# Calibration of parameters in microscopic traffic flow simulation models considering micro-meteorological information

## TL;DR

This paper improves traffic simulation models by incorporating driver behavior changes under rainy conditions, using micro-meteorological data to enhance accuracy.

## Contribution

The study introduces a driver judgment factor into traffic models and compares the performance of two improved models under rainy conditions.

## Key findings

- The I-IDM model shows lower average error and standard deviation compared to the I-Wiedemann99 model.
- The I-IDM model's maximum RMSPE is 0.4568, while the I-Wiedemann99 model's maximum RMSPE is 0.4613.
- The I-IDM model simulates following behavior more effectively than the I-Wiedemann99 model under rainy conditions.

## Abstract

Different micro-meteorological conditions can affect a driver’s judgment of road conditions, leading to changes in following behavior. On rainy days, water films on the road reduce traction, increasing the likelihood of hydroplaning and traffic accidents. While there are existing following models under various weather conditions, research on the specific impact of micro-meteorological factors is insufficient. To achieve fine management in intelligent transportation and real-time monitoring of vehicle states, it’s essential to study following behavior under different micro-meteorological conditions and establish corresponding models. This paper focuses on the Intelligent Driver Model (IDM) and the Wiedemann99 model, considering the impact of micro-meteorological conditions. By incorporating a driver’s judgment factor, λ, the IDM and Wiedemann99 models are improved, leading to the development of new models: I-IDM and I-Wiedemann99. Simulation validation is used to choose speed and following distance as performance indicators for parameter calibration of the I-IDM and I-Wiedemann99 models, with the sum of Root Mean Square Percentage Error (RMSPE) as the goodness-of-fit function. Comparisons are made between the driving paths, speeds, and accelerations of following vehicles before and after calibration, verified through simulations. The conclusions are as follows: the average error and standard deviation of the improved I-IDM model are smaller than those of the I-Wiedemann99 model, with the maximum Root Mean Square Percentage Error (RMSPE) for I-IDM model parameter calibration being 0.4568 and the minimum being 0.1324. For the I-Wiedemann99 model, the maximum RMSPE is 0.4613 and the minimum is 0.1376. The parameter calibration results of the I-Wiedemann99 model are more dispersed compared to those of the I-IDM model, indicating that the I-IDM model simulates following behavior more effectively than the I-Wiedemann99 model. The findings of this study can provide a reference for further improving the theory of following behavior, and offer a theoretical basis and IoT technology support for refined traffic management under rainy conditions.