基于支持向量机的时域虚拟路谱评价方法

Evaluation Method of Time-Domain Virtual Road Spectrum Based on Support Vector Machine

针对虚拟路谱缺乏有效评价手段的问题,利用带有高斯核函数(RBF)的支持向量机(SVM)算法以及路谱数据的客观统计值与主观评分建立了评价模型。首先,对虚拟路谱的客观统计值进行分析并在此基础上建立特征值;然后,利用网格搜索的方法搜索模型的惩罚系数和内核系数,建立时域信号分开建模的模型作为评价模型;最后,使用虚拟路谱数据对模型的评分结果进行分析,结果表明,所设计的评分模型精度达标,提高了路谱评价的效率。

For the issues of the present insufficient evaluation of the virtual road spectrum,a novel evaluation model is constructed based on the Support Vector Machine(SVM)using Radial Basis Function(RBF)along with objective statistics of load data and subjective marks.First,the eigenvalue is established based on the statistical summary of the virtual road load data.Then,the penalty coefficients and the kernel factor of the model are searched with the grid search method.Afterwards,the modeling with separate time-domain signal is constructed as the evaluation model.Finally,virtual road load data are applied to analyze the model scoring results.The results show that the designed scoring model meets accuracy standard,improves efficiency of road spectrum evaluation.