Traffic flow characteristic modal decomposition method based on generation-filtering mechanism

Abstract

The invention discloses a traffic flow characteristic modal decomposition method based on a generation-filtering mechanism, and the method comprises the steps: firstly enabling a high-speed traffic flow to serve as a closed traffic system, enabling each driver to serve as an independent particle according to the randomness of the drivers, simulating a path track, and calculating the probability distribution of the track according to the probability distribution of the track under different parameters; obtaining a corresponding traffic mode; secondly, taking different quantum random walk parameters, obtaining time evolution of traffic flow probability distribution caused by different driving modes on the stations, and then transforming the different stations to form the high-speed traffic flow mode set; and finally, screening the generated traffic modes according to the actually observed traffic flow data, and inverting to obtain the mode structure of the traffic flow. The complex structure and multi-modal characteristics of the traffic flow are disclosed from the perspective of multi-scale decomposition, a certain reference is provided for traffic management, prediction and regulation, and the method has great significance in solving various traffic problems faced by the current society.

Description

technical field [0001] The invention belongs to the fields of urban planning and traffic geography, and in particular relates to a traffic flow characteristic mode decomposition method based on a generating-filtering mechanism. Background technique [0002] Traffic flow is an important indicator for many traffic applications, and it is usually collected by sensors located at (expressway) highway entry and exit stations. Traffic flow is the main carrier of drivers with different characteristics, and its complexity and structural characteristics depend on the driver's driving mode. Assuming that driving trajectories with the same or similar driving patterns are aggregated to form a traffic flow pattern, the complex traffic flow pattern represented by overtaking leads to drastic changes in traffic flow and shows strong randomness; The flow mode has less influence on the traffic flow, and the outgoing / incoming traffic flow is closer. However, the actual traffic flow is not a s...

AI Technical Summary

Problems solved by technology

Adaptive filtering methods have poor processing capabilities for weak signals with low signal-to-noise ratios such as traffic flow, and the calculation convergence process will require a large amount of time and sequence samples as support, and even fail to converge in some cases

[0006] Due to the complex characteristics of non-stationary, nonlinear and quasi-periodic traffic flow, the existing various signal analysis methods are difficult to analyze and extract accurate trend signals, weak signals and slow-varying quasi-periodic signals in the spatio-temporal process of traffic flow. There are defects, nonlinearity and quasi-periodicity are the main reasons for the poor analysis effect of traffic flow time series data

At the same time, the above methods all start from classical statistics and do not consider the inherent characteristics of the traffic flow itself, resulting in the analyzed characteristics and modes without clear physical images, which are difficult to explain

Images