组合粗糙集和支持向量回归的船舶交通流预测

影响交通流变化的因素众多，为改进传统的船舶交通流预测精度不高，一种结合粗糙集和支持向量回归智能算法的交通流预测模型提出，通过ROSETTA软件进行属性约简预处理，筛选出影响交通流变化的关键影响因素，剔除冗余信息。筛选结果显示外轮进出艘次、对外贸易总额、港口GDP、集装箱标准箱、港口货物吞吐量为输入变量，运用Libsvm软件构建基于遗传算法参数寻优的支持向量回归模型预测2008年和2009年的交通流。算例结果表明，与BP神经网络和SVM模型相比，组合预测模型是有效和实用的预测工具。

Support vector machine regression model with rough set-based feature selection for forecasting of vessel traffic flow

Many factors contribute to vessel’s traffic flow, to improve the forecasting accuracy of the traditional method, an integrated model of Rough Sets Theory（RST）, Genetic Algorithm（GA） and Support Vector Regression（SVR）, RST-GA-SVR, for traffic volume forecasting is proposed. ROSETTA software is used in data preprocessing, through the rough set attribute reduction algorithm to find the core impact factor of volume and remove redundant information. The results reveal the No. of in-and-out foreign vessels, foreign trade volume, port GDP, container TEU, port cargo throughput as input vector. GA is used for parameters selection in SVR model and the trained LIBSVM model is used to forecast the volume in 2008 and 2009. Compared with BP neural network and SVM models, empirical results show that RST-GA-SVR model is an efficient and practical tool for vessel volume forecasting.