A Novel Environment Estimation Method of Whole Sample Traffic Flows and Emissions Based on Multifactor MFD

Area of Science:

• Environmental Science
• Transportation Engineering
• Computer Science

Background:

• Vehicle emissions significantly impact air quality and public health.
• Accurate dynamic estimation of road network emissions is crucial for air quality simulation.
• Obtaining real-time traffic volume data for emission estimation is challenging.

Purpose of the Study:

• To propose a novel method for estimating traffic volumes and emissions across an entire road network.
• To develop multifactor Macroscopic Fundamental Diagram (MFD) models for dynamic emission assessment.
• To enhance the accuracy and spatiotemporal resolution of vehicle emission estimations.

Main Methods:

• Utilized neural network and deep-learning algorithms for traffic flow pattern clustering and recognition.
• Developed multifactor MFD models incorporating road types, traffic patterns, and peak hours.
• Employed Self-Organizing Maps (SOM) and Deep Belief Network (DBN) algorithms for pattern analysis.

Main Results:

• Self-Organizing Maps (SOM) efficiently clustered 91.7% of urban road traffic states.
• Deep Belief Network (DBN) accurately recognized 92.1% of traffic patterns during peak hours.
• Multifactor MFD models achieved a high accuracy of 91.6% in estimating traffic volumes.

Conclusions:

• The proposed method provides high spatiotemporal resolution estimation of traffic volumes and emissions.
• Dynamic vehicle emission evaluation is facilitated, supporting coordinated air pollution treatment.
• The novel approach addresses the bottleneck in dynamic emission estimation for air quality simulation.