而形成现行的市场结构。在干散货航运市场上，交易信息充分透明，船东数量众多，货源地集中度高，相应地航线较为集中。政治、经济、自然地理和科技等四个宏观因素对干散货航运市场的海运格局和运量有着重要影响。 从干散货国际贸易量中产生的海运量是动态变化的，除了历史数据，对于未来发生的运量无法用精确的数值进行准确的描述。海运量与时间之间不存在象y=x之类明确的函数对应关系式。国际干散货贸易需要通过海运来进行交货，完成交易。但对其具体各时间点产生的海运量却不那么确定，对于这种不确定性，本文认为是种模糊性。 对具有全部或部分模糊性现象领域进行预测应采用模糊预测方法，其与经典预测方法本质的不同在于：各自的集合论基础不同。经典预测方法为经典集合，而模糊预测方法为模糊集合。 时间序列预测方法适用于需要了解预测目标的发展趋势而不考虑或难以考虑其它影响因素的作用，同时要求预测目标的历史数据较为完整。模糊时间序列法不同于经典时间预测之处在于其引入了隶属函数的概念，在序列的预测演算中起到重要作用。 因此，本文借鉴前人的研究成果，将模糊时间序列预测模型引入干散货海运量近期预测领域，采用模糊时间序列预测方法分别建立粮食、铁矿石和煤炭三大干散货近期运量预测模型，并从理论和示例演算两方面验证其适用性。其中，粮食为模糊平滑预测模型，分别采用模糊滑动平均法和数据色彩滑动平均法建模，通过示例演算与经典滑动平均法的比较可以看出，模糊方法比经典方法好；铁矿石和煤炭为模糊多项式预测模型，该模型为区间预测模型——示例演算结果令人满意。更多还原

【Abstract】 The dry bulk shipping market is the important part of the whole shipping market. The market’s construction is shaped to meet the need of the marine traffic of the international trade. In the market, the trade infermation is well-informed, and there are so many ship-owners, and the freight source terminals are highly concentrated, related to the shipping line. The four macroscopical elements which are politics, economy, nature&geography and technology make a heavy influence to the marine traffic and the consructure of the ocean shipping of the dry bulk shipping market.The marine traffic caused by the dry bulk international trade is dynamically changed. Except the historical statistics, it is an impossible task to have an accurate picture of the future marine traffic. There is not a clear function coincidence relation between the marine traffic and the time, such as y=x. It is certain that the international trade of the dry bulk complete the contract through delivering the cargo on time by the ocean shipping. But it is not certain to when and how much the marine traffic take place. The dissertaion consider such uncertainty as a kind of fuzziness.It should adopt the methods of fuzzy forecasting to forecast the future which have the whole or part of fuzzy characteratics. The methods of fuzzy forecasting differ from the methods of classic forecasting lies in the different sets. The classic methods are cantor set, while the fuzzy is fuzzy set.The time series forecasting are acceptable for those fields which just want to forecast the target’s future development trend and care little or hard to care the other influence elements, while the target’s historical stastics is neraly intergrity at least. The fuzzy time series forecasting differ from classic time series forecasting is lead in the conception, named membership function which contribute much to figure the method.As a result, the dissertation use the predecessor’s research findings for reference and lead the fuzzy time series forecasting model in forecasting the recent marine traffic of dry bulk, and adopt the methods to separately build the foodstuff, iron ore and coal recent marine traffic forecasting models whose applicability are proved by the theory and figuring the illustration. Hereinto, the foodstuff is fuzzy smoothness model, which use the fuzzy moving-average method and the data color moving-average method separately to build the model, which better than classic moving-average method showed by the calculated sample; while the iron ore and coal is fuzzy polynomial forecast model, which are interval prediction models, which are interval prediction model whose calculated sample are satisfied.更多还原