The invention discloses a traffic data restoration method based on non-convex low-rank tensor representation. The method comprises the following steps: S1, constructing a traffic data third-order tensor; s2, the Capped Schatten p norm of the tensor is defined, and the Capped Schatten p norm is defined; s3, constructing an improved Laplacian regular term; and S4, solving an optimization problem by using an iteration mode until a termination condition is met. According to the method, a Capped Schatten p norm of a tensor is used as a non-convex approximation of a rank, so that the inherent low-rank characteristic of traffic data is better represented; according to the method, the traffic state change characteristics of similar road sections are utilized to assist in repairing missing data, so that high-precision repairing of the traffic data is completed, the urgent requirement for high data quality in intelligent traffic system application can be met, and the method has theoretical significance and application value.
本发明公开了一种基于非凸低秩张量表示的交通数据修复方法,包括如下步骤:S1、构造交通数据三阶张量;S2、定义张量的Capped Schatten p范数;S3、构造改进的拉普拉斯正则项;S4、使用迭代方式求解最优化问题直至满足终止条件。该方法使用张量的Capped Schatten p范数,作为秩的非凸近似,以更好地表征交通数据固有的低秩特性;并整合改进的拉普拉斯正则项作为空间平滑约束,以利用相似路段的交通状态变化特征来辅助修复缺失数据,进而完成交通数据的高精度修复,能够满足智能交通系统应用中对数据高质量的迫切需求,具有理论意义和应用价值。
Traffic data restoration method based on non-convex low-rank tensor representation
一种基于非凸低秩张量表示的交通数据修复方法
2023-03-21
Patent
Electronic Resource
Chinese
Traffic forecasting with missing data via low rank dynamic mode decomposition of tensor
Wiley | 2022
|