交通流冗余数据识别和约简方法

针对交通检测器检测到的数据存在冗余现象、影响后续决策并需要进行约简的问题，提出了一种冗余数据的识别和约简方法。采用等级分组法实现对冗余数据的识别，先通过等级法计算每个交通参数的权值并按照分组思想，将大数据集分割成许多不相交的小数据集，在各个小数据集中识别冗余数据。为避免漏查，选择其他关键参数多次重复识别。识别出的冗余数据采用平均法约简。实例验证表明，等级分组法识别冗余数据具有较好的精度，随着阈值的增加，查准率和查全率减小，但仍在93%以上；同时采用平均法约简，拟合度较高，达到0.938。可见采用的冗余数据识别和约简方法能够有效地解决单数据源数据冗余问题。

Recognition and reduction of traffic flow redundant data

The detected data often appear redundant, which affects the actual application of traffic models. A method of recognizing and reducing redundant data was proposed. Redundant data were recognized based on rank-based weights and packet method. Firstly, each of traffic parameters was endowed with certain weight according to rank-based weights method. Secondly, in terms of group thought, large data sets were divided into many non-intersecting small data sets. Finally, redundant data were detected and eliminated in each small data set. To avoid missing, the above steps can be repeated. And the recognized redundant data were reduced by average method. An application example shows that, the proposed recognition method of redundant data has a good detection precision, the recall and the precision decreased with the threshold increasing, but still over 93%. The reduced data have a high fitting degree, up to 0.938. The results indicate that, the problem of single data source can be solved effectively.