基于混沌高效遗传算法优化SVM的交通量预测

Prediction of Traffic Flow Using Support Vector Machine Optimized by Chaos Higher Efficient Genetic Algorithm

针对交通量预测本身所存在的小样本、非线性和复杂性等特点,利用支持向量机建立了基于RBF核函数的SVM交通量预测模型,采用基于混沌映射和加速遗传算法的混沌高效遗传算法对SVM模型参数C,ε和δ2进行优选,结合某市1978～2008年交通量实测资料进行了仿真验证,与GA-SVM模型和BP神经网络模型的仿真预测结果对比表明:该模型取得了较好的预测效果,可有效应用于城市交通量的预测.

A SVM prediction model of traffic flow is built with kernel function of RBF,and chaos higher efficient genetic algorithm based on chaos map and genetic algorithm is used to preference the parameters С,ε and δ2 of SVM,aiming at traffic flow prediction′s feature of small sample,non-linear and complexity.At last simulation verification with observed passenger volume data over 1978-2008 was made.Compaing with GA-SVM model and tradition BP model,analysis represents that this model have better prediction result,and can be effectively applied to prediction of traffic flow.