The invention relates to the technical field of intelligent transportation, and provides a long-time traffic flow prediction model based on a low-rank map and STL (Standard Template Library) time sequence decomposition. The encoder structure network encodes a sequence of input data through a first self-correlation mechanism module, a first LGCN module and a first STL time sequence decomposition module, and the decoder structure network learns primary sub-sequence items for the encoder input data through a second self-attention mechanism, a second LGCN module and a second STL time sequence decomposition module. Outputting a secondary subsequence item by combining a coding and decoding cross attention fusion module, finally performing association rule mining on the secondary subsequence item by combining a second LGCN module and a Linear module, and superposing and outputting elements of the three subsequence items to obtain a long-time traffic flow prediction model; the method is suitable for long-time prediction of the dynamic highway traffic flow, and can achieve high accuracy of long-time prediction under the actually collected highway traffic flow data.
本发明涉及智能交通运输技术领域,提供一种基于低秩图和STL时序分解的长时交通流预测模型,采用编码器‑解码器框架结构作为模型整体框架,编码器结构网络对输入数据的序列通过第一自相关机制模块、第一LGCN模块和第一STL时序分解模块进行编码,解码器结构网络对编码器输入数据通过与第二自注意力机制、第二LGCN模块、第二STL时序分解模块学习到一次子序列项后,结合编解码交叉注意力融合模块输出二次子序列项,最终结合第二LGCN模块与Linear模块对二次子序列项进行关联规则挖掘,将三个子序列项元素叠加输出,得到长时交通流预测模型;适用于对动态高速公路交通流进行长时预测,在真实采集的高速路交通流数据下,能够对长时预测达到较高的准确度。
Long-time traffic flow prediction model and method based on low-rank map and STL (Standard Template Library) time sequence decomposition
基于低秩图和STL时序分解的长时交通流预测模型及方法
2024-02-06
Patent
Elektronische Ressource
Chinesisch
Long-time-sequence traffic flow prediction method based on graph convolution-Informer model
Europäisches Patentamt | 2021
|Europäisches Patentamt | 2023
|Ship traffic flow long sequence space-time prediction method based on ST-Informer
Europäisches Patentamt | 2023
|Short-term traffic flow prediction method based on time series decomposition
Europäisches Patentamt | 2020
|