摘要
针对交通流量数据具有非线性和非平稳性的特点,运用EMD和FOA算法实现LS-SVM核参数和惩罚系数的自适应优化选择,提出了一种基于EFLS-SVM算法的交通流量预测模型。通过EMD提取交通流量数据的细节特征和趋势特征,构建出基于EFLS-SVM的交通流量预测模型,分别进行单步、3步、5步和7步预测。通过不同交通流量预测模型的实验对比发现,EFLS-SVM算法的预测精度和预测效率均优于其他模型,从而为交通网络资源的合理配置提供科学决策的依据。
According to the traffic flow data having nonlinear and non-stationary characteristics, the EMD and FOA algorithm were used to implement self-adaptive optimization selection of LS-SVM kernel parameters and penalty coefficient, and then a traffic flow forecasting model based on EFLS-SVM algorithm was proposed. The minutiae characteristics and trend feature of traffic flow data were extracted by EMD, and a traffic prediction model based on EFLS-SVM was built, then single- step, three-step, five-step and seven-step prediction were proceeded respectively. By comparing different experiments, the results showed that prediction accuracy and prediction efficiency of EFLS-SVM algorithm were better than other models, thus scientific decision-making basis was provided for the rational allocation of transport network resources.
出处
《四川理工学院学报（自然科学版）》
CAS
2015年第6期29-35,共7页
Journal of Sichuan University of Science & Engineering(Natural Science Edition)
关键词
交通流量
果蝇优化算法
数学模型
最小二乘法支持向量机
traffic flow
fruit fly optimization algorithm
mathematical model
least squares support vector machine
