基于GRA-TOOPSO-LSSVM的港口吞吐量预测

Port throughput forecasting based on GRA-TOOPSO-LSSVM

为对港口吞吐量进行科学预测,在最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)基础上,引入灰色关联分析(Grey Relational Analysis,GRA)和二阶振荡粒子群优化(Two-Order Oscillating Particle Swarm Optimization,TOOPSO),提出一种新的GRA-TOOPSO-LSSVM算法预测港口吞吐量.采用GRA法筛选出对上海港吞吐量有重大影响的因素,并将其作为LSSVM的输入变量;采用TOOPSO法对LSSVM的参数进行寻优;运用LSSVM非线性映射的优势对上海港吞吐量进行预测.在上海港吞吐量实证研究的过程中,GRA-TOOPSO-LSSVM算法与TOOPSOLSSVM和基于交叉验证的LSSVM算法进行对比分析.研究结果表明,GRA-TOOPSO-LSSVM算法具有更好的预测精度和收敛速度,为港口吞吐量预测的研究提供了一种新的方法.

In order to forecast port throughput scientifically, Grey Relational Analysis （GRA） and Two- Order Oscillating Particle Swarm Optimization （T00PS0） are introduced on the basis of Least Squares Support Vector Machine （LSSVM） , and a new GRA-TOOPSO-LSSVM algorithm of port throughput fore-casting is proposed. GRA method is used to select the factors that have great influence on Shanghai Port throughput, and the factors are used as input variables of LSSVM. T00PS0 method is used to optimize the parameters of LSSVM. The nonlinear mapping advantage of LSSVM is used to forecast Shanghai Port throughput. In the process of empirical study on Shanghai Port throughput, GRA-TOOPSO-LSSVM algo-rithm, TOOPSO-LSSVM algorithm and LSSVM algorithm based on cross validation are compared. Results show that GRA-TOOPSO-LSSVM algorithm is of better forecasting accuracy and convergence rate, which provides a new method for forecasting port throughput.