The invention discloses a traffic road network state estimation method based on a probability diffusion model, which comprises the following steps of: 1, analyzing traffic road network environment data, and constructing a road network environment encoder; step 2, extracting road network state features, constructing a road network level traffic state estimation model Trafic DDPM, adopting a denoising probability diffusion model framework DDPM to maximize fitting probability distribution, adopting the road network environment encoder constructed in the step 1 to extract road network environment attributes and fuse the road network environment attributes into environment codes, adopting a classifier-free guider CFG to control the sampling direction of the model, and obtaining a road network state estimation result; extracting road network features by adopting a U-shaped network structure and an attention mechanism; and step 3, verifying the model from the aspects of model accuracy comparison and instance error for the Trafic DDPM obtained in the step 2, and finally estimating the road network state. The method can accurately and efficiently estimate the traffic network state under any environmental condition, and provides an important basis for decision makers to make management decision measures.
本发明公开了一种基于概率扩散模型的交通路网状态估计方法,包括:步骤1、分析交通路网环境数据,构建路网环境编码器;步骤2、提取路网状态特征,构建路网级别的交通状态估计模型Traffic DDPM,采用去噪概率扩散模型框架DDPM最大化拟合概率分布,采用步骤1所构建的路网环境编码器提取路网环境属性并融合为环境编码,采用无分类器引导器CFG控制模型采样方向,采用U型网络结构和注意力机制提取路网特征;步骤3、对步骤2得到的Traffic DDPM从模型准确度对比、实例误差方面校验模型,最终估计出路网状态。本发明可准确、高效地估计出任意环境条件下的交通路网状态,为决策者制定管理决策举措提供重要依据。
Traffic network state estimation method based on probability diffusion model
一种基于概率扩散模型的交通路网状态估计方法
2024-03-12
Patent
Electronic Resource
Chinese
Traffic toughness evaluation method based on diffusion probability model data reconstruction
European Patent Office | 2024
|TRAFFIC STATE ESTIMATION SYSTEM AND TRAFFIC STATE ESTIMATION METHOD
European Patent Office | 2020
|A Tensor Completion-Based Traffic State Estimation Model
ASCE | 2014
|Expressway road traffic state estimation method based on dynamic Bayesian network
European Patent Office | 2015
|