适用于车辆轨迹数据处理的小波去噪评价方法

Wavelet Denoising Evaluation Method for Vehicle Trajectory Data Processing

在车辆轨迹数据处理中,针对小波去噪的效果缺乏评价方法,导致小波去噪的分解层数往往基于经验或直接给定得到,缺乏理论支撑这一问题,文章提出一种适用于车辆轨迹数据处理的小波去噪复合评价指标。首先计算得到均方根误差变化量和平滑度变化量两个指标,然后使用熵权法和变异系数法进行组合赋权,将权值和归一化后指标进行线性组合后构成复合评价指标,复合评价指标越小则表明小波去噪效果越好。结果表明,文章提出的复合评价指标在不同强度的噪声情况下均能较好的指导小波去噪的分解层数选择。

In vehicle trajectory data processing,there is a lack of evaluation methods for the effect of wavelet denoising,which leads to the problem that the decomposition layers of wavelet denoising are often obtained based on experience or directly given,and lack of theoretical support.In this paper,a composite evaluation index of wavelet denoising suitable for vehicle trajectory data processing is proposed.First,two indexes,the change of the root mean square error and the change of smoothness,were calculated,and then the entropy weight method and the coefficient of variation method were used for combination weighting.The weight and the normalized index were linearly combined to form the composite evaluation index in this paper.The smaller the composite evaluation index,the better the effect of wavelet denoising.The results show that the composite evaluation index proposed in this paper can guide the selection of the decomposition layers of wavelet denoising well under different noise intensities.