Fig 1.
Field picture of the studied area.
Fig 2.
Geometric structure and two weaving traffic streams of weaving sections.
Fig 3.
Video processing by using the automated roadway conflicts identification system.
Fig 4.
Velocity fitting diagram and velocity statistical feature diagram.
Fig 5.
The initiation and completion points of lane-changing behavior.
Fig 6.
Statistical histogram of the lane-changing duration and distance.
Fig 7.
Definition of the parameters between the subject vehicle and other vehicles.
Fig 8.
Statistical histogram of the lead and lag distance.
Fig 9.
Hierarchical clustering analysis results.
Table 1.
Sample sizes of different types of weaving environments.
Fig 10.
The distribution of the six environment types in the weaving section.
Fig 11.
Box diagrams of the lane-changing duration and distance.
Fig 12.
Structure of the weaving model based on a neural network.
Table 2.
Weaving behavior data allocation for the different types.
Fig 13.
Prediction results of the weaving model without classification.
Fig 14.
Prediction results of the weaving model for type 1.
Fig 15.
Prediction results of the weaving model for type 5.
Table 3.
The loss value of lane-changing duration and distance prediction results.