The invention discloses a short-time traffic flow prediction model based on time sequence analysis and residual matching. The random characteristic of the traffic flow is the root cause of difficulty in accurately predicting the traffic flow, and in order to better fit the random fluctuation in the traffic flow data, the model adopts a decomposition thought to divide the original traffic flow data into a linear part and a residual error part. The method comprises the following steps: firstly, calculating a linear part in data by using a time sequence analysis technology, and segmenting a residual part to construct a residual vector library; and then matching the residual vectors by using a pattern matching algorithm, and further fitting random fluctuations in the traffic flow data in a mode of searching similar residual vectors. Compared with other models based on a pattern matching algorithm, the model has certain advantages in stability and prediction precision.
本发明公开了一种基于时序分析和残差匹配的短时交通流预测模型。交通流的随机特性是导致交通流难以被准确预测的根本原因,为了更好地拟合交通流数据中的随机波动,本模型采用分解的思路,将原始交通流数据划分为线性部分和残差部分。首先利用时间序列分析技术计算数据中的线性部分,并对剩余的残差部分进行切分以构建残差向量库。然后利用模式匹配算法对残差向量进行匹配,通过寻找相似残差向量的方式进一步拟合交通流数据中的随机波动。本模型相对于其他基于模式匹配算法的模型而言,在稳定性和预测精度上具有一定的优势。
Model based on time sequence analysis and residual matching
一种基于时序分析和残差匹配的模型
2021-04-16
Patent
Elektronische Ressource
Chinesisch
Feature-Based Sequence-to-Sequence Matching
British Library Online Contents | 2006
|Power battery residual endurance prediction method based on time sequence large model algorithm
Europäisches Patentamt | 2024
|Travel Time Measuring by Using Vehicle Sequence Matching between Adjacent Intersections
British Library Conference Proceedings | 2005
|Navigation using image sequence analysis and 3-D terrain matching
Tema Archiv | 1990
|